UBS Group AG (NYSE: UBS) — The Prize Was Won in 2023; the Price Now Assumes Bern Never Sends the Bill

Report date: 2026-08-01 | Price (ref.): ~$52.74 (2026-07-31 close) | Shares out: ~3,061.7M | Market cap: ~$161.5B Sector: Financials — Diversified Banks & Global Wealth Management | CIK: 0001610520 | Exchange: NYSE (UBS) / SIX Swiss Exchange (UBSG) | FYE: December | Reporting: IFRS, in USD

UBS is a foreign private issuer: it files Form 20-F annually and Form 6-K for quarterly and material events — there is no 10-K, 10-Q, 8-K or DEF 14A. Primary sources: the FY2021–FY2025 Form 20-F set (FY2025 filed 2026-03-09); the 2Q26 Interim Report and results 6-K cluster (2026-07-29) and earnings-call transcript (2026-07-30); the Swiss Federal Council / Federal Department of Finance too-big-to-fail dispatch and Capital Adequacy Ordinance (2026-04-22); UBS’s own regulatory statements and consultation responses; and published filings of comparable banks. All figures reconciled to filings; management commentary treated as hypothesis. Except for the clearly-labeled “Claude’s Take” block below, this article contains no investment recommendation and no price target — the analytical body discusses valuation only as embedded expectations and scenarios.

Data-integrity note (material — read before the numbers). Two third-party feeds are wrong on this name and were rejected. UBS reports in USD, but AZI returns a CHF-converted book value (book_value_per_share $22.93 against the filed $29.12 — below even filed tangible book of $26.89, which is arithmetically impossible) and a CHF-converted ttm_eps of $2.21 against filed TTM diluted EPS of $2.94. ROIC.ai likewise returns CHF-converted statements (its payload flags currency: "CHF", fx_applied: true; FY2025 revenue of 37,476 vs the filed USD 49,573), and its bank ratios are unusable. FactorsToday’s market cap of $175.0bn is computed on shares issued; on shares outstanding the figure is ~$161.5bn. Because USD/CHF drifted from ~1.00 (2015) to ~0.80 (2026), AZI’s own-history valuation percentiles are biased upward in recent years and are quoted here only as directional. Every ratio in this article is rebuilt from UBS’s filed figures.

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information and not investment advice, and it is not a recommendation to buy or sell any security. Everything below it — the full analytical body of this article — deliberately carries no position and no price target. Do your own work.

Verdict: AVOID here / trim into strength — emphatically not a short. The turnaround is real, delivered, and largely finished; the problem is that the price now sits above the multiple justified by UBS fully achieving its own 2028 ambition — an ambition its own filings say is void under the capital rules already sitting in the Swiss Parliament. “The prize was won in 2023; the price now assumes Bern never sends the bill.”

Directional valuation zone (the author’s own view): At $52.74 UBS trades at 1.96x tangible book ($26.89 TBVPS) and ~13.3x normalized underlying earnings. Run the standard bank test — justified P/TBV = (RoTE − g)/(CoE − g), at CoE 10% and g 3%: the 14.7% RoTE UBS actually earns today justifies ~1.67x; the 2028 ambition of ~18% RoCET1 (≈15.9% RoTE) justifies ~1.84x; and that ambition is expressly caveated in the FY2025 Annual Report as “based on the current capital framework and assuming a CET1 capital ratio of around 14%” — the very framework the Federal Council sent to Parliament on 22 April 2026. Under the proposed rules the de facto Group CET1 minimum rises to ~17.6–18.4% on UBS’s own current arithmetic, which scales today’s 16.7% underlying RoCET1 down to ~12.7–13.3% and turns the 18% ambition into roughly 13.7%, supporting about 1.3x. So the stock is priced above the best realistic case. I would want ~1.4–1.7x tangible book — roughly $38–46 — before the risk/reward turns, a zone the stock visited as recently as March 2026 ($35.49). Above ~1.9x I am a seller of strength, not a buyer of dips.

What the market is pricing correctly, and what it isn’t. Correctly: the integration genuinely worked. Underlying RoCET1 has gone 4.2% → 8.7% → 13.7% → 16.7%, the 2026 exit-rate targets were hit five months early, invested assets are a record $7.33trn, and — unusually — reported profit understates the operating improvement, because FY2025 was flattered by a 11.9% tax rate and $949m of litigation releases that 1H26 does not repeat. Incorrectly, in three places. First, quality of the inflection: 44% of the 1H26 underlying pre-tax improvement came from an Investment Bank whose profit nearly doubled in a record equities quarter the CFO himself called “unlikely to continue at that level,” while underlying operating expenses actually rose 5.6% — the entire cost/income gain is revenue-led, which is cyclical leverage, not structural. Second, the capital bill: $37bn on UBS’s own arithmetic, against just ~$2.0bn of CET1 headroom over the 14% guide; the guided $3bn buyback (a 1.9% yield) is the arithmetic residual after funding the proposed glide path, not a floor — and UBS has conditioned it, in writing, on “the deliberations by the Swiss Parliament.” Third, the US: the Americas holds 49% of wealth invested assets at a 16% pre-tax margin against APAC’s 45%. Stated precisely — because the sloppy version of this claim is wrong — UBS’s group wealth margin of 26.5% reported / 28.5% underlying is broadly in line with Morgan Stanley’s 30%, and the Americas margin is improving (9.3% FY24 → 12.7% FY25 → 15.9% in 2Q26, on management’s own ~15% target). The problem is the flows, and the cleanest number is one UBS discloses only annually: “net new assets” includes interest and dividends, and on a pure net-new-money basis GWM Americas lost $57.7bn of client money in FY2025 while reporting net new assets of just −$5.9bn. Americas net new fee-generating assets have collapsed from +$50.7bn (FY24) to +$11.7bn (FY25) to +$4.2bn (1H26).

The strongest arguments against me, stated fairly. First, the size of the capital bill is disputed by the authorities themselves, and UBS’s $37bn is an advocacy number. The Federal Council’s April 2026 dispatch puts the headline at ~$20bn but the effective shortfall at only ~$9bn, and the Swiss National Bank’s Financial Stability Report 2026 states flatly that UBS “already has sufficient capital to meet the requirements for full capital backing” — Group eligible CET1 exceeding fully-applied 2030 requirements by $9bn and the parent bank by $13bn. The two sides differ on annual cost by 3–5x ($320–560m per the Federal Council against $1.7bn per UBS). The reconciliation is that UBS is describing an accounting deduction while the authorities describe capital that must actually be raised or retained given the buffer already carried — so the solvency impact is close to nil and the honest RoCET1 haircut spans a wide 1.9 to 4.2 points, not the 4+ points a naive reading of $37bn implies. Second, there is a genuinely non-obvious rate asymmetry working in UBS’s favour: on UBS’s own disclosure a +100bp shift adds ~$1.4bn of annual banking-book net interest income and a −100bp shift also adds ~$0.9bn, because of contractual flooring benefits under negative rates. Zero is the pessimal point for Swiss net interest income, and the SNB has parked exactly there — any exit, in either direction, is a tailwind. Third, cross-sectionally UBS is not expensive. Regressing price-to-tangible-book on return on tangible equity across seven bank comps from published peer filings give a fitted line that predicts 2.20x for UBS at its 14.4% underlying RoTE — against 1.96x traded, i.e. roughly 11% cheap, at an implied cost of equity of 8.8% that is dearer than the European universals but cheaper than every US comparator. On earnings the shares are ~14.3x underlying, which is unremarkable. And the capital bill, discounted properly, is far smaller than it looks: $20.8bn of incremental CET1 earning ~3% after tax, phased from 2028 over seven years, has a present value of only about $2.33 a share. So the honest statement is that the capital rule is overwhelmingly a multiple risk, not an intrinsic-value risk — it barely touches earnings per share while mechanically compressing headline RoTE, and banks are marked on headline RoTE. Bulls quoting P/E and bears quoting P/TBV are reading the same identity (P/E = P/TBV ÷ RoTE) from opposite ends. My call rests on believing the market will mark the denominator, and on the TTM RoTE of 11.6% — which includes the ~$2.6bn a year of integration expense that has been real cash every year since 2023 — being closer to the truth than the “underlying” 14.4%. On TTM the same regression says 1.59x, and UBS is ~23% dear.

Framing. This is a late-innings, event-driven re-rating, not a momentum trade and not a value name — the factor model puts UBS’s Momentum and Value betas at essentially zero, with ~44% of variance idiosyncratic, because the price path has tracked Swiss regulatory headlines rather than style rotations. The market has already paid for regulatory relief twice (+20.0% in December 2025 on a floated compromise; +6.5% on 31 March 2026 on a softer signal). Scenario-weighting the outcomes, the traded $52.74 solves to roughly 62% bull / 23% base / 15% bear — a probability of softening well above what an upper-house committee still weighing 50%, 70% and 80% can support. Relief is not a catalyst; it is the base case in the price. Note also the objective scoreboard management could not adjust: relative TSR over the 2023–2025 long-term incentive period ran 17.45 percentage points below the G-SIB index.

Conviction: medium-high. Flips bullish if the upper-house committee (sitting 10, 11 and 31 August 2026) lands the foreign-participation requirement at or below 70% with UBS simultaneously re-basing the 2028 ambition upward at its 4Q26 strategic update — that combination would legitimize a materially larger buyback and a higher justified multiple. Flips bearish if the requirement is adopted at 100% on the proposed glide path, or if GWM Americas posts a third consecutive quarter of negative net new fee-generating assets, which would convert an advisor-attrition story into a franchise-erosion story.

📈 Stock Price Action — Five-Year Event Map

Over the trailing five years UBS has gone from ~$12 to ~$55 — a ~4.5x run that is one of the great post-crisis bank re-ratings, and one delivered in three violent legs rather than a straight line. The stock closed at $52.74 on 31 July 2026, against a 52-week range of $35.49 (27 Mar 2026) to $55.07 (15 Jul 2026), leaving it ~4.2% below its high and ~49% above its 52-week low. That July high is the highest dividend-adjusted close in the 26-year price series; the unadjusted pre-2008 peak of ~$177 is not comparable and has never been recovered.

1. The 2022 drawdown was macro, not idiosyncratic — a synchronised global de-rating of financials into the rate shock, punctuated by a −11.3% session on the 2Q22 print as investment-bank revenue collapsed. 2. The recovery through early 2023 tracked the market’s re-embrace of rate-levered balance sheets; by 6 March 2023 UBS was back at ~$19.30, its highest since 2018.

3. The Credit Suisse episode is widely misremembered. UBS fell 18% in nine sessions into the crisis on Swiss-banking contagion, bottoming at $15.85 on 17 March 2023 — before the deal. The CHF 3bn shotgun acquisition and the CHF 16bn AT1 write-down were announced on Sunday 19 March; UBS then rose 3.3% on Monday the 20th on ~10x normal volume and 12.0% on the 21st, closing the month at $18.58. Interpretation: the market immediately read the deal as an opportunistic purchase of a wealth franchise at a fraction of book, backstopped by the state.

4. Two dated catalysts drove the 2023 leg: 11 August 2023 (+5.6%), when UBS terminated the CHF 9bn loss-protection agreement and CHF 100bn public liquidity backstop, removing the taxpayer from the trade; and 31 August 2023 (+5.6%), the 2Q23 print carrying the negative-goodwill gain. 2024 was then a wide range — cost saves delivered, Swiss capital unresolved.

5. 2025 was a round trip: −7.2% on the 4Q24 print, then the April tariff shock, then a +66% recovery. The Federal Council’s 6 June 2025 TBTF consultation landed inside that rally; UBS actually rose 2.3% on the day before bleeding ~9% over the following fortnight — a multi-quarter overhang on the multiple, not a discrete crash. 6. December 2025 was the regulatory inflection: lawmakers floated a compromise on 12 December, taking the stock to a 17-year high. 7. The February–March 2026 −26% drawdown began with the 4Q25 print, a headline beat the market discounted for reserve releases and guided US wealth outflows, then compounded by macro. 8. The final leg began 31 March 2026 as lawmakers signalled softer rules; the 22 April final CAO quantified the bill; and the 29 July 2Q26 print ($2.8bn net profit, underlying pre-tax profit +45% YoY, $7.33trn invested assets, a new $3bn buyback) took the stock to within ~4% of its record.

1. Executive Summary

UBS Group AG holds $7.33 trillion of Group invested assets at 30 June 2026, of which $4.94 trillion sits in Global Wealth Management, and is since June 2023 the only large universal bank in Switzerland. It reached that position by acquiring Credit Suisse in a state-brokered rescue for CHF 3bn in stock, booking a $27.3bn negative-goodwill gain, and then spending three years and $14.2bn integrating it. Two framing corrections are due at the outset: UBS’s “world’s largest wealth manager” claim holds only on the invested-assets definition it selects — Morgan Stanley’s wealth client assets are $7.34trn — and UBS’s Swiss “dominance” is largely a myth (the Swiss domestic section): it holds 23% of domestic loans and deposits against 26%/31% for the three domestic systemically important banks combined, and it has lost 4.2 points of mortgage share and 6.4 points of non-mortgage lending share since 2021.

The integration worked, and it is essentially finished. Underlying return on CET1 capital has gone 4.2% (FY2023) → 8.7% (FY2024) → 13.7% (FY2025) → 16.7% (1H26). The underlying cost/income ratio has gone 87.2% → 79.5% → 74.4% → 70.1%. Both 2026 exit-rate targets (≈15% underlying RoCET1, <70% underlying cost/income) were reached five months early; management says it will “outperform” the return target. Internal headcount is down 12% since FY2023 and 28% since 2022; more than 90% of legacy applications are retired; all clients are migrated; Non-core and Legacy has shed 67% of its risk-weighted assets and 93% of its leverage exposure. Cumulative gross cost savings stand at $12.6bn against a $13.5bn ambition.

Three things qualify that success. First, the destination is not new: standalone UBS earned RoCET1 of 17.4% / 17.5% / 17.0% in FY2020–22, so 16.7% today is a recovery, not an advance — a point Sergio Ermotti conceded on the 2Q26 call (“we are close to achieving the same level of profitability UBS had prior to the acquisition”). Second, the composition of the 1H26 inflection is more cyclical than the headline: 44% of the underlying pre-tax improvement came from the Investment Bank, whose profit nearly doubled in a record equities quarter the CFO called “unlikely to continue at that level,” while underlying operating expenses rose 5.6% — the cost/income gain is entirely revenue-led. Third, management’s own arithmetic strips the shine: on its telling, FY2025’s 13.7% underlying RoCET1 becomes 11.5% excluding litigation releases and at a normalized tax rate.

The dominant variable is not operational; it is legislative. On 22 April 2026 the Swiss Federal Council adopted both a final Capital Adequacy Ordinance (in force 1 January 2027) and a dispatch to Parliament requiring systemically important banks to back foreign participations fully with CET1 capital, phased over seven years from 65% to 100%. UBS’s own quantification, stated at the entity levels it actually uses: ~$20bn for foreign participations plus ~$2bn from the CAO, both at UBS AG standalone (~$22bn), plus ~$15bn of Credit-Suisse-driven requirements already phasing to 2030 — ~$37bn of additional CET1 in total, at “an annual capital cost of around $3bn.” (The Group-level CAO effect is a separate ~$4bn derecognition, worth 0.8 points of the Group CET1 ratio.) That sits against CET1 capital of $72.5bn and headroom of roughly $2.0bn over the 14% guide. The figure is contested: the Federal Council’s headline is ~$20bn with an effective shortfall of only ~$9bn, and the SNB’s Financial Stability Report 2026 states UBS “already has sufficient capital to meet the requirements for full capital backing.” The two sides differ on annual cost by 3–5x. Parliament debates it from summer 2026; the upper-house committee sits on 10, 11 and 31 August 2026 and is weighing 70%, 80% or 50% in place of 100%.

The consequence is a capped distribution rather than a solvency problem. UBS generates roughly $11.7bn of annualized net profit; after RWA growth and a ~$3.9bn dividend, about $6bn is distributable, of which the proposed glide path consumes roughly half. The guided $3bn buyback — a 1.9% yield, decelerating from a front-loaded 2026 programme, with only $1bn actually committed — is the arithmetic residual, not a floor. UBS has said in writing that its pace depends on “the deliberations by the Swiss Parliament on the capitalization of foreign subsidiaries.” Critically, UBS’s flagship 2028 ambition of ~18% reported RoCET1 is expressly stated “based on the current capital framework and assuming a CET1 capital ratio of around 14%” — the framework now before Parliament. On the proposed rules the de facto Group minimum rises to ~17.6–18.4%, mechanically scaling today’s 16.7% underlying RoCET1 to ~12.7–13.3% and converting the 18% ambition into roughly 13.7% on identical earnings. Offsetting this, UBS has demonstrated real mitigation capacity: $9bn upstreamed from subsidiaries in 4Q25 cut parent RWA by $26bn, and a further ~$4.5bn of parent-bank dividend has been deliberately withheld pending the outcome.

The franchise’s weakest link is the United States — and the claim needs stating precisely, because the loose version is wrong. UBS’s group wealth pre-tax margin of 26.5% reported / 28.5% underlying is broadly in line with Morgan Stanley Wealth’s 30%. The weakness is entirely regional: GWM Americas holds $2,423bn — 49% of wealth invested assets — at a 15.9% pre-tax margin against 44.6% in APAC, 38.8% in Switzerland and 37.9% in EMEA. That margin is in fact improving (9.3% in FY2024 → 12.7% in FY2025 → 13.7% in 1Q26 → 15.9% in 2Q26, the best in the series and on management’s own ~15% target for 2026). The problem is the flows, not the margin. The single most revealing disclosure is annual only: under UBS’s own definitions “net new assets” includes interest and dividends, and on a pure net-new-money basis GWM Americas lost $57.7bn of client money in FY2025 while reporting net new assets of just −$5.9bn. Americas net new fee-generating assets have decayed from +$50.7bn (FY2024) to +$11.7bn (FY2025) to +$4.2bn (1H26) — the 2Q26 figure of −$4.6bn followed +$8.8bn in 1Q26, so it is one quarter, not yet a trend, and management notes ~$10bn of seasonal tax-related outflows. Advisor headcount is down 4% year on year. Against a 2028 ambition of more than $200bn of Group net new assets per annum (versus ~$146bn annualized in 1H26), the US is where that arithmetic has to come from.

Valuation is the binding constraint. At $52.74 the shares trade at 1.96x tangible book ($26.89 TBVPS), 1.81x book, ~17.9x trailing earnings and ~13.3x normalized underlying earnings. Rebuilt from filed figures, UBS’s own P/TBV history reads 1.12x (YE21), 1.14x (YE22), 1.27x (YE23), 1.35x (YE24), 1.73x (YE25) — today’s 1.96x is the richest of the post-crisis era. A standard justified-multiple test at a 10% cost of equity and 3% growth supports ~1.67x on the 14.7% RoTE actually earned, and ~1.84x on full delivery of the 2028 ambition under the current framework. Tangible book per share has meanwhile compounded at roughly 3–4% a year since FY2023 and was flat in 2Q26; total tangible-book return including dividends has run at ~8.1% a year since YE2023, below any defensible cost of equity. And the objective relative scoreboard — the one metric the compensation committee could not adjust — shows relative TSR over 2023–2025 running 17.45 percentage points below the G-SIB index.

This article takes no position and sets no price target. The framework verdicts that follow are: a structurally attractive but capital-taxed industry; a real but narrower moat than the “world’s largest wealth manager” label implies; high-quality asset gathering outside the US and poor economics inside it; genuine and improving financial quality with cyclical and accounting tailwinds that are running out; capital allocation that was excellent on the deal and is deteriorating on the buyback; and a valuation that embeds the good regulatory outcome.

2. Business Overview

2.1 What UBS is

UBS Group AG is a Zurich-headquartered global systemically important bank, founded 1862, reporting under IFRS in US dollars. It runs five reportable divisions plus Group Items. At 30 June 2026 it carried $1,707bn of total assets, $662.9bn of loans and advances to customers, $784.8bn of customer deposits, $503.9bn of risk-weighted assets and 112,388 internal and external personnel (99,085 internal FTE).

2.2 How UBS actually makes money

Group revenue in 2Q26 split $7,582m net fee and commission income (55%), $3,696m other net income from financial instruments at fair value (27%), $2,398m net interest income (18%) and $24m other. The business is majority fee-based, and that is the quality argument — but the fee base is levered to market levels, and the fair-value line is trading revenue.

Within Global Wealth Management the 2Q26 split is more informative:

The critical structural datum: only $2,255bn of GWM’s $4,942bn of invested assets are fee-generating — 45.6%. Mandate penetration is at a record and rising (MyWay discretionary assets exceed $40bn, +75% YoY), and net new fee-generating assets ran ~$70bn over the trailing twelve months. This is simultaneously the strongest evidence of pricing traction and the clearest statement of how much of the asset base still earns only transactional and spread economics. Recurring fee income is roughly 52% of GWM revenue; the balance is cyclical.

2.3 Segment economics, and the divergence that matters

Personal & Corporate Banking is widely assumed to be the group’s protected Swiss engine. It is not, and the evidence is unambiguous. 2Q26 pre-tax profit was CHF 676m (+21%) on an underlying cost/income ratio of 57.6% and an underlying return on attributed equity of 15.3%, from CHF 251.4bn of gross loans and CHF 252.6bn of customer deposits; UBS granted or renewed roughly CHF 40bn of Swiss loans in the quarter. The 2028 ambition is a ~48% cost/income ratio and ~19% return on attributed equity — and management has already conceded the sub-50% cost/income target will not be met by end-2026. Critically, P&C absorbs ~28% of group attributed equity ($21.4bn) — more than the entire Investment Bank ($18.4bn) — at the lowest divisional return in the group (15.3% against the IB’s 23.2%). At a zero Swiss policy rate, P&C is the group’s return-dilutive anchor, not its engine. the Swiss domestic section sets out why its competitive position is also weaker than the post-Credit-Suisse narrative implies.

The Investment Bank produced 2Q26 revenues of $3,727m (+26%) and pre-tax profit of $1,150m on $19.8bn of average attributed equity — a 23.2% return on attributed equity, against 12.2% in 2Q25. Global Markets delivered a record second quarter (+31%, Equities +53%), Leveraged Capital Markets revenue more than doubled, and Advisory fell 5%. Fixed income, rates and credit fell 21% as balance sheet was reallocated to equities. Management stresses this was achieved “without materially expanding our balance sheet.” The through-cycle ambition is ~15% return on attributed equity — i.e. the delivered 23.2% is roughly 8 points above what UBS itself thinks is normal.

Asset Management remains sub-scale in a scale industry: 2Q26 pre-tax profit of $237m underlying on ~$756m of revenue, with a 2028 ambition of a ~65% cost/income ratio and ~3% through-cycle net new money growth.

The GWM regional split is the single most important table in this article:

The Americas holds 49% of invested assets and 55% of fee-generating assets, employs 62% of the advisors, and earns roughly a third of APAC’s margin while contributing 2.5% of net new assets. Note what this is and is not. UBS’s GWM division in aggregate earns 26.5% reported / 28.5% underlying — broadly in line with Morgan Stanley Wealth’s 30% — so the “UBS wealth is low-margin” claim is false at the group level and true only regionally. The Americas margin is also improving: 9.3% (FY2024) → 12.7% (FY2025) → 13.7% (1Q26) → 15.9% (2Q26), the best in the series and at management’s own ~15% target. The unresolved problem is flows, not margin: net new fee-generating assets in the Americas have decayed from +$50.7bn (FY2024) to +$11.7bn (FY2025) to +$4.2bn (1H26), and on a pure net-new-money basis — UBS’s “net new assets” includes interest and dividends — GWM Americas lost $57.7bn of client money in FY2025 against a reported net-new-assets figure of −$5.9bn. This is the legacy PaineWebber / Wealth Management Americas problem, unresolved through two decades and now compounded by post-merger advisor attrition.

Verdict. UBS is a genuinely fee-weighted, genuinely diversified franchise with a dominant Swiss domestic position and a world-leading asset base — but “world’s largest wealth manager” describes assets, not economics. Roughly half those assets sit in a business earning a 16% pre-tax margin with no organic growth, and the divisional profit mix in 1H26 leaned harder on a cyclical Investment Bank than on the wealth franchise the equity story is sold on.

3. Industry Dynamics

UBS operates in three structurally distinct industries, and the article’s verdict differs in each. It is also, uniquely among its peer group, subject to a fourth variable that overrides all three: the jurisdiction that sets its capital denominator.

3.1 Global wealth management — a good industry at a late stage of the capital cycle

The pool is large and genuinely growing. Global financial wealth reached $333trn in 2025, +10.7%, with BCG projecting a ~7% compound growth rate to 2030. HNWI investable wealth (above $1m, excluding primary residence) was $98.3trn, +8.7% — the largest single-year rise since 2018 — across 25.3m individuals. The ultra-high-net-worth tier above $30m grew fastest for the second consecutive year (population +9.4%, wealth +9.7%) to roughly 250,000 people, and the top 1% of HNWIs now hold 34.8% of HNWI wealth. Applying BCG Expand’s benchmarked 66–72bps revenue margin to the client assets actually held at advised platforms implies a global revenue pool of roughly $390–455bn (ASSUMPTION, derived); a bottom-up sum of the sixteen largest disclosed franchises alone reaches ~$255bn.

But the single most important fact about that growth is what drives it. BCG: more than 80% of industry revenue growth in 2025 came from market performance, not from net new client money, and only 28% of wealth-manager asset growth over the past decade came from existing advisors (22% in mature markets). INTERPRETATION: the wealth-management profit pool is overwhelmingly a levered claim on global equity beta dressed as a recurring-fee annuity. That matters directly for how much of UBS’s 1H26 result should be capitalized.

Scale economics are real — and they are a cost advantage, not pricing power. BCG Expand’s benchmark of wealth managers by size is unambiguous:

A ~4x profitability gap rests on only a 6bps revenue-margin advantage — the differential is almost entirely cost. Front office is 40–45% of the cost base, and technology, at 30–35% of non-front-office cost, has outgrown revenue since 2020. Note also that net new asset growth falls as scale rises: the largest firms gather at 4% against 9% for the smallest. Scale buys margin and costs growth.

Fee compression is measurable, persistent, and currently masked. Computed from UBS’s own quarterly filings:

UBS’s fee-generating-asset margin has fallen from 86.2bps to 69.7bps in five years — a 19% decline. Morgan Stanley’s fee-based yield fell in lockstep, 88.2bps (2019) to 71.6bps (1H26 annualized), also −19%, and its revenue on total client assets fell 70.9 to 44.9bps (−37%) because the assets it is winning — workplace stock plans and self-directed accounts — monetise at a fraction of advisor-led rates. Julius Baer guides to a recurring gross margin of only 37–39bps by 2028 against a headline 87bps in 1H26. BlackRock’s effective base fee rate has gone from ~17–19bps a decade ago to 13.7bps; retail index ETFs charge 3–20bps.

The compression runs at roughly 1.5–3bps a year and is presently masked by two cyclical tailwinds: elevated transaction income (UBS’s underlying transaction-based income was +23% year on year in 2Q26) and deposit net interest income. Strip both and the recurring margin is grinding down. The industry is growing assets materially faster than it is growing revenue.

A disclosure flag the reader should note. UBS stopped disclosing GWM “gross margin on invested assets” after FY2020 and the “fee-generating asset margin” after FY2022 — in both cases while the metric was falling. Both KPIs survive in UBS’s Asset Management disclosure, so the omission is selective rather than a change of house style. The series above is therefore reconstructed by us from revenue and asset disclosures rather than taken from a UBS-presented KPI. (FACT on the withdrawal; INTERPRETATION on the motive.)

Consolidation is accelerating and priced at records. RIA transactions rose from 203 (2019) to 466 in 2025 (+27%), with 1H26 already at 262 and Q1-26 a record 142. The median EV/EBITDA multiple reached 11.6x in 2025, an all-time high and +40% since 2020, with premium platforms at 13–18x and Creative Planning/TPG reportedly ~23x. Private-equity-linked buyers accounted for 75.8% of Q2-26 deals, an all-time high. Cash at close has fallen from ~80% (2019) to 55–65% as equity consideration rises.

3.2 The Marathon capital-cycle read — capital has arrived, and it is bidding for advisors

Applying the capital-returns framework, every supply-side warning sign is flashing:

1. Record acquisition multiples with record financial-sponsor participation (11.6x, 75.8% PE).
2. New entrants and re-entrants. Citi Wealth targets 15–20% return on tangible common equity near term and above 20% medium term, from −2% in 2023, adding ~700 advisors. HSBC is redeploying $1.5bn out of Western investment banking into Asian wealth. JPMorgan grew Private Bank advisors 3,515 → 4,101 (+17% in two years) while total asset-and-wealth-management headcount rose only 1%.
3. Advisor-cost inflation — the decisive evidence. Recruiting deals have gone from ~200% of trailing-twelve-month production in 2018–19 to 300–500%, with UBS reportedly offering 550% of trailing production on a ~16-year commitment in March 2026. Capitalised recruiting books: LPL $3.3bn, up 1,300% since 2018; Morgan Stanley ~$4.86bn; Ameriprise $1.67bn; Raymond James $1.67bn. LPL’s advisor-loan book grew 8.3x from $441.7m (2019) to $3,681.5m (2025), with $1,751.7m of cash deployed into advisor loans in 2025 — three times its capital expenditure — driving negative $411.4m of operating cash flow.
4. Compensation ratios grinding up. Raymond James firmwide moved 62.8% → 64.5% (FY23–FY25), explicitly attributed in part to recruiting expense, with pre-tax margin falling 20.6% → 19.3% despite 10% revenue growth. UBS grew financial-advisor compensation 6.8% to $5,654m while cutting Group headcount 5% — and still lost 243 advisors in 2025.
5. Advisor mobility at a four-year high: 11,172 experienced advisors changed firms in 2025 (+16%), including 54 teams managing $1bn or more. Wirehouses were net −302; LPL +2,112.

INTERPRETATION: wealth management has been sold to capital as “capital-light, fee-based and recurring,” and capital has duly arrived. Because the industry has no physical capacity to build, the excess capital cannot express itself as capital expenditure — it bids up the price of the scarce input, which is the advisor. A 550%-of-revenue recruiting package is wealth management’s shipyard order book. The result is compression at the top line and inflation in the compensation line simultaneously, and the rent accrues disproportionately to labour rather than to shareholders. The characteristic 50–65% compensation ratio is the price of that.

The one contrarian tell worth recording: DeVoe’s May-2026 consolidator survey found 0% of buyers expecting prices to rise and 18% expecting declines — the first negative skew in the survey’s history, alongside a flagged “cycle of mega-recapitalizations as private-equity holding periods mature.”

3.3 Swiss domestic banking — the dominance is a myth

The most persistent misconception about post-Credit-Suisse UBS is that it enjoys a domestic near-monopoly. The Swiss National Bank’s own data says otherwise. At end-2025, domestically focused banks held 73% of domestic loans and 69% of deposits; the three domestic systemically important banks alone — PostFinance, Raiffeisen and Zürcher Kantonalbank — held 26% of loans and 31% of deposits, against UBS’s 23% and 23%. The domestic SIBs out-weigh UBS on both measures. On mortgages, cantonal banks hold 40.4%, UBS 22.7% and Raiffeisen 18.6%. The derived mortgage Herfindahl index of roughly 1,010 is below the threshold for “moderately concentrated.”

And UBS is losing share, not gaining it:

UBS’s mortgage book has been flat at roughly CHF 283bn for two years while the market added CHF 36bn — growth of +0.1% in 2025 against Raiffeisen’s +4.6% and the cantonal banks’ +3.9%. Most damningly, approximately 82% of the deposits withdrawn from Credit Suisse by domestic clients went to cantonal banks, not to UBS. The acquirer did not capture the domestic spoils.

The structural reason is a state subsidy. Twenty-one of Switzerland’s twenty-four cantonal banks carry an explicit, unlimited state guarantee, worth roughly CHF 585m a year of funding benefit against ~CHF 165m of compensation, plus ~CHF 190m a year of tax exemption, and a ratings uplift of about three notches — Zürcher Kantonalbank is rated AAA/Aaa. INTERPRETATION: in a commoditised, fully collateralised Swiss-franc product where price is the only real differentiator, UBS competes against AAA state-backed balance sheets. That is a structural handicap, not a moat, and it is a textbook Marathon case of a capital cycle that cannot clear because the state guarantees the competition.

Political consequence has, so far, been mild: the competition commission’s 2023 opinion found competition sufficient outside narrow pockets, and FINMA closed merger control on 19 June 2024 with zero conditions. No abuse-of-dominance investigation exists; the live instrument is the Price Supervisor’s observation regime since July 2024 — monitoring, not a cap.

3.4 Investment banking — a good oligopoly at an unambiguous cyclical high

LSEG data for 1H26 show global announced M&A of $2.85trn, +50% year on year — a record first half and the highest since records began in 1980. But the composition is unusual: deal count fell 9% to roughly 24,000, a six-year low, and 48 deals of $10bn or more accounted for ~46% of all value, an all-time concentration record. The fee wallet was $79.9bn (+17%), with equity capital markets +60%; IPO value more than tripled on a 3% decline in IPO count, because SpaceX alone raised $85.7bn on 12 June 2026. Excluding it, IPO value is ~$85–120bn: strong, but not a boom. These are concentration-inflated, breadth-deficient volumes.

Peer disclosure converges independently on “cyclical peak”: Goldman Sachs’ own reporting sizes the fee pool at $80–110bn normal, ~$130bn+ at peak and ~$70bn in trough, and calls the recent period “a tooth-top, not a mid-cycle point”; Morgan Stanley’s management declined to raise targets into a record quarter; Evercore reported +49% to ~$4.5trn of announced volume in 2025; Barclays’ investment bank earned 10.6% return on tangible equity on 45% of income; and HSBC has been exiting Western M&A and ECM as sub-scale.

UBS’s investment bank is at a cyclical high on every measure — revenue from $8,703m (FY23) to $12,340m (FY25) to $7,781m in 1H26 alone; pre-tax profit from $(72)m (FY23) to $2,355m (1H26); return on attributed equity of 24.0% in 1H26 against UBS’s own ~15% through-cycle ambition (1.6x); cost/income from 76.1% to 68.3%. It delivered 32% of group pre-tax profit in 2Q26 on 22.7% of group risk-weighted assets, and is hard-capped at 25% of Group RWA — headroom of only ~2.3 points, shrinking from both ends as Non-core’s residual deflates the denominator.

The compositional warning is important. In 2Q26 UBS’s Advisory revenue fell 5% year on year while Goldman’s investment-banking fees rose 55%, Morgan Stanley’s 58%, Bank of America’s 50%, Citi’s 44% and JPMorgan’s 30%; UBS’s fixed income, rates and credit fell 21% while every US peer’s FICC grew. European banks’ share of the global fee wallet has gone from 29% in 2015 to 20% in 1Q26, the lowest since 2000. INTERPRETATION: UBS is earning peak returns in the financing-led book — equities, prime brokerage, derivatives — while its advisory franchise goes sideways in the best fee half on record. It is not on the megadeals that made 46% of value, because balance-sheet-led origination favours the US bulge bracket and UBS’s balance sheet is deliberately capped. Normalising IB return on attributed equity to ~15% removes roughly $0.9–1.0bn of annualized pre-tax profit — about 12–13% of group pre-tax profit. OPEN QUESTION: is the −21% FICC print reallocation of balance sheet, or share loss?

3.5 The rate environment — and a genuinely non-obvious asymmetry

The Swiss National Bank policy rate is 0.00%, held at both the March and 18 June 2026 assessments; it has not returned to negative territory. The path ran −0.75% to a 1.75% peak in June 2023, then 175bp of cuts in fifteen months to zero since 19 June 2025. Sight deposits above a threshold earn a 0.25-point discount (an effective −0.25% on excess reserves). June forecasts were revised up and the SNB now flags that “inflation risks have increased.”

The most important rate disclosure in the file comes from UBS’s own FY2025 Annual Report: a +100bp shift adds roughly $1.4bn of annual banking-book net interest income (of which $1.1bn is Swiss franc), while a −100bp shift also adds roughly $0.9bn, with the Swiss franc contributing +$1.2bn from “contractual and assumed flooring benefits under negative interest rates.”

INTERPRETATION: zero is the pessimal point for Swiss net interest income, and the SNB has parked exactly there. Any exit — in either direction — is net-interest-income-positive on UBS’s own numbers. This is a real and under-appreciated asymmetry in UBS’s favour and it belongs in the bull case.

The damage to date is visible: P&C net interest income fell from CHF 4,987m (FY24) to CHF 4,403m (FY25), −11.7%, troughing at CHF 938m in 3Q25 and recovering to CHF 1,061m in 2Q26 — +13% off the trough, on volume rather than rate. Net interest margin sits at 170bp against 180bp. Sector-wide, domestically focused banks’ margin fell to 0.94% from 1.02%.

On the dollar and euro side, both central banks are now tightening-biased: the Fed held at 3.50–3.75% on 29 July 2026 nine-to-three with three dissents to hike, and the ECB raised its deposit rate 25bp to 2.25% effective 17 June 2026. GWM net interest income troughed at $1,655m in 3Q25 and reached $1,863m in 2Q26 (+12% YoY), guided to ~+10% for FY2026. The deposit mix-shift has run its course and reversed — UBS now reports inflows into low-cost current and savings accounts and outflows from fixed-term deposits, a margin-accretive rotation. UBS’s net interest income is a two-speed franchise: GWM in dollars (~$7.7bn run-rate, growing 10%) levered to a Fed that is not cutting, and P&C in francs (~CHF 4.4bn, flat) pinned at zero and growable only by volume in a market where UBS is losing share.

3.6 The overriding structural fact is regulatory

Barriers to entry protect nobody here. What determines returns in this industry, for this company, is which jurisdiction sets the capital denominator — and the jaws are opening in the wrong direction. Switzerland implemented final Basel III on 1 January 2025, ahead of the US, EU and UK. The United States moved the other way: the enhanced supplementary leverage ratio recalibration took effect 1 April 2026, and on 19 March 2026 the Federal Reserve, OCC and FDIC issued three proposals overhauling US capital rules and formally rescinding the 2023 Basel III Endgame framework, with a cumulative effect of reducing required Tier 1 capital by 5.6% to 7.9%. Swiss requirements are rising toward a de facto ~17.6–18.4% Group CET1 minimum on UBS’s calculation while US G-SIB requirements fall. the changes section sets out the legislative state of play in full.

3.7 Verdict — structurally good or bad?

A split verdict, and the split is unfavourable to UBS specifically.

Wealth management: good but deteriorating. An enormous, genuinely growing, asset-light pool — but with documented fee compression of 1.5–3bps a year, more than 80% of 2025 revenue growth coming from market performance rather than new money, customer captivity residing with the advisor rather than the firm, and a textbook late-stage capital influx whose only outlet is bidding up advisor compensation. High returns are attracting capital, and the capital is competing them away through the cost line.

Swiss domestic banking: structurally poor for UBS. Fragmented (Herfindahl ~1,010), commoditised, price-competitive and distorted by twenty-one state-guaranteed competitors, with UBS losing roughly 0.7–1.5 points of share a year in its core products at a policy rate that pins the liability margin at zero.

Investment banking: a good oligopoly at a cyclical peak — and UBS is in the wrong half of it, over-earning at 24% return on attributed equity in the financing book while its fee franchise goes sideways in the best fee half on record, with European wallet share at a 26-year low.

Overall: the industry is good enough to earn a mid-teens return; the jurisdiction decides whether UBS keeps it. That is an unusual verdict for a business of this quality, and it is the reason the regulatory section of this article carries the weight it does.

4. Competitive Position

The question this section answers is narrow and unforgiving: is there a mechanism that would cause UBS’s financial outcomes to deteriorate if it disappeared? Applying Greenwald’s taxonomy, the honest answer is that UBS has one genuine barrier (economies of scale, which is a cost advantage rather than pricing power), one real but geographically confined source of customer captivity (ultra-high-net-worth and cross-border), and no durable advantage at all in the two businesses that consume most of its capital.

4.1 Where UBS actually ranks

UBS’s “world’s largest wealth manager” claim depends entirely on the definition selected. On client assets:

(Definitions are not strictly comparable — Schwab and Morgan Stanley include self-directed and workplace assets.) UBS is fourth on this measure. Its defensible claims are narrower and still meaningful: the largest pure-play global wealth manager, and #1 by Euromoney’s private-banking ranking, with the deepest UHNW and cross-border franchise outside JPMorgan’s Private Bank. Even the largest player holds ~13% of HNWI investable wealth, and the top ten hold ~47% on non-comparable definitions. This is a fragmented industry with a concentrated top tier, not an oligopoly.

4.2 The Greenwald tests, run honestly

Customer captivity — WEAK at the firm level, and weakening. Capgemini’s data is the clearest evidence: the proportion of HNWIs working with a single firm collapsed from 39% in 2019 to 19% in 2025, and 88% now use multiple firms specifically for alternatives access, with $1.5trn flowing to non-traditional competitors between 2022 and 2025. The decisive counter-evidence to any firm-level captivity claim is portability: the fact that firms pay 300–550% of trailing-twelve-month production to move a book proves that the client relationship attaches to the advisor, not to the institution. UBS’s own experience proves it in both directions — it lost at least 27 teams managing roughly $28bn in 1H26 (against 24 teams and $16.8bn in 1H25) to Wells Fargo, RBC, Rockefeller and Merrill, while simultaneously recruiting teams into Tampa and Austin.

The instructive contrast is Morgan Stanley, which has manufactured firm-level captivity where none naturally exists, by originating relationships institutionally through its Workplace and E*TRADE funnel — more than $400bn of advisor-led assets sourced from that channel since 2020. UBS has no equivalent. Its client acquisition runs through the advisor, which is precisely the asset it does not own.

Captivity does exist, genuinely, at the top of the wealth spectrum. In UHNW and cross-border, search costs are high, the relationship is institutional rather than personal (family-office structures, multi-jurisdictional booking, lending against concentrated or illiquid holdings), and the balance sheet is part of the product. This is UBS’s real moat, and it is why APAC and EMEA earn 45% and 38% pre-tax margins. It is not the mass-affluent US wirehouse book.

Economies of scale — the industry has them; UBS does not capture them. BCG’s benchmark shows 27bps of pre-tax margin above $100bn of client assets against 7bps below $20bn — a fourfold profitability gap resting on a 6bps revenue-margin advantage. UBS is unambiguously above that threshold, with $4.94trn of GWM invested assets, a global booking-centre network, and a differentiated capability in structuring, cross-border lending and multi-jurisdictional compliance that a CHF 100bn Swiss private bank cannot replicate.

But run the test properly and it fails. UBS GWM’s underlying pre-tax margin reached 28.5% in 2Q26 (26.5% reported) — broadly in line with Morgan Stanley Wealth’s 30%, on a wealth asset base roughly 60% of Morgan Stanley’s. GWM’s cost/income ratio ran 83.2% / 84.1% / 79.8% across FY2023–25 and 73.6% in 2Q26. The fair statement is therefore not that UBS is uniquely inefficient — it is that the largest wealth manager in the world has no cost advantage over a competitor half again its size. Greenwald is unambiguous that scale is an advantage only where it shows up in cost per unit in the relevant market, and the relevant market here is each booking centre and client segment, not “global wealth.” UBS has size, not scale. Per Greenwald, scale without captivity is not a barrier to entry in any case — and UBS shares what scale it has with Morgan Stanley, Bank of America, Schwab, Fidelity and JPMorgan.

And the revenue mix is lower-quality than the peer it is measured against. Only about 52% of GWM revenue is recurring (2Q26: recurring net fee income 52.2%, net interest income 26.2%, transaction-based 21.3%), against Morgan Stanley Wealth’s 58.7% asset-management fees in FY2025. Nearly half of UBS’s wealth revenue is rate-driven or activity-driven. The recurring share has risen only 1.2 points in three years, and the improvement came from net interest income falling as the SNB and Fed cut, not from fees winning.

Per-unit pricing is eroding roughly 3.9% a year on the fee-paying base. UBS masks that by converting more invested assets into fee-generating assets — penetration up from 42.4% in FY2023 to 45.6% — which is a genuine lever but a finite one. Asset Management is the uglier picture: gross margin on invested assets has gone 19bps (FY2023) → 16bps (FY2025) → 14bps (2Q26), −26% in under three years, with $1,040bn of $2,005bn (51.9%) now in passive strategies and $601bn (29%) sourced from UBS’s own wealth business rather than won in the open market. That mirrors BlackRock’s ~13.7bps effective base fee — but BlackRock has the scale to defend it and UBS Asset Management does not.

Market-share stability — FAILS decisively. Greenwald’s test is that a genuine barrier produces stable shares over time. From 2019 to 2026, Morgan Stanley Wealth client assets grew +199%, LPL +241%, and Schwab to $13.1trn — while UBS GWM grew +88% including the entire Credit Suisse wealth book, and Julius Baer +28%. Shares have moved by far more than the five points Greenwald sets as the failure threshold. Much of the movement came through acquisition — E*TRADE, TD Ameritrade, Credit Suisse, Commonwealth — which is itself the point: in this industry share is bought, not earned.

Cost advantage — real in Switzerland, structurally impaired. UBS’s Swiss cost base is now materially lower post-integration, but the Swiss domestic section established that it competes there against twenty-one state-guaranteed cantonal banks funding themselves at AAA, and has lost 4.2 points of mortgage share and 6.4 points of non-mortgage lending share since 2021.

The returns test — FAILS on history, passes only from 2026. Greenwald’s diagnostic is whether the claimed advantage shows up in returns above the cost of capital. UBS’s underlying return on tangible equity was 4.1% (FY2023), 8.5% (FY2024) and 12.1% (FY2025) — a three-year average of roughly 8.3%, which sits inside Greenwald’s “6–8%: advantages absent” band — against a 10–11% cost of equity. On that record UBS destroyed value in FY2023 and FY2024 and only crossed the hurdle in FY2025. The 1H26 run-rate of 14.4% underlying RoTE finally clears it. INTERPRETATION: a business that averaged ~8% returns over three years does not, on this framework, have demonstrated competitive advantages — it has an accounting event and an integration bill. The 2026 numbers are the first evidence to the contrary, and they are one half-year old.

The government-granted barrier — real, but the rent is expropriated. FINMA, not the competition commission, cleared the merger under Article 10(3) of the Cartel Act on 19 June 2024 without conditions or divestitures. The Swiss state granted UBS the barrier. It is simultaneously taxing away the rent through the too-big-to-fail regime — and UBS itself confirms the linkage by conditioning its buyback pace on “further visibility on the deliberations by the Swiss Parliament on the capitalization of foreign subsidiaries.” A barrier to entry whose economic rents accrue to the regulator is not a moat for the shareholder.

4.3 The Credit Suisse question — what was actually acquired

UBS bought a distressed franchise at roughly 0.12x tangible book, and the acquisition unambiguously delivered scale: Group invested assets went from $3,957bn (FY2022) to $7,326bn — a rise of 85%, and the Swiss universal-banking field went from four large players to one. It also delivered a permanent cost annuity — $12.6bn of gross savings realised against a $13.5bn ambition, with more than 90% of cost synergies now booked and 90% of legacy applications retired.

What it did not deliver is a higher return on capital or captured domestic share. Underlying return on CET1 capital of 16.7% remains marginally below the 17.0–17.5% standalone UBS earned in FY2020–22. Roughly 82% of the deposits domestic clients pulled from Credit Suisse went to cantonal banks rather than to UBS. And in the Americas, the combination has coincided with accelerating advisor attrition rather than consolidation benefits.

Two disciplines on this section. First, UBS has never disclosed a Credit Suisse asset- or revenue-retention percentage. Figures circulating in the market trace only to low-credibility sources and are not used here. Second, the widely-repeated claim that independent Swiss private banks systematically harvested ex-Credit-Suisse relationship managers and assets could not be verified from any primary source and is therefore not asserted, notwithstanding that Julius Baer reached a record CHF 547bn of assets under management and record first-half profit.

INTERPRETATION, through the Marathon lens: this was a genuine distressed purchase rather than an asset-growth-anomaly acquisition — the price was a fraction of book, the seller was a failing competitor, and consolidation reduced industry capacity. That is the favourable half of the capital cycle and it is why the deal was correct. But the supply-side victory was captured by the Swiss cantonal banks on the domestic side and, so far, taxed by the Confederation on the international side. UBS won the consolidation and someone else is collecting a large share of the rent.

4.4 Direct comparison with the key competitors

The Morgan Stanley comparison is the one that matters, and it is unflattering in the specific place it counts: MS earns roughly 7 points more return on tangible equity, in a regulatory regime moving in its favour, with a wealth business earning a 30% pre-tax margin against UBS GWM’s 26.5% (and against the Americas’ 15.9%), and it has built the institutional client-acquisition funnel UBS lacks.

One contrarian correction, against the received view. The long-standing criticism that UBS’s Investment Bank is a capital-consuming appendage management keeps promising to shrink is stale on the numbers. FY2025 IB return on attributed equity was 15.3% reported / 14.8% underlying — marginally above GWM’s 15.2%. In 2Q26 the IB earned 23.5% underlying against GWM’s 21.8%, contributing 30% of group underlying pre-tax profit on 22.7% of risk-weighted assets, inside its self-imposed 25% cap, with IB risk-weighted assets actually falling $2.0bn while revenues rose 31%. The correct criticism is not that the capital is idle — it is that these are peak-cycle trading returns being capitalized as though structural (the investment-banking section), and that the IB is the direct cause of the foreign-participation capital charge that caps the buyback.

4.5 Verdict

UBS has a real but narrower moat than its scale implies, and it does not extend to the businesses that consume most of its capital.

The durable advantage is genuine and worth owning: global UHNW and cross-border wealth management, where search costs, multi-jurisdictional capability and balance-sheet lending create real institutional captivity, evidenced by 45% and 38% pre-tax margins in APAC and EMEA and by net new asset growth of 4.7% and 6.2%. Combined with a fourfold scale cost advantage over sub-$20bn managers, this is a defensible franchise.

But three of Greenwald’s four tests fail. Firm-level customer captivity is weak and deteriorating industry-wide (single-firm relationships down from 39% to 19%), and demonstrably resides with the advisor — proven by UBS losing 27 teams and ~$28bn in a single half-year, and by the 550%-of-production packages the industry pays. Market-share stability fails outright. And in Swiss domestic banking, UBS is a share-losing 23% player competing against state-guaranteed balance sheets, not a monopolist.

Applying this article’s own standard — if a “moat” claim cannot be tied to a financial outcome that would deteriorate without it, it is not a moat — UBS passes in UHNW and cross-border and fails in the US mass-affluent book and Swiss retail. Since GWM Americas holds 49% of wealth invested assets and P&C absorbs 28% of group attributed equity, roughly half the company sits outside the moat.

The honest label: a two-speed business wearing one name. In Switzerland, APAC and EMEA UBS has genuine customer captivity — cross-border, credit-linked, custodial relationships with high search and switching costs — and the returns prove it, at 33–45% regional pre-tax margins on 4.3–9.4% organic growth. Layered on top is a government-granted Swiss barrier whose rents the state is expropriating through a ~$22bn capital surcharge, and which is eroding anyway as cantonal banks take mortgage share. What UBS does not have is the advantage its scale implies: no cost advantage over Morgan Stanley, organic growth below Schwab, Goldman’s wealth arm, Ameriprise and BlackRock, a wealth revenue mix only ~52% recurring, a fee margin on fee-paying assets compressing ~3.9% a year, an asset manager that has lost 26% of its fee rate in three years while becoming half-passive and 29% internally funded, and a US franchise — half its wealth assets — that earned a 12.7% margin on negative net new money in FY2025.

This is not a bad business. But it should be called what it is: a scale player in a structurally competitive industry, with a real but geographically confined captivity moat, a regulator-owned Swiss barrier, and a returns record that has only just stopped destroying value. The cost-cycle story reaches its terminus when the integration closes at end-2026; from there the growth must come from fee penetration (45.6% and rising against a finite ceiling) and from fixing an American business UBS has not fixed in twenty-five years.

5. Growth History and Forward Opportunities

5.1 The growth record

Almost all of the reported revenue growth between FY2022 and FY2024 was acquired, not organic. Revenue rose from $34.6bn to $48.6bn across the Credit Suisse consolidation; the organic component is best seen in the FY2024→FY2025 step of $48.6bn to $49.6bn (+2.0%) and then the 1H26 acceleration to a $55.9bn annualized run-rate (+12.7%) — the latter driven substantially by a record Investment Bank half and by market-level appreciation of the asset base.

Invested-asset growth is overwhelmingly market-driven, not flow-driven. Group invested assets rose 10.7% year on year to $7,326bn, while GWM net new assets ran at 3.0% annualized. That gap is the industry norm — BCG found more than 80% of 2025 industry revenue growth came from market performance rather than net new money — but it means the growth line is a levered claim on equity markets rather than evidence of share gain. This is the single most important qualification on UBS’s growth quality.

5.2 Quality of the flows

The flow picture divides cleanly by geography and by asset type:

Outside the United States, UBS is genuinely gathering assets at 4.7–6.5% annualized with strongly positive mandate flows and record mandate penetration. Inside the United States — 49% of the asset base — it is barely growing. The precise claim matters: the Americas margin is improving (9.3% FY2024 → 15.9% in 2Q26, on target), and 1H26 net new fee-generating assets were still positive at +$4.2bn, with 2Q26’s −$4.6bn following +$8.8bn in 1Q26 and reflecting roughly $10bn of seasonal tax-related outflows. What is deteriorating is the rate of fee-generating gathering — +$50.7bn (FY2024) → +$11.7bn (FY2025) → +$4.2bn (1H26) — and, most starkly, the underlying money flow: on a net-new-money basis (stripping the interest and dividends that UBS’s “net new assets” definition includes), GWM Americas lost $57.7bn of client money in FY2025. Group-level transaction-based income grew 23% year on year, its twelfth consecutive quarter of double-digit growth, but transaction income is the lowest-quality revenue line and its strength is a function of client activity levels, not of relationship depth.

The genuine structural growth lever is mandate penetration: only 45.6% of GWM invested assets are fee-generating. Every point of conversion moves revenue from transactional to recurring at a materially higher and more durable margin. Management reports record penetration across all regions, with APAC up 5 points year on year and MyWay discretionary assets exceeding $40bn (+75%). This is real and it is working — everywhere except the Americas, where net new fee-generating assets are negative.

5.3 Forward opportunities, tested against the targets

UBS’s 2028 divisional ambitions are more than $5.5trn of GWM invested assets, more than $200bn of net new assets per annum, a ~68% GWM cost/income ratio, ~48% for P&C, ~65% for Asset Management, and ~15% through-cycle return on attributed equity for the Investment Bank.

• Invested assets: a soft target. $4,942bn today against “>$5.5trn by end-2028” requires only ~11% over two and a half years — achievable on market drift alone. This should not be read as evidence of share gain.
• Net new assets: the hard target, and it depends on the US. 1H26 annualizes to roughly $146bn against a >$200bn ambition — a required step-up of ~37%. With the Americas contributing 0.2% annualized growth and negative fee-generating flows, that arithmetic requires either a genuine US inflection or a disproportionate acceleration in APAC, EMEA and Switzerland from already-strong 4.7–6.5% rates.
• Cost/income: partly conceded. Management has already acknowledged the P&C sub-50% cost/income target will not be met by end-2026.
• Rate optionality: genuinely favourable. As the rate section established, UBS’s own disclosure shows both a +100bp and a −100bp shift add net interest income. Zero is the worst point and the SNB is sitting on it.
• Operational-risk harmonization: the underrated option. The CFO has argued that Switzerland’s application of the operational-risk internal loss multiplier produces materially higher risk-weighted assets than the UK, EU and US implementations, where the multiplier is expected to be set at 1 — worth “$40 billion lower risk-weighted assets.” That is worth more than the parliamentary compromise currently being debated, and it is not in consensus numbers.

5.4 Verdict — high- or low-quality growth?

Mixed, and lower-quality than the headline. The durable, high-quality growth is real but geographically confined: mandate penetration rising from a 45.6% base, 4.7–6.5% net new asset growth in APAC, EMEA and Switzerland, and genuine UHNW and cross-border share gains at 38–45% pre-tax margins. Against that, the reported revenue growth of the last three years is overwhelmingly acquired; the invested-asset growth is overwhelmingly market-driven (>80% industry-wide); the fastest-growing revenue line is the lowest-quality one (transaction-based, +23%); the recurring fee margin is compressing 1.5–3bps a year; and the largest region by assets is contributing essentially nothing and shedding fee-generating assets.

The 2028 net-new-asset ambition is the honest test of whether this is a growth business or an asset-appreciation business, and on current run-rates it requires a 37% step-up that must come from the one region that is not delivering. Watch net new fee-generating assets in the Americas, not headline invested assets.

6. Financial Quality

6.1 The multi-year picture — reported and underlying

Trailing-twelve-month diluted EPS is $2.94 (2.36 + 1.81 − 1.23), placing the shares at ~17.9x trailing earnings. Annualizing 1H26 underlying pre-tax profit at a 23% tax rate gives normalized underlying EPS of ~$3.96 — ~13.3x.

6.2 Normalizing the distortions — five of them, and they do not all cut the same way

The negative goodwill. FY2023’s gain — provisionally $28,925m, restated to $27,264m as purchase accounting was finalized — makes that year’s reported 36.9% return on equity meaningless; UBS’s own underlying RoTE for FY2023 was 4.1%. Tangible book per share went $16.28 (YE22) → $24.34 (YE23); at the then share count the gain alone is roughly $7.9 per share. Essentially the entire FY2023 tangible-book step-up is purchase accounting, not compounding. (FACT on the figures; INTERPRETATION on the characterization.)

Purchase price allocation accretion — a real tailwind that is visibly expiring. Accretion of acquisition-date fair-value discounts added $824m to 1H26 revenue against $1,170m in 1H25 — down 30% year on year — of which $613m sits in net interest income. That is roughly 13% of Group 1H26 net interest income that is acquisition accounting and amortizes to zero. The trajectory (2,877 → 1,892 → ~1,650 annualized) implies two to three more years of materially shrinking accretion. UBS discloses no forward run-off schedule; this is an open question. The liability side unwinds as cost: in 4Q25 UBS repurchased $7.7bn principal of legacy Credit Suisse senior debt for ~$8.5bn and booked a net loss of $457m “including the release of purchase price allocation adjustments of USD 427m.”

Two FY2025 flatterers — and here the bias runs bullish. FY2025’s reported $7,767m was helped by roughly $1.0bn of abnormal tax benefit (an 11.9% effective rate; 2Q25 carried a negative 9.5% rate) and $949m of net litigation-provision releases — together about a quarter of the figure. 1H26 contains neither: a 21.1% tax rate and a small net litigation expense. Normalized to 23% tax, 1H25 net profit would be ~$3,330m and 1H26 ~$5,724m — +72%, not the reported +43%. This is the rare case where the headline understates the operating improvement, and the article credits it. Management concedes the corollary: on its own arithmetic, FY2025’s 13.7% underlying RoCET1 becomes 11.5% excluding litigation releases and at a normalized tax rate, and 3Q25’s headline 16.3% becomes 12.7%.

6.3 Where the improvement actually came from

The Investment Bank supplied 44% of the entire underlying pre-tax improvement, lifting its share of Group underlying profit from 23.2% to 30.2%, on a 31% revenue increase and a pre-tax margin that went from 20.9% to 30.9%. Its return on attributed equity went from 12.2% to 23.2% in a year, against a ~15% through-cycle ambition. INTERPRETATION: the 2026 return inflection is, more than anything else, an Investment Bank trading result — the most cyclical and least capitalizable earnings stream in the mix, earned with Group invested assets at a record $7,326bn.

The cost story does not survive contact with the numbers. Underlying operating expenses ran $34,061m (FY23) → $36,432m (FY24) → $35,595m (FY25) → $37,846m annualized in 1H26 — up 6.3% on FY2025 and 5.6% year on year — even as internal headcount fell 12.2% from 112,842 to 99,085. The $12.6bn of “cumulative gross cost savings” is a gross figure measured against a frozen 2022 combined baseline; management has always guided to net saves of “around 75% of that amount,” and the balance is consumed by inflation, variable compensation on higher revenue, technology investment and currency. The entire underlying cost/income improvement from 76.4% to 70.1% is revenue-led: underlying revenue +15.1% against operating expenses +5.6%, jaws of +9.5 points. Positive jaws are real operating leverage — but leverage powered by a trading boom and record asset levels is cyclical, and it reverses.

A further reversible item: management upgraded FY2026 GWM net interest income guidance from “low single digits” (4Q25) to “around 10% versus 2025” (2Q26), citing “higher US dollar rates than previously assumed.” That is a rate windfall, not a structural gain.

6.4 Capital, balance sheet and asset quality

Deposits fell ~$46bn (−5.8%) and loans ~$60bn (−9.4%) through 2024 — the Credit Suisse attrition and Non-core wind-down year — and have since rebuilt above the 2023 combined level. Deposits have been broadly flat for eighteen months while the loan book is up 14.3% since YE24, which is where the net interest income growth is coming from. Liquidity buffers have been steadily run down (LCR 215.7% → 177.3%; NSFR 124.1% → 115.1%, the tightest in the series though comfortably above 100%) as excess cash is redeployed into lending. INTERPRETATION: a legitimate net interest income tailwind and a legitimate reduction in the safety margin — both real.

Credit is running at roughly half a sensible through-cycle charge. Six basis points on a $663bn book, with credit-impaired exposure of $7,689m against expected-credit-loss allowances of $3,189m. Normalizing to ~20bps would cost ~$0.9bn of annual pre-tax profit — about 7% of underlying pre-tax profit. Not thesis-breaking, but current returns are not clean of a benign credit environment.

6.5 Returns versus the cost of capital

ASSUMPTION: cost of equity ~10% in USD — a US 10-year around 4.2–4.3%, a measured beta of 1.036, and an equity risk premium of ~5.5%. A 9–11% range is defensible; UBS’s 26% realized volatility and ~44% idiosyncratic variance share argue against the low end.

Does UBS earn its cost of capital? On an underlying basis: no in FY2023 (4.1%), no in FY2024 (8.5%), marginally in FY2025 (12.1%), and yes in 1H26 (14.4%). Across the whole post-acquisition period it did not. The confirming market test is unforgiving: from YE2023 to 2Q26 tangible book per share went $24.34 → $26.89 while dividends of $0.70 + $0.90 + $1.10 were paid — a total tangible-book return of +21.6%, or roughly 8.1% a year, below cost of equity.

And the pre-deal comparison remains uncomfortable. Standalone UBS earned RoCET1 of 17.5% (FY21) and 17.0% (FY22); 1H26’s underlying 16.7% is still marginally below, after $14.2bn of integration spend and on a 58% larger risk-weighted asset base. In fairness: FY2021–22 was a capital-markets boom; Switzerland’s final Basel III implementation (effective 1 January 2025) inflates today’s RWA denominator by roughly $60bn on management’s telling; the absolute profit pool is far larger; and Non-core and Legacy still carries a $149m 1H26 drag that disappears.

6.6 Book value, share count and cash conversion

Shares issued 3,277,805,164 less treasury 216,063,015 = 3,061,742,149 outstanding, down 3.4% year on year, with 63.8m cancelled in 2Q26. Tangible equity is $82.3bn; goodwill and intangibles are only ~$6.8bn — itself a direct consequence of the negative goodwill. At $52.74: P/TBV 1.96x, P/B 1.81x.

Cash conversion is the strongest part of the file. The 1H26 equity walk runs: opening $90,213m; treasury purchases $(3,337)m; dividends $(3,404)m (the $1.10 paid 23 April 2026 — this answers why equity fell despite a profitable quarter); share-based compensation +$652m; treasury delivered +$159m; obligation to purchase own shares $(50)m; total comprehensive income +$4,839m (net profit $5,840m less $1,001m of other comprehensive loss, being $642m of currency translation and $442m of cash-flow hedges); closing $89,165m. UBS distributed $6,741m against $5,840m of earnings — a 115% payout — and yet CET1 capital still rose $1,202m while risk-weighted asset growth consumed ~$1,500m. Reported profit converts to regulatory capital at more than 100%. There is no accrual-versus-cash divergence here. (A bank’s operating cash flow is dominated by balance-sheet movement and is not a meaningful free-cash-flow proxy; CET1 generation is the correct sector analog, and it is clean.) Share-based compensation expensed was $652m, ~2.3% of revenue — not distorting.

But per-share book is not compounding. Tangible book per share went $26.93 → $26.89 in the half despite $5.8bn of earnings, because buybacks executed above tangible book are TBVPS-dilutive by construction. At 1.96x that trade is getting worse, not better (the capital-allocation section).

Verdict: do the economics improve with scale? Partially, and less than the headline claims. The genuine structural gains are real — a permanently lower headcount, a retired legacy technology estate, more than 90% of cost synergies realized, and a Non-core division that has released 67% of its risk-weighted assets. Distributable-capital conversion is excellent and the balance sheet is genuinely strong (CET1 14.4%, TLAC 38.4%, LCR 177%). Against that: the absolute underlying cost base is rising, so the cost/income improvement is revenue-led and therefore cyclical; 44% of the profit inflection came from a peak trading quarter; ~13% of net interest income is expiring purchase accounting; credit is running at half a normal charge; the tax rate is normalizing from 11.9% toward 23%; and after four and a half years — including a $27.3bn accounting windfall — tangible book per share has compounded below the cost of equity. The improvement in returns is part durable operating leverage and part cyclical and accounting windfall, arriving precisely as the accounting tailwinds run out.

7. Capital Allocation

7.1 The Credit Suisse acquisition, netted out

What was paid. 3,949m Credit Suisse shares at 1 UBS share per 22.48 CS shares = 176m UBS shares at CHF 18.35 = CHF 3,223m / $3,547m; plus $162m of assumed share-based compensation and $135m to settle pre-existing relationships = $3,845m of total consideration. UBS acquired $108,510m of cash and central-bank balances in the transaction.

What was received. Negative goodwill of $28,925m provisional, restated to $27,264m as purchase accounting was finalized.

What the state support cost. The CHF 9bn Loss Protection Agreement (UBS bore the first CHF 5bn) and the CHF 100bn Public Liquidity Backstop were both voluntarily terminated on 11 August 2023 — 60 days after closing — and neither was drawn against the Confederation. Total fees: CHF 40m to establish the LPA, CHF 214m of PLB fees to 31 July 2023, and a CHF 476m emergency-liquidity risk premium to the Swiss National Bank on the CHF 50bn facility repaid 10 August 2023. Roughly CHF 0.73bn (~$0.8bn) — 2.7% of the negative goodwill — to remove a CHF 109bn state safety net inside two months.

Cumulative integration expenses reached $14.2bn at 30 June 2026, guided to ~$15bn by year-end (both “assuming… constant foreign-exchange rates compared with 30 September 2023,” per UBS), against cumulative gross savings of $12.6bn — a cost-to-achieve ratio of roughly 1.1x. Note that the savings ambition rose from ~$13bn to ~$13.5bn while the cost migrated ~$13bn → ~$14bn → ~$15bn.

Verdict on the deal: a competently executed balance-sheet arbitrage, not an operating triumph. Buying tangible assets at roughly 0.12x book and cancelling the state backstop within 60 days for CHF 0.73bn is genuinely excellent capital allocation, and it is the single best decision in UBS’s modern history. But the return evidence is sober. Pre-deal standalone UBS earned RoCET1 of 17.0% (FY22) and 17.5% (FY21); underlying RoCET1 in 1H26 is 16.7% — still marginally below. And the metric management could not adjust says the same: relative total shareholder return over the 2023–2025 long-term incentive period ran 17.45 percentage points below the listed G-SIB index. Three years and $15bn bought enormous scale and a permanently lower cost base; they have not yet bought a higher return on capital than UBS already had.

Nor did the deal buy compounding. Tangible book per share stepped from $16.28 (FY22) to $24.34 (FY23) — almost exactly the negative goodwill per share — and has since crawled to $26.89, roughly 3.2% a year, falling sequentially in 2Q26. Applying Marathon’s capital-cycle lens, the asset-growth anomaly cuts for UBS: headcount down 19% since FY2023, Non-core leverage exposure down 93%, share count down 3.4% — contraction of this kind historically precedes excess returns, and the Swiss consolidation from four universal banks to one is the most favourable supply-side shift in the franchise’s history. The complication is that the state responded to that supply-side victory by taxing it. The regulator is now the marginal competitor.

7.2 Buybacks — and the price-to-tangible-book actually paid

Dividends declared per share: 2022 $0.55 · 2023 $0.70 · 2024 $0.90 · 2025 $1.10 (+22%, $3,404m paid 23 April 2026), with 2026 accruing at “a mid-teen percentage increase” (~$1.25–1.27, ~$3.9bn). FY2025 total capital returns were $6.4bn. Cumulative 2021 → July 2026: ~$16.3bn of buybacks and ~$12.2bn of dividends, ~$28.5bn — almost exactly the negative-goodwill gain. UBS has, in effect, distributed the Credit Suisse windfall.

The arithmetic on the current programme. Tangible equity at 30 June 2026 was $82,330m on 3,061.7m shares. Executing $3.0bn at $52.74 retires 56.9m shares (1.86%) and leaves TBVPS at (82,330 − 3,000) / (3,061.7 − 56.9) = $26.40, a 1.8% reduction. Roughly half the cash buys nothing but premium over book.

Is it still value-creating? Using justified P/TBV = (RoTE − g)/(CoE − g): 1H26 underlying RoCET1 of 16.7% converts to roughly 14.7% RoTE (CET1 $72.5bn over tangible equity $82.3bn), which at CoE 10% and g 3% justifies ~1.67x. Full delivery of the 2028 ambition (~18% reported RoCET1 ≈ 15.9% RoTE) justifies ~1.84x. The stock is at 1.96x — above both. The buyback’s earnings yield is ~6.9% on annualized 1H26 reported EPS and ~8% underlying, against a ~10% cost of equity.

Verdict on the buyback. The 2021–2024 repurchases at 1.06–1.27x tangible book were textbook, and the suspension in March 2023 to fund the acquisition was itself an act of discipline. 2025 at ~1.44x was defensible. The July 2026 programme at ~1.96x is not obviously value-creating and is marginally value-destructive on current returns — accretive only if one underwrites the 2028 ambition in full and a softened Swiss rule. Peer context from published filings: Morgan Stanley trades ~4.1x on a ~25% peak RoTCE, Goldman ~3.0x on ~22%, HSBC ~2.05x on ~15%, where the same repurchase arithmetic turns tangible-book-dilutive — and Barclays ~1.27x on ~12%. UBS sits on the HSBC point. Applying Greenwald, a buyback creates value only where price is below earnings power value; at 1.96x tangible book on ~14.7% RoTE, price is at or slightly above a defensible EPV.

7.3 The collision between capital return and Swiss regulation

CET1 headroom is thin: $72.5bn of capital on $503.9bn of risk-weighted assets is 14.4% against a ~14% guide ($70.5bn) — roughly $2.0bn of excess, and that is already after deducting the full $3bn reserve for the new programme, which cost 60 basis points of the ratio. Against UBS’s own ~$37bn requirement at “an estimated annual capital cost of $3bn,” the scenarios are stark. Using filed inputs — ~$11.7bn of annualized net profit, ~$1.8bn consumed by RWA growth, a ~$3.9bn dividend, leaving ~$6.0bn distributable before regulatory build:

INTERPRETATION: UBS can fund both the rule and the buyback — and only both. The guided $3bn is the arithmetic residual after funding the proposal on its stated glide path, not a floor. The market is capitalizing a 1.9% buyback yield as a base case when it is closer to a ceiling. Solvency is never in question; the distribution is. Note also that the new programme is a deceleration: the February 2026 $3bn was front-loaded roughly $2bn into the first half, whereas the new $3bn stretches over up to four quarters with only $1bn actually committed — and management attributes the pacing to the regulatory outcome, not to earnings.

In fairness, mitigation is real and already demonstrated. UBS upstreamed $9bn from subsidiaries in 4Q25 (~$4bn from Credit Suisse International in the UK, ~$3bn from the US intermediate holding company), cutting parent-bank risk-weighted assets by $26bn — “effectively $6.5 billion times four” — and has deliberately withheld a further ~$4.5bn of parent-bank dividend “to see what happens in terms of the Swiss regulatory capital framework.” There is also a genuine upside option: the CFO has argued that Switzerland’s application of the operational-risk internal loss multiplier drives materially higher operational-risk RWA than the UK, EU and US implementations, where authorities are expected to set it at 1 — which “would mean $40 billion lower risk-weighted assets.” Harmonization there would be worth more than the compromise being debated. No redomiciliation threat appears in any UBS transcript reviewed, and none is asserted here.

7.4 Compensation and incentives

¹ Nine months. Pay was flat year on year in a year when net profit rose 53% — a point in management’s favour.

The actual long-term incentive metrics, verbatim: “average reported return on CET1 capital (RoCET1) and relative total shareholder return (rTSR) over a three-year performance period” — RoCET1 50% / rTSR 50%. For the 2025 award the RoCET1 range was raised to 8–16% from 7.5–14%: 16% pays out in full, 8% pays 33%, below 8% pays nothing. Relative TSR is measured against a listed G-SIB index, with ±25 percentage points as the doubling and forfeiture triggers. The CEO’s annual scorecard weights Group pre-tax profit 20%, cost/income 20%, RoCET1 20%, non-financial 30%, behaviours 10%.

The 2022 award covering 2023–2025 vested at 71.56%: RoCET1 scored 100% after the committee stripped “both the positive and negative impacts related to the integration (such as the negative goodwill, or gain, of USD 27.7bn for 2023),” while relative TSR scored 43.12%, running 17.45 points below the index.

INTERPRETATION. By this framework’s standards the metric set is good: there is no revenue, no asset-growth, no invested-assets and no size metric anywhere — managers are paid on return on capital, cost efficiency, absolute profit and relative TSR. Three criticisms stand. First, there is no tangible-book-value-per-share metric, while RoCET1 carries roughly 70% of combined short- and long-term weight and is mechanically improved by shrinking the capital base — an incentive that tilts toward buying stock at almost any price. Second, an 8% RoCET1 floor paying 33% is undemanding for a bank that earned 17.5% standalone. Third, “reported” RoCET1 was committee-adjusted in 2023, 2024 and 2025; the adjustments removed the windfall as well as the charges and are defensible, but reported is not reported. The honest half of the scorecard came from the metric that could not be adjusted. Politically, the proxy adviser Ethos recommended rejecting both the compensation report and the buyback; the 2026 AGM nonetheless approved compensation with roughly 93.5% support. For global context, Ermotti’s ~$18m compares with Morgan Stanley’s Ted Pick at ~$45m — underpaid globally, and a lightning rod domestically.

7.5 Insider and ownership behaviour — there is no signal, and none is manufactured

As a foreign private issuer, UBS’s own directors and officers are exempt from Section 16 and file no Forms 3, 4 or 5. Every one of the 35 Form 4s and 6 Form 3s under UBS’s CIK is UBS Group AG reporting as a greater-than-10% beneficial owner of other issuers — the three most recent (29 May 2026) are disposals of auction preferred stock in Eaton Vance funds — as is the Schedule 13G/A cluster filed 31 July 2026. US insider-transaction signal for UBS itself is zero.

From the FY2025 Annual Report’s audited ownership tables: Ermotti held 2,930,928 shares (0.222%) at 31 December 2025, up from 2,755,505; the Board held 1,941,862 in aggregate (0.147%) versus 1,747,176; Kelleher held 642,893 versus 552,218; two directors elected in 2025 held none. Every increase is explained by the mandatory settlement of at least 50% of board fees in blocked shares and by vesting — Ermotti’s realized-compensation table shows CHF 0 from equity plans in both 2024 and 2025. There is no evidence of a single discretionary open-market purchase by any UBS director or executive in 2024–2025, and equally no selling signal. Zero conviction information in either direction. (OPEN QUESTION: individual SIX Exchange Regulation Article 56 management-transaction filings could not be retrieved from the register; absence is inferred from the Annual Report, not proven.)

7.6 What Non-core and Legacy has actually released

Non-core and Legacy has delivered on the profit-and-loss and leverage promises: risk-weighted assets of $27.7bn against $83.8bn at inception (−67%), leverage exposure of $13.8bn against $208.7bn (−93%), underlying operating expenses of $96m a quarter (−88% versus the 2022 baseline), and a 2Q26 underlying pre-tax loss of only $52m. But of the $27.7bn of residual risk-weighted assets, $24.0bn is operational risk — credit, market and other risk together are only ~$3.7bn, and average attributed equity is $2.7bn. INTERPRETATION: the capital release from Non-core is largely already banked. Anyone modelling a further $20bn+ of RWA release from this division is modelling the wrong number; the remaining prize is $2.7bn of attributed equity, and the residual is a largely permanent operational-risk charge, not a coiled spring.

7.7 Verdict — has management allocated capital intelligently?

Yes on the deal. Yes historically on the buyback. No on the buyback today.

The Credit Suisse acquisition was correct and competently executed: $19.6bn all-in for $27.5bn of day-one tangible book, a permanent cost-saving annuity, and a franchise that took Group invested assets from $3,957bn to $7,326bn (+85%), with the state backstops cancelled in 60 days for CHF 0.73bn. But it was a balance-sheet arbitrage rather than an operating triumph — underlying RoCET1 is still below the pre-deal level and relative TSR ran 17.45 points below the G-SIB index over the incentive period. Buying assets at 0.12x book is not a repeatable skill.

The discipline frays on the current buyback. $3bn at ~1.96x tangible book pays a ~$1.47bn premium to book, cuts tangible book per share by ~1.8%, and yields ~7–8% against a ~10% cost of equity — management is buying its own stock at the richest price-to-book of the post-crisis era while simultaneously telling its regulator that it needs $37bn more capital. The incentive structure quietly encourages exactly this: RoCET1 carries ~70% of combined weight and is flattered by shrinking the denominator, and nothing in the scorecard penalizes paying twice book to do it. The unanswered question — which belongs in any management meeting — is why a $3bn buyback at 1.96x tangible book is preferable to a special dividend of the same size. Management has neither been asked nor volunteered an answer.

Net: strong capital allocators who made one very large correct decision and a series of correct small ones, now facing the harder problem — what to do with excess capital when your own stock is expensive and your regulator wants the money. Their answer so far is to buy the stock anyway. That is the weakest link in the file.

8. Changes and Headwinds — Last Two Years

8.1 The Swiss capital file — dated, and live this month

This is the dominant development of the period and the decisive open variable in the thesis.

The critical distinction, and the most under-appreciated point in the file: there are two instruments on two different legal routes.

• The Capital Adequacy Ordinance tranche is ALREADY LAW. It carries no parliamentary veto. Software is amortised over a maximum three years for capital purposes and prudential valuation adjustments are tightened; effective 1 January 2027 (valuation) and 1 January 2029 (software). Cost: ~$4bn of derecognition at Group — −0.8 points on the Group CET1 ratio — and ~$2bn at UBS AG standalone. This is not contingent and should not be scenario-weighted. Note that the deferred-tax-asset deduction in the original draft was dropped — a genuine, already-banked win for UBS.
• The Banking Act tranche — the foreign-participations deduction — is NOT decided. ~$22bn on UBS’s figure, seven-year phase-in, entry into force 2028 at the earliest, full effect around 2035.

Where it stands as of this report date: Parliament has decided nothing. The Council of States takes the bill first; WAK-S postponed in May and resumes 10–11 and 31 August 2026 — days from now. A floor vote in the Council of States is expected in the autumn session at the earliest, the National Council follows, and final resolution runs into 2027. A Banking Act revision is an ordinary federal statute and is therefore subject to the optional referendum (50,000 signatures within 100 days); none has been launched, but the route exists and would push entry into force well beyond 2028.

The quantum is disputed by the authorities, and this cuts against UBS’s public framing:

The most damaging datapoint to UBS’s advocacy comes from the Swiss National Bank’s own Financial Stability Report 2026, in which Vice Chairman Antoine Martin calls full CET1 backing “important… targeted… and proportionate,” and the report states that UBS “already has sufficient capital to meet the requirements for full capital backing” — Group eligible CET1 exceeding fully-applied 2030 requirements by $9bn and the parent bank by $13bn. INTERPRETATION: the reconciliation between “~$9bn” and “~$22bn” is accounting deduction versus capital that must actually be raised or retained given the buffer already carried. Neither party has published its methodology, and both cannot be describing the same quantity. The practical consequence is that the solvency impact is close to nil while the return and distribution impact is real.

What it costs, on the range of estimates:

Against roughly $12bn of underlying earnings, FY2026 capital returns of ~$6.4bn leave ~$5.6bn retained, less ~$2.1bn consumed by 3% risk-weighted-asset growth — ~$3.5bn a year of net accretion against a $3.1–3.4bn a year requirement. The rule absorbs essentially the entire organic capital-generation surplus, and roughly the whole current buyback run-rate, for the best part of a decade. UBS’s own figure agrees: ~$3bn a year of annual capital cost. This is not a solvency or dilution event. It is a capital-return event plus a permanent 2–4 point return haircut.

The countervailing move is in the opposite direction abroad: the US enhanced supplementary leverage ratio recalibration took effect 1 April 2026, and on 19 March 2026 the Fed, OCC and FDIC formally rescinded the 2023 Basel III Endgame framework, cutting required Tier 1 capital by 5.6–7.9%. September-2025 reports that UBS explored a US redomicile drew a Senate Banking letter from Senator Warren to Treasury. ASSUMPTION: redomicile is a very low-probability tail — tax, licensing, client-consent and Swiss-political costs are enormous — and no UBS transcript reviewed contains a redomiciliation threat. But the option’s existence has been a demonstrably effective lobbying instrument: the committee has drifted from 100% toward 50–80%.

8.2 Integration milestones

8.3 Litigation — materially de-risked, with one unquantified tail

Litigation and regulatory provisions fell from $3,602m (31.12.24) to $2,200m (31.12.25) to $1,964m (30.6.26); IFRS 3 acquisition-related contingent liabilities from $2,122m to $531m to $316m; total provisions and contingent liabilities from $8,409m to $5,035m to $4,631m. Estimated losses in excess of provisions stand at $0–1.5bn, unchanged from FY2025 to 2Q26 — the lowest post-acquisition range disclosed.

INTERPRETATION. The book was materially de-risked in 2025: the two largest cash items (France, DOJ tax) are settled and paid, provisions fell 39% year on year, and the IFRS 3 contingent liability fell ~$1.6bn “mainly as a result of releases upon resolution.” What remains is one unquantified, unprovisioned item — the AT1 case, where CHF 16bn of notional is the theoretical ceiling and the remedy was expressly left open. Note the asymmetry of scale: the $0–1.5bn litigation range is now small relative to the $4–24bn regulatory-capital range. Any model still carrying DOJ RMBS as a forward liability is stale — it is now a credit, not a liability.

8.4 Management’s own risk ranking has shifted

INTERPRETATION (high confidence): management’s own ordering says the binding risk has migrated from integration execution to Swiss capital reform. That is the single clearest internal confirmation of where this thesis now sits.

8.5 Other developments

Leadership was reshuffled in 2024 (Ivanovic to Asset Management, Khan to APAC, Karofsky to the Americas and co-head of GWM), formalising the succession contest. Legacy Credit Suisse debt tender offers in late 2025 booked an $885m loss (net $457m after a $427m purchase-accounting release). The Swisscard card portfolios were transacted with American Express, producing gains in FY2025 and 1Q26. Rescission offers were made in early 2026 to cure a Securities Act registration defect inherited from Credit Suisse Group. Revised Swiss anti-money-laundering legislation and a beneficial-ownership transparency register enter into force 1 October 2026, adding due-diligence cost. In the EU, a June 2026 delegated act temporarily adjusts the Fundamental Review of the Trading Book from 1 January 2027 for three years; the UK PRA’s Basel 3.1 implementation date remains 1 January 2028; and an EC banking-competitiveness report published in July 2026 promises legislative proposals in 1Q27.

8.6 Verdict — do these strengthen or weaken the thesis?

Operationally they strengthen it decisively; structurally they weaken it. The integration was delivered on time and close to budget, the state backstops were cancelled within sixty days for CHF 0.73bn, the litigation book has been reduced by 39% with the two largest cash items settled, and the deferred-tax-asset deduction was dropped from the final ordinance. Set against that: the Capital Adequacy Ordinance is now law and costs 0.8 points of CET1 ratio from January 2027 regardless of Parliament; the Banking Act tranche is undecided and worth 1.9 to 4.2 points of return on CET1 capital depending on where it lands; management’s own risk hierarchy now puts regulation first; and the buyback has been made explicitly conditional on a committee that begins deliberating nine days after this report date.

9. Risk Analysis

9.1 Risk matrix

9.2 What is not a material risk

Solvency and funding are not in question. CET1 stands at 14.4%, the going-concern ratio at 19.0%, total loss-absorbing capacity at 38.4%, the liquidity coverage ratio at 177.3% and the net stable funding ratio at 115.1%; reported profit converts to regulatory capital at more than 100%; and the SNB’s own Financial Stability Report 2026 states that UBS already holds sufficient capital to meet the fully-applied 2030 requirements, with Group eligible CET1 exceeding them by $9bn.

Dilution risk is effectively nil. UBS is retiring shares, not issuing them — share count is down 3.4% year on year with 63.8m cancelled in 2Q26 — and nothing in the capital proposals requires an equity raise. The requirement is met by retention, not issuance.

Legacy litigation is largely resolved. Provisions are down 39% year on year, the two largest cash items (France €835m, US cross-border tax $511m) are settled and paid, the DOJ RMBS matter is now a credit rather than a liability, Greensill has been dropped from the disclosed item list, and the reasonably-possible-loss range sits at its post-deal low of $0–1.5bn. The AT1 case is the sole exception and is carried separately at row 2.

9.3 The risk that actually dominates

Rows 1 and 3 interact, and that interaction — not either risk alone — is the real exposure. A capital rule adopted at 100% is survivable in a strong capital-markets year and considerably less so in a weak one. UBS’s ~$3.5bn of annual net capital accretion assumes roughly $12bn of underlying earnings, which currently embeds an Investment Bank earning 1.6x its own stated through-cycle ambition. Normalize the IB and the capital-accretion surplus narrows toward the requirement, at which point the buyback is not merely capped but suspended (the competitive-position section, scenario c). These two risks are usually modelled independently; they are correlated through the same denominator, and a bad regulatory outcome is most likely to arrive in the same year as a weak market.

10. Valuation Discussion — Embedded Expectations

No price target and no recommendation appear in this section. Enterprise value is not computed: EV is a category error for a deposit-funded bank. The analysis uses price-to-tangible-book and price-to-earnings only.

10.1 The earnings base — and why the multiple set looks self-contradictory

At $52.74 on 3,061.7m shares outstanding, UBS carries a market capitalization of $161.5bn, against filed book value per share of $29.12 (P/B 1.81x) and tangible book value per share of $26.89 (P/TBV 1.96x). Three defensible earnings bases give three different pictures:

The TTM bridge reconciles two ways: $2,481m (3Q25) + $1,199m (4Q25) + $3,040m (1Q26) + $2,800m (2Q26) = $9,520m, or summing quarterly diluted EPS of 0.76 + 0.37 + 0.94 + 0.87 = $2.94.

The identity that dissolves the apparent contradiction. For a bank, P/E = (P/TBV) ÷ RoTE. At 1.96x: 1.96/0.144 = 13.6x; 1.96/0.116 = 16.9x — bracketing the observed multiples. INTERPRETATION: the entire “cheap on earnings but at a record on book” debate is a fight over the RoTE denominator and nothing else. Bulls quote the P/E; bears quote the P/TBV; both are reading the same equation from opposite ends.

10.2 Cost of equity — argued, not assumed

CAPM gives a USD risk-free rate of ~4.0% plus a beta of 1.036 times a ~5.0% equity risk premium = 9.2%. Adjust upward for 26.1% trailing realized volatility with ~44% of variance idiosyncratic — and that idiosyncrasy concentrated in a single binary legislative event for which no factor hedge exists; ~20.8x assets to tangible equity; and an unquantified, unprovisioned AT1 exposure before the Swiss Federal Supreme Court. Adjust downward for roughly 57% of underlying divisional profit coming from capital-light wealth and asset management, a 177.3% liquidity coverage ratio, a 14.4% CET1 ratio, and a credit book dominated by Swiss mortgages.

ASSUMPTION: base cost of equity 10.0%; bull 9.0%; bear 11.0%. For calibration, a comparable Spanish/LatAm bank supports ~11%, while the US premium comps below embed 7.3–7.5%. UBS belongs between. ASSUMPTION on growth: g = 3%. Note that sustainable g = RoTE × (1 − payout) would give ~6.6%; UBS’s actual tangible book per share has compounded at ~4% a year since FY2023 and fell in 2Q26. Three percent is the observed number, not the arithmetic one.

10.3 Embedded expectations — what the price requires

Implied sustainable RoTE at P/TBV 1.96 = P/TBV × (CoE − g) + g:

At 1.96x tangible book the market underwrites a permanent RoTE of roughly 16.7%. UBS earned 14.4% underlying, 13.9% reported, and 11.6% on a trailing-twelve-month basis. There is no cell in that grid — not even the most generous corner at 9% cost of equity and 4% growth — where the market is paying only for the return UBS actually delivered.

Measured against the three relevant benchmarks:

• Against 1H26 underlying (14.4%): 230 basis points below the central embedded figure. At CoE 10% and g 3%, a 14.4% RoTE justifies 1.63x — 17% below the traded multiple.
• Against the pre-deal baseline (FY20–22 RoTE 12.8% / 14.1% / 14.9%; RoCET1 17.4% / 17.5% / 17.0%): the market paid 1.13–1.14x for precisely this order of return. It now pays 1.96x — a ~73% re-rating on a return that has not yet been regained. This is the load-bearing observation of the entire valuation.
• Against what survives the capital bill: 11.6–12.6% (the valuation section).

Inverting, justified P/TBV = (RoTE − g)/(CoE − g), at g = 3%:

10.4 What the capital bill actually does — and the important subtlety

Risk-weighted assets of $503,923m against CET1 of $72,464m give 14.38%. Holding 16.5% requires ~$10.7bn more; holding 18.5% requires ~$20.8bn — bracketing UBS’s own 20-F range. Tangible equity rises to $93.0bn or $103.1bn. Crediting the incremental capital a 3% after-tax return (ASSUMPTION), profit rises to $12.2bn or $12.5bn, so sustainable RoTE falls to 13.1% or 12.1%; normalizing the tax rate from 21.1% to 24% takes it to 12.6% or 11.6%. Those justify 1.37x and 1.23x respectively at a 10% cost of equity.

But the bull has a legitimate and important rebuttal. $20.8bn of CET1 earning ~3% after tax against a 10% cost of equity destroys roughly 0.30x book, or ~$14.6bn — if it happened today. It does not: the phase-in begins in 2028 at 65% and rises five points a year to 100%, an average delay of ~7.5 years. Discounted at 10%, the present value is roughly $7.1bn, or about $2.33 a share — and $1.63–1.86 under a 70–80% compromise.

INTERPRETATION — the crux of this article. A de-rating from 1.96x to 1.43x is roughly $14 a share, against an intrinsic economic loss of ~$2.33 a share. The reconciliation is that the capital bill is overwhelmingly a multiple risk, not an intrinsic-value risk. It barely touches earnings per share — it slightly raises them — while mechanically compressing headline return on tangible equity. Banks are marked on headline RoTE. An investor who believes the market prices discounted cash flows should be relatively relaxed; an investor who believes banks trade off headline returns should not be. This article takes no position on which is right, but flags that the two framings are an order of magnitude apart and that the difference between them is the debate.

10.5 Scenario analysis

Probabilities are explicit ASSUMPTIONS. The per-share zones are scenario outputs, not price targets, and no recommendation attaches to them.

INTERPRETATION — the sharpest single output. Using scenario midpoints of roughly $35 / $45 / $60 and holding the bear at a 15% weight, the traded $52.74 solves to approximately 62% bull, 23% base, 15% bear. The price therefore discounts the favourable regulatory outcome at a probability well above what an upper-house committee that meets on 10, 11 and 31 August 2026 — still weighing 50%, 70% and 80% — can reasonably support. Regulatory relief is not a catalyst; it is already the base case in the price, having been paid for twice already (+20.0% in December 2025, +6.5% on 31 March 2026).

Sensitivity, ordered by impact on justified P/TBV: (1) the backing percentage, 100% versus 70%, worth ~1.5 points of RoTE ≈ 0.20x; (2) cost of equity, 100bp ≈ 0.20–0.25x; (3) sustainable RoTE, 100bp ≈ 0.14x; (4) normalized tax, 21.1% versus 24% ≈ 0.06x; (5) Investment Bank profit −⅓ ≈ 1.0 point of group RoTE ≈ 0.14x.

10.6 Peer cross-section — price-to-tangible-book against return on tangible equity

All peer figures are drawn from each company’s own public filings at the reference dates shown.

Julius Baer is the natural Swiss pure-play comparator, but its Merrill IWM goodwill makes P/TBV structurally uncomparable and its figures were not independently verified — flagged as an open question rather than quoted.

Regressing P/TBV on RoTE across the seven genuine bank comparables gives a slope of 0.2166x per point of RoTE and an intercept of −0.921. Predicted UBS P/TBV: 2.20x at 14.4% RoTE, 2.09x at 13.9%, 1.90x at 13.0%, and 1.59x at the 11.6% TTM figure. Implied cost of equity at g = 3% across the set: Goldman 7.3%, Northern Trust 7.3%, Morgan Stanley 7.5%, HSBC 8.0%, Raymond James 8.8%, UBS 8.8%, Barclays 9.5%, Santander 9.7%.

INTERPRETATION — the key cross-sectional test, and it cuts against the own-history read. On the fitted line, UBS at 14.4% underlying RoTE is roughly 11% cheap (1.96x traded versus 2.20x predicted), priced at an 8.8% implied cost of equity — dearer than the European universals, cheaper than every US comparator. Cross-sectionally UBS is fair to slightly cheap; it is extreme only against its own history. Both readings are correct, and they reconcile on the single variable no peer carries: an unlegislated capital regime capable of removing 250–450 basis points from the denominator. On the TTM 11.6% figure the line says 1.59x and UBS is ~23% dear — so the cross-sectional verdict rests entirely on accepting “underlying,” which excludes ~$2.6bn a year of integration expense that has been real cash every year since 2023.

The Morgan Stanley anchor is instructive but is not an argument for UBS: 4.1x tangible book on a 21.6% ROTCE reflects 7.2 points more return, in a settled regulatory regime, with a wealth business earning a 30% pre-tax margin against GWM Americas’ 15.9%. It sizes the prize and shows how far UBS is from claiming it.

10.7 Sum-of-the-parts

Earnings-based; 2Q26 underlying divisional pre-tax profit annualized, taxed at 22% (P&C at the Swiss 20%).

INTERPRETATION. The sum-of-the-parts spans roughly $42–56 a share, midpoint ~$49. The traded $52.74 requires close to the top of the range on every division simultaneously — a peak Investment Bank multiple and a full wealth multiple on a business half of which is low-margin US and no discount for the capital regime. Deducting the ~$2.33 present value of adoption at 100% takes the midpoint to ~$46.7. The sum-of-the-parts undercuts the price at central multiples and supports it only at bull multiples; there is no hidden conglomerate discount to unlock. Valuing GWM Americas on its actual economics rather than a blended wealth multiple removes roughly $5–8bn against a naive 14–15x applied to all of GWM — 49% of GWM invested assets producing about 28% of GWM pre-tax profit.

10.8 Own-history context

AZI’s own-history percentiles place UBS in the top few percent of its own ~10-year range on price-to-book and price-to-sales, and around the 82nd percentile on price-to-earnings. The ranks are retained only as directional — and are best stated as “the richest on book in the post-crisis era” rather than as a precise percentile — while the underlying levels are rejected (AZI’s BVPS of $22.93 and P/B of 2.30 conflict with the filed $29.12, and the CHF conversion error grows through the sample — see the data-integrity note).

Rebuilt independently from filed tangible book values and period-end market capitalizations: FY2021 1.13x → FY2022 1.14x → FY2023 ~1.29x → FY2024 ~1.35x → FY2025 ~1.87x → 2Q26 1.96x. This is the highest ratio in the filed series and, given that UBS traded at or below tangible book for most of 2015–2020, almost certainly the highest of the decade. The AZI rank is directionally confirmed even though its level is not. Note that the price-to-earnings percentile is only the 82.4th — the same P/E-versus-P/TBV divergence, and further confirmation that the re-rating has been a book-multiple event.

Why the book multiple rose is itself the story: tangible book per share went from $16.28 to $26.89, but essentially all of that came from Credit Suisse purchase accounting ($27,264m of negative goodwill against a ~$27.5bn rise in tangible book). Since FY2023, tangible book per share has compounded at ~4% a year and fell in 2Q26 — 63m shares repurchased at ~$43.06 against a TBVPS of $26.93 cut it by ~1.2%. At $52.74, each dollar repurchased destroys roughly $0.49 of tangible book value per share. The buyback has stopped being a value lever, and it is the one lever UBS has explicitly made hostage to Parliament.

10.9 What the market is pricing correctly, and what it may not be

Correctly. The integration is genuinely finished — $12.6bn of the $13.5bn savings ambition, internal headcount down 19%, Non-core credit and market risk-weighted assets down to $4bn, and both end-2026 targets met five months early at 16.7% and 70.1%. Absolute earnings power is more than 50% larger than pre-deal ($7,326bn of invested assets against ~$4,200bn). Roughly $2.6bn a year of integration expense stops in 2027 (~$0.64 a share). Litigation has been materially de-risked, with provisions down 39% and the reasonably-possible range at its post-deal low of $0–1.5bn. Solvency is not in question.

Possibly not. (1) The re-rating has outrun the return — 16.7% underlying RoCET1 today against 17.0–17.5% pre-deal, at 1.96x against 1.13–1.14x. (2) The ~62% embedded probability of a favourable vote, against a committee that has not voted. (3) The denominator switch — the 280 basis points between 14.4% underlying and 11.6% trailing is real cash integration expense plus a tax rate that ran 11.9% in FY2025 and 3.1% in FY2023. (4) GWM Americas does not earn the wealth multiple applied to all of GWM. (5) The buyback is TBVPS-dilutive and conditional — total capital return of ~1.9% plus ~2.1% is ~4%. (6) The 2027 step-up is double-counted by any model that adds back integration expense without also deducting the ~$1.2–1.5bn after-tax purchase-accounting run-off.

Not deducted anywhere above: the AT1 exposure before the Swiss Federal Supreme Court, which is unquantified and unprovisioned and sits outside every zone in this section.

11. Variant Perception

11.1 The consensus view

Consensus holds that UBS is a successful, largely de-risked turnaround: the Credit Suisse integration has been delivered on time and close to budget, underlying return on CET1 capital has quadrupled from 4.2% to 16.7%, the 2026 exit-rate targets were met five months early, and the company now owns the largest pure-play global wealth franchise with $7.33trn of invested assets. The Swiss capital file is understood as a known overhang trending toward compromise — the Council of States committee has tabled only weakening variants — and the residual is treated as a timing question rather than an economic one. On roughly 14x underlying earnings, with a ~4% total capital-return yield and integration costs about to stop, the shares look reasonably priced for a quality compounder emerging from a self-inflicted three-year detour.

Most of that is correct. The tape agrees: +44% over twelve months, within ~4% of a record, a 1.62 one-year Sharpe ratio, and an aligned 21/50/200-day trend.

11.2 The strongest bull case

1. The turnaround is genuine and the evidence is not soft. Underlying cost/income has gone 87.2% → 70.1%; internal headcount is down 19% since FY2023; more than 90% of cost synergies and 90% of legacy application retirement are done; all clients are migrated; Non-core has shed 67% of risk-weighted assets and 93% of leverage exposure. $12.6bn of $13.5bn in gross savings is banked.
2. Reported earnings understate the improvement. FY2025 was flattered by an 11.9% tax rate and $949m of litigation releases; 1H26 has neither. Normalized to a 23% tax rate, 1H26 net profit growth is +72%, not the reported +43%.
3. The capital bill is smaller than UBS says. The Federal Council’s effective shortfall is ~$9bn, not $22bn, and the SNB states UBS “already has sufficient capital to meet the requirements for full capital backing,” with Group eligible CET1 exceeding fully-applied 2030 requirements by $9bn. The two sides differ on annual cost by 3–5x. Properly discounted — seven-year phase-in from 2028 — the present value of adoption at 100% is only about $2.33 a share.
4. Cross-sectionally the shares are not expensive. On a regression of price-to-tangible-book against return on tangible equity across seven bank comparables, UBS at 14.4% underlying RoTE is ~11% cheap (1.96x traded against 2.20x predicted), at an implied 8.8% cost of equity — dearer than the European universals, cheaper than every US comparator.
5. Rates are a two-sided option in UBS’s favour. On UBS’s own disclosure, +100bp adds ~$1.4bn of annual banking-book net interest income and −100bp also adds ~$0.9bn, thanks to contractual flooring benefits. Zero is the worst point and the SNB is sitting on it.
6. An unpriced regulatory option exists. If Switzerland harmonizes its operational-risk internal loss multiplier with the UK, EU and US, the CFO says it is worth “$40 billion lower risk-weighted assets” — worth more than the parliamentary compromise being debated, and in nobody’s numbers.
7. The 4Q26 strategic update is a live upward catalyst. Management declined to re-base the 2028 ambition on the 2Q26 call and deferred it to early 2027, having already outperformed the 2026 targets.

11.3 The strongest bear case

1. The destination is not new. Standalone UBS earned RoCET1 of 17.4% / 17.5% / 17.0% in FY2020–22 and the market paid 1.13–1.14x tangible book. Today it earns 16.7% underlying — still marginally below — and the market pays 1.96x. Ermotti conceded the point himself: UBS is “close to achieving the same level of profitability UBS had prior to the acquisition.” A ~73% re-rating for a return that has not been regained.
2. The quality of the inflection is cyclical. 44% of the 1H26 underlying pre-tax improvement came from an Investment Bank earning 1.6x its own through-cycle ambition in a record equities quarter the CFO said would not repeat. Meanwhile underlying operating expenses rose 5.6% — the entire cost/income improvement is revenue-led, not cost-led. “Gross” savings of $12.6bn have not reduced the absolute cost base; management guides to net saves of only ~75% of gross.
3. The accounting tailwinds expire together. ~13% of Group net interest income is purchase-accounting accretion falling 30% a year to zero; the tax rate normalizes from 11.9% to ~23%; credit runs at 6bps against a ~20bps through-cycle charge worth ~$0.9bn. All three reverse into 2027–28.
4. The 2028 ambition is void as written. UBS’s ~18% RoCET1 ambition is expressly “based on the current capital framework and assuming a CET1 capital ratio of around 14%” — the framework already before Parliament. At a ~17.6–18.4% effective minimum, the same earnings yield ~12.7–13.3%.
5. Per-share compounding has been poor. Essentially all of the tangible-book step-up came from purchase accounting; since FY2023 TBVPS has compounded ~3–4% a year and fell in 2Q26. Total tangible-book return including dividends has run at ~8.1% a year since YE2023 — below cost of equity — and relative TSR over the 2023–2025 incentive period ran 17.45 points below the G-SIB index.
6. The buyback is a residual, not a floor, and it is dilutive to book. $3bn is a 1.9% yield, explicitly conditioned on Parliament, decelerating from a front-loaded 2026 programme with only $1bn committed; each dollar spent at $52.74 destroys ~$0.49 of TBVPS.
7. Two “moat” businesses aren’t. Swiss domestic is a 23%-share, share-losing position against 21 state-guaranteed cantonal banks (HHI ~1,010), and ~82% of deposits leaving Credit Suisse went to them rather than to UBS. US wealth lost $57.7bn of net new money in FY2025 and at least 27 teams with ~$28bn in 1H26.

11.4 The 3–5 assumptions that actually matter

11.5 Where consensus may be offsides — the positioning read

The factor evidence says this is not a crowded trade in the usual sense. UBS’s Momentum beta is essentially zero (−0.02) and its Value beta is zeroed entirely, despite a +44% twelve-month return, with roughly 44% of return variance idiosyncratic and the largest non-market loadings being Country: Switzerland (+0.34) and Sector: Financials (+0.31). Its nearest factor neighbours are broad developed-international ETFs at ~0.86–0.89 similarity, with no single-stock factor twin. In factor space UBS trades as a proxy for developed-international equity beta, not as an idiosyncratic bank — which is itself a caution on the comp set, since the tape does not corroborate any European-bank peer group.

Two implications follow. First, the current momentum unwind cannot hurt UBS directly — over the trailing month Momentum (−3.7%, z −1.41), Quality (−5.0%, z −1.84) and Growth (−3.0%, z −1.09) are being sold hard, and UBS has no exposure to any of them, while the factors it does load on (DividendYield, Financials, Broker-Dealers, Value, LowVol) are working. Second, and more important, the risk here is not a factor unwind but idiosyncratic headline risk, for which there is no factor hedge — and the headline in question is a committee vote beginning nine days from this report date.

Note also that UBS delivered its +44% year against a sector headwind: Financials as a factor returned −10.7% over 252 days (z −1.47) and Broker-Dealers −6.9% (z −2.12, an extreme reading). That gap is the +0.21 alpha and the 44% idiosyncratic variance share. The market re-rated UBS specifically, not European banks generically — which is consistent with the re-rating being about the Credit Suisse story and the regulatory outcome, and inconsistent with it being a sector beta trade.

The variant perception, stated plainly. Consensus is right that the turnaround worked and right that the capital bill is smaller than UBS’s advocacy implies. Where it may be offsides is in three places at once: it is marking UBS on an “underlying” return that excludes $2.6bn a year of real cash cost; it is capitalizing a peak Investment Bank quarter into a permanent multiple; and — scenario-weighting the regulatory outcomes — the traded price solves to roughly 62% bull / 23% base / 15% bear, a probability of softening well above what an unvoted committee supports. Regulatory relief is not a catalyst here; it is already the base case in the price, having been paid for twice (+20.0% in December 2025, +6.5% on 31 March 2026). The asymmetry is therefore poor in a specific and unusual way: the good outcome is discounted, the bad outcome is not, and the one genuinely unpriced item — the AT1 case at the Federal Supreme Court, unprovisioned against CHF 16bn of notional — sits outside every scenario anyone is running.

12. Fact vs. Interpretation

13. Open Questions

1. Where does the Council of States land on the backing percentage, and when? The committee sits on 10, 11 and 31 August 2026 with a decision targeted for September; a floor vote follows in the autumn session, then the National Council, with final resolution running into 2027. Everything material in the valuation hinges on this. An optional referendum (50,000 signatures in 100 days) remains available and would push entry into force well beyond 2028.
2. Whose capital number is right — UBS’s ~$22bn or the Federal Council’s ~$9bn effective shortfall? Neither has published its methodology, and the annual-cost estimates differ by 3–5x ($320–560m against $1.7bn). This single dispute is worth roughly 0.20x on the justified multiple.
3. What is the forward run-off schedule for purchase-accounting accretion? UBS discloses the period amount but no schedule. At ~13% of Group net interest income and falling 30% a year, the shape of the decay materially affects 2027–28 estimates.
4. Does GWM Americas inflect on a net-new-money basis in FY2026? The −$57.7bn FY2025 figure is disclosed annually only, so the next genuine read is the FY2026 Annual Report. Quarterly net new fee-generating assets are the interim proxy.
5. Is the Investment Bank’s −21% FICC print in 2Q26 deliberate balance-sheet reallocation into equities, or share loss? Every US peer grew FICC in the same quarter.
6. Will the 2028 ambition be re-based at the 4Q26 strategic update, and in which direction — upward on integration outperformance, or downward for the capital framework? Management explicitly deferred the question.
7. What is the AT1 remedy? The Federal Administrative Court found FINMA’s order unlawful but expressly left the remedy open; UBS carries no provision and offers no estimate against CHF 16bn of notional. Timing of the Federal Supreme Court judgment is unknown.
8. What was the unexplained ~$8bn GWM item the CFO was asked about on the 2Q26 call and undertook to detail (“We’ll come back on the details on that one”)? Context suggests an invested-asset exit.
9. How much of GWM’s invested assets are genuinely UHNW? UBS does not quantify it, and it is the segment where the moat argument actually holds.
10. Would Switzerland harmonize the operational-risk internal loss multiplier with the UK, EU and US? The CFO puts it at “$40 billion lower risk-weighted assets” — larger than the parliamentary compromise being debated, and in nobody’s estimates.
11. Do UBS insiders buy? As a foreign private issuer UBS files no Forms 3/4/5, and individual SIX management-transaction filings could not be retrieved; the absence of discretionary purchases is inferred from Annual Report ownership tables, not proven from the register.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

14.3 The single cleanest test

If one indicator had to carry the thesis: the Council of States’ decision on the backing percentage, and UBS’s buyback guidance in the quarter that follows it. A committee outcome at or below 70% combined with a buyback stepped up materially above $3bn would falsify the bear case in one move — it would demonstrate both that the capital bill is the smaller number and that the guided $3bn was a floor rather than a residual. Adoption at 100% with the buyback held at $3bn confirms it.

15. Source Appendix

The full source appendix — approximately ninety numbered entries organised by tier, each with title, publisher, URL, date, a one-line statement of what it supports, and a PRIMARY / SECONDARY / TERTIARY designation — is carried as Appendix B to this report.

Its structure is: (1) UBS filings — the five Form 20-Fs for FY2021–FY2025, the interim and capitalization 6-Ks, eighteen material-event 6-Ks, and the documented negative result on Forms 3/4/5; (2) Swiss regulatory primary sources — the Federal Council dispatch and Capital Adequacy Ordinance, the too-big-to-fail hub, the September 2025 consultation, the Credit Suisse state-support accounting, two FINMA AT1 releases, and the SNB Financial Stability Report 2026 and banking statistics; (3) transcripts and investor presentations; (4) industry data — BCG Expand, Capgemini, SNB; (5) trade and financial press, explicitly flagged as the only evidence for the parliamentary state of play; (6) peer primary filings; (7) comparable-company public filings used for cross-read; (8) quantitative feeds and their accept/reject status; (9) the data-quality conflicts table reproduced in the header note above; (10) flagged claims requiring correction or re-sourcing; and (11) coverage gaps recorded honestly.

Three sourcing disciplines govern this article and are worth restating. First, every ratio has been rebuilt from UBS’s filed figures; the AZI and ROIC.ai feeds return CHF-converted values for this USD-reporting issuer and were rejected. Second, the entire parliamentary state of play — the August committee dates, the 50/70/80% variants, the partial-AT1 substitution — rests on trade press, not on primary documents, and is labelled as such wherever it appears. Third, claims that could not be verified to a primary source were cut rather than softened: no Credit Suisse asset-retention percentage is quoted, no Julius Baer relationship-manager capture is asserted, and no redomiciliation threat is attributed to management.

Except for the labeled “Claude’s Take” block, this article contains no investment recommendation and no price target. It is general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Report date: 2026-08-01 · CIK: 0001610520 · Price (ref.): $52.74 (2026-07-31) · Reporting: IFRS, in USD Supplemental to the research memo; not counted toward its length standard. Labels: FACT / INTERPRETATION / ASSUMPTION / OPEN QUESTION. Where a question does not map to a bank’s business model, the correct sector analog is given.

General

What thoughtful questions have other investors asked about this company?

The 2Q26 analyst call is the best available record, and four themes dominated it.

1. US wealth stabilization — Jeremy (Barclays) asked directly about “significant further advisor exits in the quarter” and “where you are in the stabilization of the US wealth management franchise.” Amit Goel (Mediobanca) pressed harder, on why “the net new assets were positive, but then the net new fee-generating assets were negative. When I look at Q2 in prior years, net new fee-generating assets have held up better.” FACT. Management’s answer was a deflection — “nothing I would call out… I wouldn’t overread into one quarter versus the other” — while conceding “a lag effect from previously announced FA movement that will continue to show up in flows for a few quarters.”
2. Buyback structure as a regulatory tell — Giulia Miotto (Morgan Stanley) asked why the $3bn programme runs to June 2027 rather than year-end, and whether “some sort of compromise in Parliament” could allow earlier completion. INTERPRETATION: the sell side has correctly identified that the buyback is a regulatory option, not an earnings statement.
3. Cyclical versus structural returns — Anke Reingen (RBC) asked whether 2026’s structural progress implies upside to the 2028 target, “just trying to distinguish between cyclical versus structural progress.” The CFO declined and deferred to the 4Q26 strategic update. This is the single most important unanswered question in the file.
4. Investment Bank revenue without balance-sheet growth — analysts probed how UBS grew Global Markets revenue 31% without expanding the balance sheet. The answer (equities intermediation, prime-brokerage financing, reallocation out of FICC) is credible but confirms the earnings are financing- and activity-led.

A fifth question the market has not pressed, and should: why is a $3bn buyback at ~1.96x tangible book preferable to a special dividend of the same size? Management has neither been asked nor volunteered an answer.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? A cyclical high, and identifiably so. FACT/INTERPRETATION. Three independent markers. (i) The Investment Bank earned a 24.0% return on attributed equity in 1H26 against UBS’s own ~15% through-cycle ambition — 1.6x — and supplied 44% of the entire underlying pre-tax improvement; the CFO conceded APAC volumes were “unlikely to continue at that level.” (ii) Credit loss expense is running at 6bps against 19bps in FY2023; normalizing to ~20bps costs ~$0.9bn of pre-tax profit. (iii) Invested assets are at a record $7,326bn, and BCG found more than 80% of 2025 industry revenue growth came from market performance rather than net new money. Offsetting these: the effective tax rate is normalizing upward (11.9% in FY2025 to a guided ~23%), and ~$2.6bn a year of integration expense stops at end-2026 — so not every line is at a peak.

Driven by the external environment or internal actions? Both, in roughly equal measure — and the split is the crux. INTERPRETATION. Internal: $12.6bn of gross cost savings, internal headcount down 19% since FY2023, >90% of legacy applications retired, Non-core risk-weighted assets down 67%. External: a record M&A and equities half, record asset levels, and a favourable dollar rate path. The diagnostic that settles it is that underlying operating expenses rose 5.6% year on year — so the entire cost/income improvement from 76.4% to 70.1% was revenue-led. Real cost action happened; it did not reduce the absolute cost base.

How stable are revenues? Moderately, and less than the “wealth manager” label implies. FACT: only ~52% of GWM revenue is recurring (2Q26: recurring fees 52.2%, net interest income 26.2%, transaction-based 21.3%), against Morgan Stanley Wealth’s 58.7% asset-management fee share. At group level, 2Q26 revenue was 55% net fee and commission income, 27% fair-value/trading, 18% net interest income. Recurring fee income is itself a levered claim on market levels.

Outlook for products/services? Structurally positive in wealth: HNWI investable wealth grew 8.7% to $98.3trn in 2025 and the UHNW tier fastest of all (+9.4%). Mandate penetration — only 45.6% of GWM invested assets are fee-generating — is the genuine internal lever, at a record and rising, with MyWay discretionary assets above $40bn (+75%). Against that, per-unit pricing is eroding ~3.9% a year on the fee-paying base, and Asset Management’s gross margin has fallen 19bps → 14bps in under three years while becoming 51.9% passive.

How big will this market be — growing, shrinking, domestic or international? FACT. Global financial wealth reached $333trn in 2025 (+10.7%), with BCG projecting ~7% compound growth to 2030; HNWI investable wealth $98.3trn across 25.3m individuals. The market is growing and overwhelmingly international — and the geography is shifting against Switzerland: Hong Kong ($2.95trn) overtook Switzerland ($2.90trn) as the largest cross-border booking hub in 2025. UBS’s own growth is concentrated outside its home market, with APAC, EMEA and Switzerland gathering at 4.7–6.5% annualized while the Americas gathers at 0.2%.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. FACT-based INTERPRETATION. Every supply-side marker of Marathon’s late capital cycle is present: RIA transaction multiples at a record 11.6x EV/EBITDA (+40% since 2020) with private-equity buyers in 75.8% of Q2-26 deals, an all-time high; deal count up 129% since 2019; Citi Wealth rebuilding toward a >20% return target from −2% in 2023; HSBC redeploying $1.5bn into Asian wealth; JPMorgan’s Private Bank advisor count +17% in two years. Most tellingly, recruiting packages have gone from ~200% of trailing-twelve-month production in 2018–19 to 300–500%, with UBS itself reportedly offering 550% on a ~16-year commitment. Capitalised recruiting books have exploded (LPL $3.3bn, +1,300% since 2018).

How profitable is the business (ROIC, ROE)? For a bank the correct measures are return on tangible equity and return on CET1 capital; ROIC and enterprise value are category errors for a deposit-funded institution and are not computed. FACT:

Against a ~10% cost of equity, UBS destroyed value in FY2023 and FY2024, roughly covered it in FY2025, and earned it in 1H26. The three-year average of ~8.3% sits inside Greenwald’s “6–8%: advantages absent” band. Management’s own adjustment is sobering: excluding litigation releases and at a normalized tax rate, FY2025’s 13.7% becomes 11.5%. The market test agrees — tangible book per share plus dividends has compounded at ~8.1% a year since YE2023, below cost of equity.

How profitable is the industry — how many competitors, what barriers to entry? Profitable but fragmenting. BCG’s benchmark shows pre-tax margins of 27bps of client assets above $100bn against 7bps below $20bn — genuine scale economics, but a cost advantage rather than pricing power. Even the largest player (Schwab, $13.1trn) holds ~13% of HNWI investable wealth and the top ten ~47% on non-comparable definitions. This is a fragmented industry with a concentrated top tier, not an oligopoly. UBS ranks fourth on client assets ($4,942bn, ~5%), behind Schwab, Morgan Stanley ($8,084bn) and JPMorgan ($7,663bn), though it is the largest pure-play global wealth manager and ranks #1 on Euromoney’s private-banking measure.

Can the business be easily understood? Only partially, and that is a genuine mark against it. INTERPRETATION. The wealth business is simple. The consolidated accounts are not: reported results carry purchase-accounting accretion, integration expense, intangible amortization and Swisscard items, requiring a reported-versus-underlying bridge every quarter; a $27,264m non-cash negative-goodwill gain (itself restated twice from $28,925m) makes FY2023 uninterpretable; the effective tax rate has ranged from 3.1% to 24.6% in three years; and the binding valuation variable is a Swiss legislative process. A generalist cannot value this company from the income statement alone.

Can it be undermined by foreign low-cost labour? Not materially. The front office is relationship-based, licensed and jurisdiction-bound. The relevant analogous threat is the opposite — labour inflation, not labour arbitrage: advisors are the scarce input and their price is rising 300–550% of production. Back- and middle-office offshoring is already substantially done and is part of the delivered $12.6bn.

Do brands matter? Yes, but less than assumed and insufficiently to constitute a moat. Swiss private-banking heritage carries real weight in cross-border and UHNW. But Credit Suisse had the same brand and it did not save it — a 167-year-old institution lost CHF 110.5bn of assets in a single quarter (4Q22) and 5% of AUM in 1Q23. Brand is necessary, not sufficient.

What is the nature of competition? Competition for advisors, not for clients. Because the client relationship attaches to the individual rather than the institution, share is transferred by moving people and by acquisition rather than won on merit. FACT: 11,172 experienced US advisors changed firms in 2025 (+16%), including 54 teams managing $1bn+; UBS lost at least 27 teams with ~$28bn in 1H26 alone. Secondarily, competition is on price — visible in the ~3.9% annual fee-margin erosion.

Customers’ switching costs? Genuinely high in UHNW and cross-border; genuinely low in US mass-affluent. In multi-jurisdictional booking, family-office structuring and lending against concentrated or illiquid holdings, search and transfer costs are real, and the returns prove it (APAC 44.6%, EMEA 37.9%, Switzerland 38.8% pre-tax margins on 4.3–9.4% organic growth). In the US the assets follow the advisor out of the door. FACT: the proportion of HNWIs working with a single firm collapsed from 39% in 2019 to 19% in 2025, and 88% now use multiple firms for alternatives access.

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet? Yes, and they are the valuable ones. INTERPRETATION. The $7,326bn of invested assets are client property and appear nowhere on UBS’s balance sheet, yet they generate the majority of revenue — the classic wealth-management understatement. Similarly unrecognised: the Swiss deposit and payments franchise, the client-advisor relationships (which, being unowned, are also the vulnerability), and — perversely — the Credit Suisse brand and platform, most of which was written off or never capitalised. Goodwill and intangibles are only ~$6.8bn against $89.2bn of equity, a direct consequence of the negative goodwill: UBS’s balance sheet is unusually clean, with tangible equity at 92% of total equity.

Off-balance-sheet liabilities? The material ones are disclosed. Litigation and regulatory provisions of $1,964m at 30.6.26 (down from $3,602m at end-2024), IFRS 3 acquisition-related contingent liabilities of $316m (down from $2,122m), and estimated losses in excess of provisions of $0–1.5bn — the lowest post-acquisition range disclosed. The one genuine off-balance-sheet exposure is the AT1 matter: the Federal Administrative Court held FINMA’s CHF 16bn write-down order unlawful on 2025-10-01 with the remedy left open, both FINMA and UBS have appealed to the Federal Supreme Court, and UBS carries no provision and offers no estimate, stating only that an adverse outcome “could be material to UBS.” CHF 16bn of notional is roughly 10% of market capitalization. This sits outside every valuation scenario in the article. Separately, the off-balance-sheet loan portfolio carries its own expected-credit-loss allowances, disclosed in the credit notes.

How conservative is the accounting? Mixed, and the reader should hold both facts. Conservative: reported profit converts to CET1 capital at more than 100% (1H26 distributions of $6,741m against $5,840m of earnings, yet CET1 still rose $1,202m); share-based compensation is fully expensed at $652m (~2.3% of revenue); the reported-versus-underlying bridge is disclosed in detail every quarter; and the FY2023 negative goodwill was revised downward twice as purchase accounting finalised. Less conservative: purchase-accounting accretion of $824m flowed through 1H26 revenue with no forward run-off schedule disclosed; FY2025 net profit was flattered by ~$1.0bn of abnormal tax benefit (including a $747m deferred-tax revaluation and $215m arising simply from a higher share price — a reflexive gain) and $949m of net litigation releases; and UBS withdrew two GWM margin KPIs — gross margin on invested assets after FY2020 and the fee-generating-asset margin after FY2022 — in both cases while the metric was falling, while retaining the equivalent disclosure in Asset Management. That last item is the clearest disclosure-quality flag in the file.

How CapEx-hungry is the business? Not capital-expenditure hungry; regulatory-capital hungry, which is the correct sector analog. Physical and technology capex is modest relative to a $1.7trn balance sheet, though technology has risen to 30–35% of non-front-office cost industry-wide and has outgrown revenue since 2020. The binding constraint is CET1: UBS carries $72.5bn against $503.9bn of risk-weighted assets, roughly $2.0bn above its ~14% guide after reserving the full $3bn buyback, and faces up to ~$37bn of additional requirement on its own estimate. In this industry the “capex” line is the regulator’s.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? Free cash flow is not meaningful for a bank — operating cash flow is dominated by balance-sheet movement (1H26: $8,054m, uninformative). The correct analog is organic CET1 generation, and it is clean: roughly $7.9–9.4bn in 1H26 against $5,840m of reported profit. Uses in 1H26: $3,404m of dividends ($1.10 a share, paid 23 April 2026), $3,337m of treasury purchases, and ~$1,500m consumed by risk-weighted asset growth. UBS distributed $6,741m against $5,840m of earnings — a 115% payout — and still grew CET1 capital. The stated philosophy is a ~14% CET1 ratio, a progressive dividend, and buybacks as the residual. Cumulatively 2021 to July 2026: ~$16.3bn of buybacks and ~$12.2bn of dividends, ~$28.5bn — almost exactly the negative-goodwill gain. UBS has, in effect, distributed the Credit Suisse windfall.

Significant acquisitions recently? One, and it defines the company: Credit Suisse, completed 12 June 2023 for 176m UBS shares (CHF 3,223m / $3,547m), plus $162m of assumed share-based compensation and $135m to settle pre-existing relationships — $3,845m all in. Against that, $27,264m of negative goodwill and $108,510m of cash acquired. State support (the CHF 9bn Loss Protection Agreement and CHF 100bn Public Liquidity Backstop) was voluntarily cancelled on 11 August 2023, sixty days after closing, for total fees of roughly CHF 0.73bn, with nothing drawn against the Confederation. All-in outlay including ~$15bn of integration expense is roughly $19.6bn for $27.5bn of day-one tangible book, a $13.5bn annual cost-saving annuity, and a franchise that took Group invested assets from $3,957bn to $7,326bn (+85%). Smaller items: the Swisscard card portfolios transacted with American Express (a $163m pre-tax gain in 1Q26), the Credit Suisse China securities stake sale, and Select Portfolio Servicing.

Buying back shares? Yes, and this is the weakest link in the capital-allocation file.

The 2021–2024 repurchases at 1.06–1.27x tangible book were textbook, and the suspension in March 2023 to fund the acquisition was itself an act of discipline. The July 2026 programme is not: at ~1.96x it pays a ~$1.47bn premium over book, cuts tangible book per share by ~1.8% (each dollar deployed destroys ~$0.49 of TBVPS), and yields ~7–8% against a ~10% cost of equity. It is accretive only if one underwrites the 2028 ambition in full and a softened Swiss rule.

Issuing large amounts of new shares to insiders? No. Share-based compensation was $652m in 1H26, ~2.3% of revenue — modest for a bank. Share count is falling: shares outstanding are down 3.4% year on year, with 63.8m cancelled in 2Q26 and 115m of the 216m treasury shares held for cancellation. Dilution is not a risk here.

Compensation policy of directors/management? Better designed than most, with two real flaws. FACT: Ermotti’s 2025 award was CHF 14,921,193 (~$18.0m), flat year on year in a year net profit rose 53%, of which only CHF 4,920,000 was realized; Kelleher’s chair fee was CHF 5,500,000 with 50% in blocked shares; the Group Executive Board aggregate was CHF 145.3m across 15 members. The long-term plan is 50% average reported RoCET1 / 50% relative total shareholder return over three years, with the 2025 RoCET1 range raised to 8–16% and rTSR measured against a listed G-SIB index. The annual CEO scorecard weights pre-tax profit 20%, cost/income 20%, RoCET1 20%, non-financial 30%, behaviours 10%.

There is no revenue, asset-growth, invested-assets or size metric anywhere — a genuine positive by this framework’s standards. The flaws: (i) no tangible-book-value-per-share metric, while RoCET1 carries ~70% of combined short- and long-term weight and is mechanically flattered by shrinking the capital base — an incentive pointing directly at buying stock at any price; (ii) an 8% RoCET1 floor paying 33% is undemanding for a bank that earned 17.5% standalone. Note the honest half of the scorecard: the 2022 award vested at 71.56% because relative TSR scored 43.12%, running 17.45 percentage points below the G-SIB index — the one metric the committee could not adjust. Ethos recommended rejecting both the compensation report and the buyback; the 2026 AGM approved compensation with ~93.5%.

Motivations of management? Ermotti returned specifically to execute the integration, which completes at end-2026 — key-person and succession risk is highest precisely when the mandate is fulfilled, and the 2024 Group Executive Board reshuffle (Khan to APAC, Karofsky to the Americas) formalised the contest. OPEN QUESTION on alignment: as a foreign private issuer UBS’s directors and officers are exempt from Section 16 and file no Forms 3/4/5 — every Form 4 under UBS’s CIK is UBS reporting as a >10% owner of other issuers. Ermotti held 2,930,928 shares (0.222%) at end-2025, up from 2,755,505, but every increase is explained by mandatory fee-settlement in blocked shares and by vesting; his realized-compensation table shows CHF 0 from equity plans in both 2024 and 2025. There is no evidence of a single discretionary open-market purchase by any UBS director or executive in 2024–2025 — and equally no selling signal. Individual SIX Article 56 filings could not be retrieved, so this is inferred from the Annual Report rather than proven from the register.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? UBS Group AG registered shares trade on the SIX Swiss Exchange as UBSG and on the NYSE as UBS. The NYSE line is a direct listing of the ordinary registered share, not a sponsored depositary receipt with a ratio — one NYSE share equals one Swiss registered share, which is why UBS’s own market-capitalization disclosure reconciles directly to the NYSE price. It is not an MLP and issues no K-1; US holders receive Form 1099. Swiss withholding tax of 35% applies to dividends, reducible to 15% for US residents under the US–Switzerland treaty via reclaim or at-source relief — a real friction for US taxable accounts and a reason the effective yield to a US holder is below the ~2.1% headline. ASSUMPTION: individual treatment varies; this is not tax advice.

Dividend policy? Progressive and consistently delivered. Declared per share: $0.55 (2022), $0.70 (2023), $0.90 (2024), $1.10 (2025) — the last paid 23 April 2026 at a cost of $3,404m — with 2026 accruing at a guided “mid-teen percentage increase” (~$1.25–1.27, ~$3.9bn). Trailing yield ~2.08%. Combined with the ~1.9% buyback yield, total capital return is roughly 4% — modest against US peers, and the buyback half is explicitly conditional on the Swiss Parliament.

How profitable is the business? See above: underlying return on tangible equity of 14.4% and return on CET1 capital of 16.7% in 1H26, having been 4.1% and 4.2% in FY2023. Divisionally the dispersion is extreme — GWM at a 21.8% return on attributed equity (2Q26 underlying), P&C at 15.3%, the Investment Bank at 23.5%, and within GWM a regional spread from 15.9% (Americas) to 44.6% (APAC) pre-tax margin.

Is net income diverging from cash from operations? No — and this is the strongest part of the financial file. For a bank the meaningful test is conversion of reported profit into regulatory capital, and UBS converts at more than 100%: 1H26 reported profit of $5,840m against organic CET1 generation of roughly $7.9–9.4bn, with CET1 capital rising $1,202m despite a 115% distribution payout. There is no accrual-versus-cash divergence. The divergence that does exist is elsewhere: between earnings and per-share book value. Tangible book per share went from $26.93 to $26.89 in the half despite $5.8bn of earnings, because buybacks executed above tangible book are dilutive to it by construction; TBVPS has compounded at only ~3–4% a year since FY2023.

Risks & Downside

What factors would cause the stock to decline? In descending order of expected impact: (1) the Swiss Banking Act adopted at 100% foreign-participation backing, scaling underlying RoCET1 from 16.7% toward ~12.5–13.1% and capping the buyback for a decade — the Council of States committee sits 10, 11 and 31 August 2026; (2) an adverse AT1 ruling from the Federal Supreme Court against CHF 16bn of unprovisioned notional; (3) a capital-markets slowdown normalizing the Investment Bank from 24% toward its own ~15% ambition, removing ~12–13% of group pre-tax profit; (4) an equity-market drawdown, given that >80% of industry revenue growth is market-driven and invested assets are at a record; (5) a failure of GWM Americas to inflect; (6) the mechanical 2027 headwinds — tax normalizing to ~23%, purchase-accounting accretion (~13% of net interest income) running off, credit normalizing from 6bps.

Risk of a catastrophic loss? Low, and materially lower than at any point since 2023. INTERPRETATION. CET1 stands at 14.4%, the going-concern ratio at 19.0%, total loss-absorbing capacity at 38.4%, LCR at 177.3% and NSFR at 115.1%; the SNB’s own Financial Stability Report 2026 states UBS “already has sufficient capital to meet the requirements for full capital backing,” with Group eligible CET1 exceeding fully-applied 2030 requirements by $9bn. Litigation is materially de-risked — provisions down 39%, the two largest cash items settled and paid, Greensill dropped from the disclosed list, and the reasonably-possible range at $0–1.5bn. The credit book is dominated by Swiss mortgages and Lombard lending against liquid collateral, with a 1.0% credit-impaired ratio.

The genuine tail risks are three: the AT1 remedy (CHF 16bn notional, unprovisioned, ~10% of market capitalization — the only unquantified item in the file); a Swiss political rupture that made the domicile untenable (very low probability — Kelleher told the 2026 AGM “We want to remain headquartered in Switzerland,” and no UBS transcript reviewed contains a redomiciliation threat); and the generic G-SIB tail of a trading or operational-risk event, noted because $24.0bn of Non-core and Legacy’s $27.7bn residual risk-weighted assets is operational risk. Note also that liquidity buffers, while ample, are the tightest of the post-acquisition period (LCR 215.7% → 177.3%; NSFR 124.1% → 115.1%).

Chance of a total loss? Remote — but not zero, and the honest answer must acknowledge that this company’s own counterparty demonstrated the mechanism three years ago. Credit Suisse was a 167-year-old G-SIB with a passing capital ratio that lost CHF 110.5bn of assets in one quarter and was resolved over a weekend, wiping out CHF 16bn of AT1 holders entirely. Bank solvency is a confidence variable, not only an arithmetic one. Against that: UBS’s funding is far more granular and wealth-deposit-weighted than Credit Suisse’s institutional base was; its LCR is 177%; the reason for the pending capital legislation is precisely to eliminate the double-leverage weakness that felled Credit Suisse; and the equity ranks behind $343.6bn of issued debt and a 38.4% loss-absorbing-capacity stack. A permanent total loss of equity requires a systemic event, and there is no evidence in the current disclosure pointing to one. The realistic downside is a de-rating from 1.96x toward ~1.2–1.4x tangible book on an adverse regulatory outcome, not impairment.

Recent News & Events

Has the business environment changed recently? Yes, decisively, and on 22 April 2026. The Federal Council adopted two instruments simultaneously: a dispatch to Parliament requiring systemically important banks to back foreign participations fully with CET1 capital (seven-year phase-in from 65% to 100%, entry into force 2028 at the earliest), and final amendments to the Capital Adequacy Ordinance which do not require Parliament and take effect 1 January 2027 — worth ~$4bn of Group derecognition and 0.8 points of CET1 ratio. UBS quantifies the total at ~$37bn; the Federal Council’s effective shortfall is ~$9bn; the annual-cost estimates differ by 3–5x. Parliament debates from summer 2026, with the Council of States committee sitting 10, 11 and 31 August 2026 and weighing 50%, 70% or 80% in place of 100%. Separately and in the opposite direction, the US rescinded the 2023 Basel III Endgame framework on 19 March 2026 and recalibrated the enhanced supplementary leverage ratio from 1 April 2026, cutting required Tier 1 capital 5.6–7.9%. The regulatory jaws are opening against UBS. Management’s own risk hierarchy confirms the shift: the FY2025 20-F promoted “substantial changes in regulation” to the top risk factor, demoting the Credit Suisse acquisition to second.

Significant acquisitions? None new. The 2Q26 development is the completion of the integration itself — all clients migrated, more than 90% of legacy applications retired, >90% of cost synergies realised, and the programme closing at year-end with ~$750m of remaining expense split evenly across 3Q and 4Q.

Change in accounting policies? One presentational change and one estimate revision, both disclosed. From 4Q25, income from certain fair-valued financial instruments not directly linked to client activity moved from transaction-based income to “other revenues” (the line was renamed from “Other income”), applied prospectively — so multi-year transaction-income series are not fully comparable across that boundary. Separately, the FY2023 negative goodwill was revised twice as purchase accounting finalised: $28,925m provisional → $27,748m → $27,264m final. Three vintages circulate in the market; use the restated figure. No change to the reported-versus-underlying framework, which remains disclosed in detail.

Recent changes — new markets, facilities, management? The Swiss booking-centre migration completed in 1Q26, making 2Q26 the first full quarter operating on UBS platforms. UBS was granted an OCC national bank charter on 20 March 2026 — a deliberate doubling-down on the US, notwithstanding that business’s flow problems. The 2024 Group Executive Board reshuffle (Ivanovic to Asset Management, Khan to APAC, Karofsky to the Americas and co-head of GWM) formalised the succession contest. Revised Swiss anti-money-laundering legislation and a new beneficial-ownership transparency register enter into force 1 October 2026, adding due-diligence cost. A strategic update is scheduled for the 4Q26 results in early 2027, at which management has said it will revisit the 2028 ambition — the single most important scheduled catalyst in the file.

Supplemental to the UBS Group AG research memo dated 2026-08-01. No investment recommendation and no price target appears in this appendix.

APPENDIX B — Source Appendix

Report date: 2026-08-01 · CIK: 0001610520 · Filer status: foreign private issuer (files 20-F / 6-K; no 10-K, 10-Q, 8-K or Section 16 Forms 3/4/5 for its own insiders) All URLs accessed 2026-08-01 unless stated.

Authority hierarchy applied throughout. UBS’s own SEC filings and Swiss official documents are PRIMARY and govern. Third-party aggregators (AZI, FactorsToday, ROIC.ai) are TERTIARY and were used only where a filing does not reach; where they conflicted with a filing, the filing won and the conflict is documented in the risk section. Press is SECONDARY and is used only for the parliamentary state of play, where no English-language primary document yet exists.

1. UBS Group AG filings and company primary sources

1.1 Annual reports on Form 20-F (five-year corpus, mirrored locally to output/UBS/sources/20-F/)

1.2 Interim reports and capitalization filings on Form 6-K

1.3 Material-event 6-Ks (the Credit Suisse and capital-reform timeline)

1.4 Form 3/4/5 — the negative result

2. Swiss regulatory and official primary sources

3. Earnings-call transcripts and investor presentations

4. Industry and market data

5. Trade and financial press (secondary — used only where no primary document exists)

Every item below is a secondary source. The parliamentary state of play (5.1–5.4) is the one material area of the article where no English-language primary document yet exists, because a Swiss parliamentary committee’s deliberations are not published contemporaneously. These claims must be labelled INTERPRETATION or “reported” in the article — never FACT.

6. Peer-company primary filings (comp set)

7. Comparable-company public filings consulted (cross-read only)

This is fresh coverage — an initiation. No prior published analysis by this author exists on UBS Group AG or Credit Suisse.

The following listed comparables were consulted through their own public filings for peer multiples, industry framing and cross-read: Morgan Stanley, Goldman Sachs, Charles Schwab, HSBC, Barclays, Santander, Raymond James, Northern Trust, BlackRock, Ameriprise, Evercore, JPMorgan, Bank of America, Citigroup, Wells Fargo, LPL Financial, Julius Baer, Pictet and Lombard Odier. Every peer figure quoted in this article traces to that company’s own reported results at the reference date shown in the comparison table; none is a substitute for reading the peer’s filing directly.

8. Quantitative data feeds

9. Data-quality conflicts — the filed figures govern

Three third-party feeds conflict with UBS’s own filings. In every case the filing wins. This section is the source for the article’s data note.

10. Flagged claims — unsourced, misstated, or requiring correction before publication

Ordered by materiality. F-1 through F-4 are corrections that must be made; F-5 onward are labelling or single-sourcing flags.

F-1 — CORRECT AN ARITHMETIC ERROR IN THE USD 37bn DECOMPOSITION. The components are not “USD 20bn + USD 4bn + USD 15bn” (that sums to 39). Per 1.28, the composition is: at UBS AG standalone, ~USD 20bn (foreign participations) + ~USD 2bn (CAO) = ~USD 22bn; plus ~USD 15bn of Credit-Suisse-driven requirements (USD 9bn concession removal + USD 6bn progressive add-on) = ~USD 37bn. The ~USD 4bn is the Group-level CAO derecognition (−0.8pp on the Group CET1 ratio), a different entity level — adding it to the UBS AG stack double-counts. The Industry-analysis scratch file _scratch/swiss_tbtf_regulatory_timeline.md carries this error (“USD 22bn + USD 2bn = USD 24bn incremental”) and its downstream scenario table (“+USD 24bn: 12.5% RoCET1”) inherits it. Rebuild the scenarios off USD 22bn / USD 20bn / USD 12bn, not USD 24bn.

F-2 — THE “16% AMERICAS MARGIN vs MORGAN STANLEY’S HIGH-20s” COMPARISON IS RIGHT BUT THE FRAMING MUST BE PRECISE. The peer figure is now primary-sourced (6.1): MS Wealth Management pre-tax margin 30% in 2Q26, 30% in 1Q26, 28% in 2Q25 (FY 25%/27%/29%). But UBS GWM in aggregate is not the laggard: 2Q26 GWM reported PBT 1,883 on revenues 7,112 = 26.5%, underlying 1,997 on 6,997 = 28.5% — essentially in line with MS. The gap is entirely regional: Americas 15.9%, APAC 44.6%, EMEA 37.9%, Switzerland 38.8%. The article must say “UBS’s US wealth business earns roughly half the margin of Morgan Stanley’s wealth franchise,” not “UBS’s wealth business is low-margin versus MS.” The latter is not supported by the data and would be a category error.

F-3 — THE HEADLINE “AMERICAS NNFGA OF −USD 4.6bn” IS ONE QUARTER, NOT A TREND. THERE IS A BETTER, MORE DAMNING DATUM. Verified from the regional tables:

Three corrections follow. (i) The margin is improving, not stagnant — 9.3% → 12.7% → 15.9% is the best print in the series, and management’s own target is ~15% in 2026 rising to ~18% by 2028, i.e. it is currently being met. (ii) The −4.6bn NNFGA is a single quarter that followed +8.8bn in 1Q26; 1H26 Americas NNFGA is +4.2bn. Calling it a trend on this evidence is not defensible; calling it a deterioration is (FY2024 +50.7bn → FY2025 +11.7bn → 1H26 +4.2bn annualising to ~8bn). (iii) The genuinely damning number is one the log never uses: on the strict net new money basis — which, per the FY2025 20-F APM definitions, excludes interest and dividends whereas “net new assets” includes them — GWM Americas lost USD 57.7bn of client money in FY2025 while reporting NNA of only −5.9bn. The USD ~52bn wedge is interest and dividends. UBS discloses net new money by region only annually (it is absent from the 1Q26 and 2Q26 interim reports), which is itself worth saying. Lead this section on the FY2025 net-new-money figure and the FY2024→FY2025 NNFGA collapse; carry the 2Q26 NNFGA as a datapoint to watch in 3Q26, not as the proof. Note also that management explicitly declined to explain the NNA/NNFGA gap on the call (“nothing I would call out”) and that 2Q26 Americas NNA was struck after “around USD 10 billion of seasonal tax-related outflows” (Tuckner) — a mitigant that must be reported alongside the number.

F-4 — THE “16.5%–18.5%” REGULATORY CET1 RANGE IS STALE. USE 17.6%–18.4%. The FY2025 20-F (1.1, published 2026-03-09) sketched a Group ratio rising to ~18.5% and then falling to 16.5% because the June-2025 proposal would have eliminated ~USD 11bn of Group CET1 via DTAs on temporary differences, capitalised software and PVAs. The final CAO of 2026-04-22 dropped the DTA measure entirely (1.28: “The treatment of deferred tax assets arising from temporary differences remains unchanged”), cutting the Group elimination from ~USD 11bn to ~USD 4bn. UBS’s own current numbers are therefore a de facto Group minimum of ~18.4%, or ~17.6% after the CAO derecognition. The Lead’s RoCET1 scaling arithmetic (“16.7% scales to 12.6–14.2% on a 14/16.5 to 14/18.5 basis”) is built on the superseded range and should be rebuilt on 14/17.6 to 14/18.4, i.e. ~12.7–13.3%.

F-5 — THREE VINTAGES OF THE NEGATIVE GOODWILL CIRCULATE. STATE WHICH ONE YOU ARE USING. USD 28,925m (4Q23 report, provisional, 1.12) → USD 27,748m (FY2023 20-F as filed, 1.3) → USD 27,264m (final, after IFRS 3 measurement-period adjustments, restated in the FY2024 and FY2025 20-Fs, 1.1). Use USD 27,264m and say “as finalised.” The same applies to the FY2023 return figures: reported RoTE was 42.6% (4Q23 report), 41.3% (FY2023 20-F) and 40.8% (FY2025 20-F, restated) against an underlying 4.0%/4.1%/4.1%. The task’s “40.8% vs 4.1%” is CONFIRMED on the restated basis — cite the FY2025 20-F, not the FY2023 one. Note also that FY2023 net profit appears as 27,366 (attributable to shareholders) and 27,382 (total, incl. non-controlling interests); the SEC-sweep entry uses the latter and the Lead the former. Standardise on “attributable to shareholders” throughout (FY2025 7,767 / FY2024 5,085 / FY2023 27,366).

F-6 — CUT OR RE-SOURCE THE 1Q26 ERMOTTI QUOTES. “There is a broad agreement… that the announced measures would require UBS to hold around CHF 22 billion in additional capital” and “that CHF 22 billion of capital would be trapped and unproductive” rest solely on an Investing.com transcript aggregation (3.6). ROIC carries no 1Q26 or 2Q25 UBS transcript. These quotes must be verified against UBS’s own 1Q26 webcast, results presentation (3.7) or 6-K before appearing in the article — or cut. The substance (USD 22bn) is independently primary-sourced at 1.1 and 1.28 and is unaffected; note also that the aggregator renders the figure in CHF while UBS’s own filings state it in USD, which is itself a signal the transcription is unreliable.

F-7 — THE PARLIAMENTARY STATE OF PLAY IS SECONDARY-SOURCED ONLY. LABEL IT. Every claim about the WAK-S committee — the 10/11/31 August meeting dates, the 50%/70%/80% variants, partial AT1 substitution, a September decision — comes from press (5.1–5.4), not from a primary parliamentary document. reuters.com is not fetchable from this environment, so 5.1 could not be independently verified; 5.2 (FStech) and 5.3 (SRF) provide corroboration and should be cited alongside it so the claim is not single-sourced. In the article these belong as INTERPRETATION / “reported”, never as FACT. The primary and therefore FACT-grade statements are only: the dispatch was adopted and the CAO finalised on 2026-04-22 (2.1); the CAO enters force 1 January 2027 (2.1); the phase-in is seven years from 65% (1.28, 2.3); and “Parliament will be able to debate the legislative proposal from summer 2026” (2.1, verbatim).

F-8 — “INVESTED ASSETS 7,326bn vs ~4,200bn PRE-DEAL” USES THE WRONG BASE. UBS’s invested assets were 3,957bn at FY2022 (1.4), 4,596bn at FY2021 and 4,187bn at FY2020. “~4,200bn” is the FY2020 figure, not the pre-deal FY2022 one. Use USD 3,957bn (FY2022) — which makes the point more forcefully, not less (7,326 / 3,957 = +85%, versus the +74% the log’s number implies).

F-9 — THE “~USD 22bn PHASED OVER SEVEN YEARS” SHORTHAND IS LOOSE. Only the ~USD 20bn foreign-participations tranche phases over seven years (from 2028 at the earliest, at 65% rising 5pp/yr). The ~USD 2bn/USD 4bn CAO effect is already final law effective 1 Jan 2027 (PVAs) and 1 Jan 2029 (software), with no parliamentary veto; the ~USD 15bn CS-related tranche began phasing 1 January 2026 and completes 1 January 2030. Split the USD 37bn by contingency and timing, not by a single phase-in.

F-10 — MINOR RECONCILIATION ITEMS. (i) The 1Q26 reported-to-underlying revenue bridge is USD 599m (USD 472m PPA/integration plus USD 128m Swisscard as a separate line), not “USD 472m incl. USD 128m”; the log’s parenthetical is wrong. (ii) The “In July 2026, we completed our latest share repurchase program of USD 3bn” quotation is verbatim — but from the “Share information and earnings per share” section; the “Other developments — Capital returns” version of the same sentence omits “of USD 3bn”. Cite the former. (iii) Treasury shares held for cancellation are 63,776,550 (2024 program) + 52,582,575 (2025 program) = 116.4m, not “115m”. (iv) The FY2021 dividend per share of USD 0.50 was not verified from a filing this session — either verify it against the FY2021 20-F (1.5) or drop it. (v) UBS’s “gross cost savings” and integration-expense figures are stated “assuming constant foreign-exchange rates compared with 30 September 2023” — quote the caveat when quoting the USD 13.5bn / USD 14.2bn / USD 15bn numbers.

F-11 — DATE-INTEGRITY CHECK. No source materially relied upon is older than ~18 months except where deliberately used for historical context: the FY2021/FY2022 20-Fs (the pre-deal baseline), the 2023 event 6-Ks (the acquisition timeline), the FINMA 2023-03-19 AT1 communication, and the 2011 Deutsche Bank Drive file (rejected as unusable, 7.9). The FY2025 20-F is 4.8 months old and, on the capital file specifically, is superseded by the 2026-04-22 documents (1.28, 2.1) — do not quote its 18.5%/16.5% figures as current (see F-4).

11. Coverage gaps recorded honestly

• No US insider signal exists (1.35). SIX Art. 56 LR management-transaction filings could not be retrieved (JavaScript portal), so the absence of open-market director purchases is inferred from the FY2025 20-F ownership and realized-compensation tables, not proven.
• No 1Q26 or 2Q25 earnings-call transcript is available from ROIC or Drive (3.6); the filed 1Q26 investor presentation (3.7) is the substitute.
• No non-earnings event transcripts (investor day, conference fireside, the separate post-results media Q&A) were located anywhere (3.9).
• No usable internal sector primer for wealth management or Swiss banking exists in the Drive (7.9).
• No primary parliamentary document on the WAK-S deliberations (F-7).
• The AT1 Supreme Court exposure is unquantified and unprovisioned by UBS; Note 17 does not size it, and the CHF 16bn notional (2.5) is a theoretical ceiling, not an estimate. Whether liability would fall on FINMA/the Confederation or on UBS as CSG successor is unresolved.
• Short interest / days-to-cover for the ADR is not available from the feeds used; the variant-perception section positioning rests on the factor read alone.
• UBS has not updated the USD 37bn estimate in its 1Q26 or 2Q26 interim reports, and has not restated the foreign-subsidiary shortfall for the ~USD 9bn (4Q25) + ~USD 3bn (UK) + ~USD 2bn (US IHC) of subsidiary upstreaming, which the CFO conceded would make the number “lower” without giving it.