Sony Group Corporation (NYSE: SONY) — The Loss That Never Happened

Report date: 26 July 2026 · Primary listing: Tokyo Stock Exchange Prime (6758) · ADR: NYSE: SONY (1 ADS = 1 ordinary share) Fiscal year end: 31 March · Reporting currency: JPY · Basis: IFRS · SEC status: foreign private issuer (Form 20-F / 6-K) Last reported period: fiscal year ended 31 March 2026 (Sony labels this “FY2025”) · Next print: 6 August 2026 Reference price: ¥3,409 (TSE, 24 July 2026) / $20.98 (ADR, 24 July 2026)

Note on fiscal-year labels: Sony calls the year ended 31 March 2026 “FY2025.” To eliminate ambiguity this article uses period-end labels throughout — “FYE Mar-26” means the year ended 31 March 2026. Where guidance is discussed, “FY2026” is used only as management’s own label for the year ending 31 March 2027, and is flagged as such.

⚡ Claude’s Take

This block is the author’s own independent opinion. It is general information, not investment advice, and it is not a recommendation to buy or sell any security. The analysis in Sections 1–15 below takes no position, contains no price target, and makes no buy or sell recommendation. Only this block does.

Verdict: BUY — accumulate, with patience. Attractive accumulation zone ¥3,000–3,500 (≈$18.50–$21.50 ADR at ¥162/USD), which is roughly 10.5–12.0x EV/EBIT and 15.5–18.0x earnings on management’s own FY2026 guidance. Base-case fair-value zone ¥4,000–4,400. Conviction: MEDIUM.

Tag: the screen says loss-making; the cash flow statement says a 7% yield.

Sony reported a net loss of ¥326.9bn for the year ended 31 March 2026. It did not lose ¥326.9bn. The entire loss — and more — is a single non-cash accounting entry: when Sony executed the partial spin-off of its Financial Services business on 1 October 2025, ¥1,377.8bn of accumulated other comprehensive income tied to Sony Life’s bond portfolio was recycled out of the equity reserve and through the income statement. Those unrealised bond losses had already been charged against book value in prior years. The demerger just moved them from one line to another. The proof is on the balance sheet: retained earnings fell ¥1,383.3bn, other reserves rose ¥1,795.8bn, and equity attributable to owners was essentially unchanged at ¥8,119.0bn. Meanwhile the continuing business earned record operating income of ¥1,447.5bn, generated ¥1,487.9bn of free cash flow, and ended the year with ¥1,166.9bn of net cash. Trailing reported EPS is negative, so Sony now returns null on every P/E screen in the market — I verified this on two independent data vendors. A large, liquid, profitable company has been mechanically removed from the value screens at precisely the moment its balance sheet became the cleanest it has been in thirty years.

That is the opportunity, and I want to be careful not to oversell it, because the operating case is only good, not spectacular. Management guides FY2026 operating income to ¥1,600bn, which sounds like 10.5% growth but is flat once you normalise last year’s ¥120.1bn Bungie impairment, ¥27.1bn Pixomondo write-off, ¥36.4bn of sensor restructuring, ¥44.9bn Sony Honda Mobility loss and the offsetting ¥34.7bn Peanuts gain and ¥43.9bn land gain — underlying FYE Mar-26 operating income was roughly ¥1,597bn. So you are paying ~11.9x EV/EBIT for a flat year in which memory prices tax the console and TV businesses, image sensors are guided down, and PS5 shipments just fell 46% in a quarter. The bear case is neither stupid nor fully priced out. What tips it for me is the mix underneath: PlayStation set a record operating profit while hardware collapsed, because Network Services grew 13.9% and Digital Software grew 5.4%; Music compounded 15.1% with a 19.4% clean margin against a non-substitutable rights position; and the TSMC joint venture converts the segment the market hates most from a fixed-cost fab into a variable-cost design house. Add a ¥500bn buyback being executed into the weakness, a raised dividend, net cash, an unvalued 16.4% stake in the now-listed financial arm, and a Spotify holding nobody counts, and the downside looks well-supported around 10x EV/EBIT.

The framing is contrarian value in a business whose earnings quality is improving while its reported earnings look broken — explicitly not momentum, and, on the evidence, no longer a falling knife either. The factor model is unambiguous: momentum loading is slightly negative (−0.06), value only marginally positive (+0.03), and the tape has been punishing — one-year return −14.1%, Sharpe −0.54, drawdown −36.0%. But the three-month annualised return has flipped to +14.5% against a six-month figure of −17.0%; the stock bottomed at $19.32 on 25 June and is 8.6% off that low while still 30.5% below the November 2025 high. The decline has stopped. The most uncomfortable factor finding, and the one that keeps conviction at medium rather than high, is that Sony’s closest factor neighbours are Japan index ETFs and a Japanese broker — empirically the market trades this as Japan-beta with an AI kicker (Country: Japan +0.600, Robotics & AI +0.310, Quality −0.10), not as the entertainment compounder management describes. You are underwriting a re-rating that requires the market to change its mind about what kind of company this is.

Conviction: MEDIUM. What would flip me bullish (high conviction): the August and November prints showing Network Services and Music streaming still compounding at high single digits or better with I&SS holding its guided ¥400bn while the TSMC JV moves to a definitive agreement with disclosed capex relief — that would confirm the annuity is carrying the cycle and the capital intensity is genuinely falling. What would flip me bearish: evidence that Music streaming growth decelerates below mid-single digits, or that Sony deploys a meaningful slice of the ~¥800bn undeployed strategic-investment frame on another large studio or platform acquisition. This management is superb at structural surgery and demonstrably poor at buying operating businesses — Bungie has now been impaired twice on a ~$3.6bn purchase, Pixomondo was bought and shut, and Sony Honda Mobility was abandoned. Another Bungie would break the thesis faster than any memory-price cycle.

📈 Stock Price Action — Five-Year Event Map

Over the trailing five and a half years the ADR has traced a violent round trip: a $11.95 low on 12 October 2022, a $30.18 high on 12 November 2025, and $20.98 today (24 July 2026) — 30.5% below the high and 8.6% above the 52-week low of $19.32 set on 25 June 2026. The 52-week range is $19.32–$30.18. The shares trade at the 50-day EMA ($20.98) and roughly 6.5% below the 200-day EMA ($22.44). On the Tokyo line, 6758 closed at ¥3,409 on 24 July 2026.

Cycle narrative. (1) The 2021 advance was the PS5 launch working exactly as intended, with image sensors recovering alongside. (2) The 2022 collapse was mostly macro — a global multiple compression in which Sony had a specific idiosyncratic shock, Microsoft’s January 2022 agreement to acquire Activision Blizzard, which the market read as a direct threat to PlayStation’s content position. (3) The 2023 recovery tracked PS5 supply normalising into what proved to be the peak shipment year. (4) The 2023–24 plateau reflected a maturing console cycle and a Pictures segment absorbing the Hollywood labour stoppages. (5) The 2024–25 advance was the re-rating that mattered: Sony obtained METI approval for the Corporate Restructuring Plan in February 2024, listed SFGI on the TSE Prime Market in September 2025, executed the partial spin-off on 1 October 2025, split the stock 5-for-1, and lifted buybacks from ¥203.0bn to ¥285.5bn to ¥522.1bn across three years. The market paid up for a simpler, more focused Sony. (6) The decline from 12 November 2025 is the subject of this article, and Sony’s own CEO named the two causes when a Nikkei reporter asked him directly on 8 May 2026 — the memory shortage (“growth is really — might be inhibited… there might be a deterioration in the cost structure”) and the fear that AI-generated content abundance will compete away entertainment’s share of user time. Notably, the FY2025 results and the TSMC announcement on 8 May were received well: the ADR went from $19.89 on 7 May to $22.79 on 13 May, a 14.6% gain. The subsequent slide to a new low on 25 June was therefore not a verdict on the results. (7) Since late June the stock has stabilised, with the ¥500bn repurchase running (37.1m shares for ¥127.5bn through 30 June) and the inaugural US-registered bond pricing at +70bp / +90bp — spreads that imply a solidly investment-grade credit.

Caveat, stated rather than assumed away: the adjusted ADR series shows no visible discontinuity across the 1 October 2025 SFGI dividend-in-kind distribution (29 Sep $29.02 → 30 Sep $28.71 → 1 Oct $28.66). Either the distributed value was small relative to Sony’s capitalisation or the series does not adjust for it. Total-return comparisons spanning that date therefore carry a modest, unquantified break.

This block is factual price history. It carries no recommendation and no price target.

1. Executive Summary

Sony Group Corporation is, for the first time in three decades, a comprehensible company. On 1 October 2025 it completed the partial spin-off of its Financial Services business — Sony Life, Sony Bank and the associated insurance operations — distributing the majority of SFGI to shareholders in kind and retaining 16.40%. The effect on the balance sheet was violent and entirely beneficial: total assets fell from ¥35,293.2bn to ¥15,683.5bn, total liabilities from ¥26,783.0bn to ¥7,169.9bn, and net debt of ¥617.8bn became net cash of ¥1,166.9bn. The tangible common equity ratio went from 13.2% to 33.6%. What remains is an entertainment and imaging company: Game & Network Services, Music, Pictures, Entertainment Technology & Services, and Imaging & Sensing Solutions, with management noting that entertainment, IP and creation technology now represent 67% of consolidated sales.

The accounting cost of that transaction is the central fact of this article. On execution, ¥1,377.8bn of accumulated other comprehensive income directly related to the disposal group was transferred into the income statement as a loss — ¥1,640.1bn of unrealised losses on fair-value-through-OCI debt instruments held at Sony Life, partly offset by ¥263.3bn of insurance finance income. The result is a headline consolidated net loss attributable to stockholders of ¥326.9bn for the year ended 31 March 2026, against net income of ¥1,141.6bn the prior year. This loss is non-cash, was already reflected in book equity, and destroyed nothing. Continuing operations earned ¥1,030.9bn attributable net income on ¥12,479.6bn of sales, with record operating income of ¥1,447.5bn (11.6% margin) and free cash flow of ¥1,487.9bn.

The operating year underneath was strong but uneven. Three segments set records. I&SS grew sales 19.6% to ¥2,151.5bn with operating income up 36.8% to ¥357.3bn. Music grew sales 15.1% to ¥2,120.1bn with operating income up 25.1% to ¥447.0bn. G&NS set a record ¥463.3bn operating income on essentially flat sales — a genuinely important result, because it was achieved while hardware revenue fell 12.1% and March-quarter PS5 shipments fell roughly 46% to 1.5m units, the console’s weakest quarter since launch. Network Services revenue rose 13.9% and Digital Software & Add-on Content rose 5.4%. Pictures was weak (operating income −10.6% to ¥104.9bn, a 7.0% margin) and ET&S was weaker (sales −6.2%, operating income −16.9%, Displays −20.3%).

The forward year is flat, and management’s framing obscures it. FY2026 guidance of ¥1,600bn operating income represents 10.5% growth against reported FYE Mar-26 — but FYE Mar-26 carried roughly ¥150bn of net one-time charges (Bungie ¥120.1bn, Pixomondo ¥27.1bn, I&SS restructuring ¥36.4bn, Sony Honda Mobility ¥44.9bn, less the Peanuts gain ¥34.7bn and a ¥43.9bn land gain). Normalised, FYE Mar-26 operating income was approximately ¥1,597bn, and the guide is flat. Management effectively concedes this segment by segment on the call. Absorbed inside that flat guide are ~¥30bn of memory-cost impact in ET&S, ~¥20bn of TCL joint-venture implementation costs, ~¥30bn of further Sony Honda Mobility losses, and increased next-generation platform investment.

Two structural decisions define the forward thesis. The TSMC joint venture (non-binding MOU, 8 May 2026) would place next-generation image-sensor development and production in Sony’s new Kumamoto fab with Sony as majority and controlling shareholder — Sony’s first move to a “fab-light” model after decades as an integrated device manufacturer. Management states it will improve I&SS cash flow, reduce invested capital and improve profitability. The TCL partnership (definitive agreement, March 2026, JV operational April 2027) hands the television, B2B display, home theatre and home audio businesses into a joint venture with a Chinese panel-integrated competitor. Both are exits from capital intensity in businesses where Sony’s advantage is eroding or absent.

Capital allocation is the sharpest divide in the analysis. Structurally it is excellent — the demerger, the fab-light pivot, the TV exit, and a shareholder-return programme that took buybacks from ¥203.0bn to ¥285.5bn to ¥522.1bn over three years, with a ¥500bn facility and a ¥35 dividend for FY2026 and ¥127.5bn already repurchased by 30 June. On operating M&A it is poor: Bungie has been impaired twice against a ~$3.6bn purchase price, Pixomondo was acquired and wound down, Sony Semiconductor Israel was sold at a ¥19.9bn loss, and Sony Honda Mobility’s Afeela EV has been abandoned. Roughly ¥800bn of the ¥1.8tn strategic-investment frame remains undeployed.

At ¥3,409 the shares carry a market capitalisation of approximately ¥20.3tn and an enterprise value of approximately ¥19.1tn — 17.6x FY2026 guided earnings, 11.9x FY2026 guided EV/EBIT, 1.55x EV/sales, and a 7.3% trailing free-cash-flow yield, before crediting the retained 16.40% SFGI stake or the Spotify holding, neither of which is valued here. The market is underwriting a company whose operating profit does not grow in real terms. Whether that is correct is the question Sections 10 and 11 address.

No recommendation and no price target appear in this section or anywhere in Sections 1–15.

2. Business Overview

Sony is a Japanese holding company operating five reportable segments plus an “All Other” category, following the deconsolidation of Financial Services on 1 October 2025. It employed 94,900 people at 31 March 2026, down from 112,300 a year earlier — a reduction of 17,400, of which 14,300 were the departing Financial Services headcount and the balance came from ET&S restructuring in Japan and I&SS divestitures outside Japan. Roughly 8% of employees are union members.

Segment composition, fiscal year ended 31 March 2026

Segment sales are stated before elimination of intersegment transactions and therefore sum to more than consolidated sales. Source: Form 20-F, fiscal year ended 31 March 2026.

How each business actually makes money

Game & Network Services is a two-sided platform. Sony sells PlayStation consoles at or near cost — sometimes below — and monetises the installed base through three annuity streams: a platform fee on third-party software sold through the PlayStation Store, PlayStation Plus subscriptions, and first-party software published by Sony’s own studios. The revenue disclosure makes the model visible. Of ¥4,570.1bn of external sales in FYE Mar-26, Digital Software & Add-on Content contributed ¥2,415.3bn (+5.4%), Network Services ¥763.1bn (+13.9%), and Hardware & Others ¥1,391.6bn (−12.1%). Digital and network revenue together are 69.5% of segment external sales and growing; hardware is 30.5% and shrinking. Monthly active users reached a record 125 million accounts in March 2026 (+1%), total quarterly play time rose 1%, and cumulative PS5 shipments exceeded 93 million units.

Music is three businesses: Recorded Music (owning master recordings and taking a share of streaming, physical, live and merchandising revenue), Music Publishing (owning compositions and collecting mechanical, performance and synchronisation royalties), and Visual Media & Platform (anime production via Aniplex, mobile game applications, and related platforms). The revenue split for FYE Mar-26 was Recorded Music streaming ¥852.7bn (+8.1%), Recorded Music other ¥492.7bn (+21.0%), Music Publishing ¥419.9bn (+10.5%), and Visual Media & Platform ¥325.3bn (+33.1%). This is the group’s most profitable segment at a 21.1% reported margin, and it is asset-light in operating terms — the assets are copyrights, carried as intangibles.

Pictures comprises Motion Pictures (production and distribution of theatrical film), Television Productions (scripted and unscripted series for third-party networks and streamers), and Media Networks (television channels and, critically, the Crunchyroll anime direct-to-consumer platform). The mix shifted materially in FYE Mar-26: Motion Pictures fell 18.8% to ¥495.7bn while Television Productions rose 11.6% to ¥512.4bn and Media Networks rose 11.5% to ¥478.3bn. Crunchyroll now has more than 21 million paid subscribers and a library exceeding 50,000 episodes. Sony Pictures is notable among the major studios for having no general-entertainment streaming service of its own — it is an arms dealer, selling content to Netflix, Disney and others, while owning a focused DTC platform in anime.

Entertainment, Technology & Services is the consumer electronics legacy: Imaging (Alpha cameras and professional video, ¥722.5bn, −2.1%), Sound (¥278.8bn, −4.0%), Displays (BRAVIA televisions, ¥476.3bn, −20.3%), Network Services (¥188.3bn, +4.8%) and Other (¥518.9bn, −6.9%). Cameras remain a genuinely strong franchise; televisions are being exited into the TCL joint venture.

Imaging & Sensing Solutions designs and manufactures CMOS image sensors, overwhelmingly for smartphone cameras, with growing positions in automotive, industrial and security. Sales grew 19.6% to ¥2,151.5bn on higher average selling prices, improved mix and higher unit volume. This is the group’s only genuinely capital-intensive business: Sony invested ¥227.4bn and ¥246.7bn in the last two fiscal years principally to expand sensor capacity.

All Other houses disc manufacturing, recording media, and — significantly — the equity-method interest in Sony Honda Mobility, which drove the segment’s ¥74.6bn operating loss via a ¥44.9bn share of loss on the discontinuation of the Afeela EV programme.

Recurring versus non-recurring revenue

Sony does not disclose a recurring-revenue metric, so this must be built. Genuinely recurring or highly repeatable revenue comprises PlayStation Network Services (¥763.1bn), the large majority of Digital Software & Add-on Content (¥2,415.3bn — free-to-play add-on content and live-service spend is highly recurring; premium title purchases less so), Recorded Music streaming (¥852.7bn), Music Publishing (¥419.9bn), Crunchyroll subscriptions within Media Networks (¥478.3bn), and ET&S Network Services (¥188.3bn). Conservatively treating half of Digital Software and half of Media Networks as recurring gives approximately ¥3.7tn, or roughly 30% of consolidated sales, on a durable subscription or royalty basis — with a further large slice in repeatable-but-not-contracted catalogue and licensing income. Non-recurring revenue is concentrated in console hardware (¥1,391.6bn), television and consumer electronics hardware, image sensors (which are recurring at the design-win level but not contracted), and theatrical film.

Verdict on the business model. Sony is now a portfolio of one excellent business (Music), one very good platform business in a bad phase of its hardware cycle (G&NS), one good but cyclical and capital-hungry semiconductor business (I&SS), one poor business being restructured (Pictures ex-Crunchyroll), and one bad business being exited (ET&S Displays). That is a materially better portfolio than the pre-demerger conglomerate, and the disclosure is now good enough for an outside investor to see which is which. The revenue mix is shifting decisively toward annuity income, and the FYE Mar-26 result — a record G&NS profit despite a 46% collapse in quarterly console shipments — is the cleanest possible demonstration of it.

3. Industry Dynamics

Sony operates across four distinct industries with very different structures. Averaging them is the analytical error to avoid.

Recorded music and publishing — the best industry Sony is in

Recorded music is a rights oligopoly selling a non-substitutable input into a fragmented and competitively disadvantaged distribution layer. Publicly reported streaming economics establish the mechanics: the three majors — Universal, Sony and Warner — plus the independent aggregator Merlin control roughly 72% of streams; approximately 68% of Spotify’s total revenue flows to rights holders; and label contracts are structured to protect the labels through most-favoured-nation clauses, royalties set at the greater of a revenue percentage or a per-stream floor, minimum guarantees, audit rights and change-of-control provisions that function as kill-switches. No credible streaming service can launch without the major catalogues, and no major catalogue can be replicated.

The industry is also growing. Sony’s own dollar-basis streaming revenue rose 9% in Recorded Music and 14% in Music Publishing in FYE Mar-26, and management guides the mid-to-long-term market to mid-to-high single-digit growth. The recent addition of subscription price increases by the platforms flows disproportionately to rights holders under the revenue-share structure.

Structural verdict: an excellent industry. High and stable concentration upstream, non-substitutable product, contractual protections, and secular growth. The one genuine question is whether generative AI content dilutes the pool — addressed in Section 11.

Interactive entertainment / console platforms — a good industry entering a difficult transition

Console gaming is a three-player oligopoly (Sony, Microsoft, Nintendo) with enormous barriers to entry: an installed base numbering in the hundreds of millions, developer relationships built over decades, exclusive first-party studios, and an ecosystem of accounts, saves, friend graphs and entitlements that creates real switching costs. Profit pools sit with the platform holder’s store take-rate and subscription revenue, not with hardware.

The industry is in an awkward phase. Console hardware unit volumes are declining industry-wide as the installed base matures and mobile and PC absorb marginal engagement. Sony’s PS5 shipped roughly 16m units in FYE Mar-26 against 18.5m the prior year and a 20.8m peak, with the March quarter at 1.5m versus 2.8m — a ~46% decline and the worst quarter since launch. The memory-price shock is directly relevant here: memory is a material component of console bill-of-materials, and prices reportedly roughly doubled in a quarter with a further sharp rise forecast. Sony has already raised the PS5 price to $650 from $550 and says no further increase is planned; management will instead set FY2026 hardware volumes based on “the volume of memory we can procure at reasonable prices.”

Structural verdict: a good industry with a difficult cyclical overlay. The oligopoly and the annuity economics are intact — and are demonstrably working, given the record segment profit. But the platform holders face a genuine strategic question about the next hardware generation’s cost structure, which Totoki addressed candidly: memory prices are expected to remain high in FY2027, and Sony has not decided the timing or price of the next console, is considering “how can we reduce the other costs of the hardware other than the semiconductor,” and may consider “new ways of selling the product” and “changing business models.”

CMOS image sensors — a concentrated, capital-hungry, cyclical good business

The image sensor market was approximately $25.6bn in 2025 on 8.1bn units, forecast at roughly $27.4bn in 2026 and $39.9bn by 2031. Concentration is high: the top five suppliers hold roughly 83%, with Sony the clear leader, Samsung around 20% and OmniVision around 11%. Published share figures for Sony vary widely by methodology — measures in the 40s on unit-inclusive bases and above 50% on high-end and revenue bases are both in circulation — so the honest statement is that Sony is the dominant supplier in the high-value segment of a concentrated market, without a precise point estimate.

Barriers to entry are real but of a specific type: they are process and analog-design know-how accumulated over decades, plus the capital to build and continually re-equip fabs. Totoki’s own description is the useful one — “we scrutinize and optimize every aspect of the sensor from the pixel structure, stacking and layering technologies through to the processes and final packaging… deep expertise cultivated over many years in analog domain spanning design, development and manufacturing.” That is a genuine cost/supply advantage in the Greenwald sense.

Applying the Marathon capital-cycle lens is instructive. Sony has been a heavy net investor in sensor capacity — ¥227.4bn and ¥246.7bn in the last two years — into a market whose near-term demand it now describes cautiously: management expects the trend toward larger smartphone sensors to moderate and has embedded a slight year-on-year decrease in mobile sensor sales in the FY2026 forecast. Heavy capital deployed into a decelerating cycle is precisely the setup Marathon warns about. The TSMC joint venture is the correct response: it converts the marginal capacity decision from a Sony balance-sheet decision to a shared one, and Sony explicitly expects it to “reduce invested capital.”

Structural verdict: a good industry, but the least attractive of Sony’s three core positions on a risk-adjusted basis — capital-intensive, cyclical, exposed to smartphone unit demand and to a single very large customer base in high-end handsets, and now entangled in semiconductor industrial policy on both sides of the Taiwan Strait.

Film and television production — a structurally poor industry

Theatrical film has no durable barrier to entry beyond capital and relationships. Returns are hit-driven, marketing costs are enormous, and the streaming transition transferred bargaining power to the distribution platforms. Sony’s own numbers testify: a 7.0% segment operating margin, a ¥27.1bn write-off of an acquired VFX business, and Motion Pictures revenue down 18.8% in a year. Sony’s 20-F risk factors state the problem plainly — “increasing concentration of digital music distributors and creation of content by distributors themselves” and the risk that “the prevalence of digital streaming networks and other new media on traditional television and in-theater motion picture viewership could adversely affect the operating results of the Pictures segment.”

Sony’s chosen position — arms dealer to the streamers rather than operator of a competing general-entertainment service — is the right call for a sub-scale player and avoids the capital destruction visible at several peers. But it does not create a moat; it avoids a worse outcome.

Structural verdict: a bad industry, correctly played. The exception is Crunchyroll, which operates in anime — a genuinely growing global category where Sony holds licensing relationships with Japanese publishers, a 50,000-episode library, and 21m+ paid subscribers. That is a different and better business inside a worse one.

Consumer electronics / televisions — a bad industry

Displays fell 20.3%. The 20-F cites “intensified competition in Displays and additional U.S. tariffs.” Sony competes against vertically integrated Chinese and Korean panel makers who own the most expensive input. There is no path to a durable advantage, and Sony is exiting into a joint venture with TCL — one of those very competitors.

Structural verdict: a bad industry, being exited. Correct.

Cross-cutting regulatory and geopolitical factors

Sony carries genuine, unhedgeable geopolitical exposure: a Japanese manufacturing base, Chinese and Taiwanese supply chain dependencies, a US end market and US tariff exposure, and a semiconductor business now explicitly a subject of industrial policy. The TSMC JV investment and further capital investment at Sony’s existing Nagasaki plant are premised in part on Japanese government support. Asked directly about US reciprocal tariffs following a court ruling, Totoki offered only that uncertainty had increased and Sony would “be quite flexible.” That is an honest answer and an uncomfortable one.

4. Competitive Position

The question this section must answer is whether Sony possesses durable competitive advantages that show up in financial outcomes, and whether those advantages would deteriorate if removed. Taking each in turn, using the Greenwald taxonomy.

PlayStation — economies of scale reinforced by customer captivity. Real, and financially proven.

The moat mechanism is specific. A 125-million-monthly-active-user installed base makes PlayStation a mandatory distribution channel for any major third-party publisher, which guarantees content, which sustains the installed base — a self-reinforcing loop. On top of it sits customer captivity: a decade of purchased entitlements, saved progress, trophies, friend graphs and PS Plus subscription state that do not transfer to a competing console. The economics of the platform fee on third-party sales are near-pure margin.

The financial proof is the point. In FYE Mar-26, hardware revenue fell 12.1% and March-quarter shipments fell roughly 46%, yet segment operating income set a record ¥463.3bn — and would have been approximately ¥583bn excluding the ¥120.1bn Bungie impairment, up around 45% year on year on management’s framing. Network Services grew 13.9% and Digital Software & Add-on Content grew 5.4% straight through the hardware collapse. That is exactly the financial signature a genuine platform moat should produce, and it is the strongest single piece of evidence in this article. If the moat were not real, a 46% collapse in console shipments would have crushed the segment; instead it set a record.

Against whom? Microsoft has effectively conceded the console-exclusivity battle, publishing former exclusives broadly, and competes primarily through Game Pass and cloud. Nintendo competes in a differentiated first-party family segment rather than head-on. The genuine competitive threat is not another console — it is mobile and PC absorbing engagement hours, and, prospectively, AI-lowered barriers to game creation flooding the market with content. Totoki addressed the latter directly and, in my reading, correctly: lower creation costs mean more games, more games mean discovery becomes the scarce resource, and “our platform’s role will be critical in ensuring players find the right content in an increasingly crowded landscape.” A content glut is, on balance, good for a curation-and-distribution platform that takes a percentage — provided it retains the audience.

Verdict: a durable competitive advantage. It would visibly deteriorate if removed: third-party publishers would defect, the store take-rate would collapse, and Network Services revenue would fall. It is not permanent — no platform is — but it is real and it is currently strengthening in mix terms even as it weakens in unit terms.

Sony Music — intangible-asset customer captivity of the strongest kind. The best asset in the group.

Sony Music’s advantage is ownership of copyrights that cannot be replicated. Not “hard to replicate” — cannot. A streaming service cannot substitute away from a major’s catalogue, which is why roughly 68% of a platform’s revenue goes to rights holders and why the contractual terms so favour the labels. Barriers to entry are absolute for the existing catalogue and very high for new signings, because the majors control the global marketing, radio, playlist and sync relationships that turn an artist into a hit.

Financially: 21.1% reported operating margin (19.4% excluding the ¥34.7bn Peanuts remeasurement gain), 15.1% revenue growth, and streaming growth of 8.1% in recorded music and 10.5% in publishing on a yen basis (9% and 14% on a dollar basis). Sony has been adding to the position deliberately — acquiring the Pink Floyd and Queen catalogues and announcing a music-rights acquisition joint venture with GIC, Singapore’s sovereign wealth fund, which brings third-party capital to catalogue purchases without consuming Sony’s balance sheet.

Verdict: a durable competitive advantage, and the one I would least want to be short. Removing it would eliminate the entire economic rent — Sony Music without owned copyright would be a services business earning a distribution fee.

Imaging & Sensing Solutions — a real supply/cost advantage, but the most fragile of the three

Sony’s sensor advantage is genuine and technical: stacked and layered pixel architectures, analog design depth, process integration and packaging accumulated over decades and defended by continuous capital investment. The market position — clear leader in the high-value segment of a market where the top five hold ~83% — is the evidence.

Two things qualify it. First, the customer structure. High-end smartphone sensors mean concentration in a small number of very large handset makers with enormous purchasing leverage; Sony’s own commentary that “our main customer base and demand in the high-end segment remains strong” while the “volume-driven low-end smartphone market is impacted by the rising cost of memory” describes both the strength and the concentration. Second, and more fundamentally, semiconductor advantages require continuous reinvestment to defend, and Sony is now choosing to stop doing half of it in-house. The TSMC joint venture is explicitly designed to “reduce invested capital” and lower “investment in production facilities.”

I judge this the right trade, and Totoki was refreshingly honest about the terms when a sell-side analyst pushed him on margin: shifting fixed cost to variable “the margin will go down, that is right, as you have said, but the risk would go down as well. So in — since it’s a trade-off.” Sony is exchanging some moat depth for materially better capital efficiency and downside protection. But it is a trade. A designer who buys process from a merchant foundry has a narrower advantage than an integrated device manufacturer, and the same foundry is available to Samsung and OmniVision.

Verdict: a real but narrowing competitive advantage, being deliberately converted into higher returns on a smaller capital base. Financially defensible; strategically a reduction in the moat’s width.

Pictures — no moat at the studio; an emerging one at Crunchyroll

A 7.0% operating margin, an 18.8% decline in Motion Pictures revenue, and a written-off VFX acquisition are what an absence of competitive advantage looks like in the numbers. Sony Pictures has a valuable library and good franchise IP (Spider-Man, Jumanji), and the Spider-Man trailer surpassing one billion views in four days is real evidence of franchise strength. But franchise IP is an asset, not a moat — every studio has some, and the studio business remains hit-driven with no barrier to entry.

Crunchyroll is different and deserves separate treatment: 21m+ paid subscribers, 50,000+ episodes, exclusive relationships with Japanese anime publishers, and a category — anime — growing globally. That is a scale-plus-captivity position in a defined niche. Its contribution is visible in Media Networks growing 11.5% while the studio shrank.

Verdict: no durable advantage in the studio; a genuine one forming at Crunchyroll. The CEO’s own comment that “pictures ROIC is low” and that Sony “constantly carry out structural reform” in low-ROIC segments is the most direct acknowledgement available.

ET&S — advantage in cameras, none in televisions

Sony’s Alpha camera franchise and professional imaging business have real brand and technical strength, and Imaging revenue fell only 2.1% in a difficult year. Televisions have no advantage and fell 20.3%. The TCL joint venture is the appropriate answer.

The share-stability test

Greenwald’s most reliable moat test is stability of market share over time. On that test: PlayStation’s share of console has been stable-to-improving for two generations; Sony Music’s share of the majors’ oligopoly has been stable; Sony’s image-sensor leadership has been stable at the high end for over a decade; Sony’s TV share has been eroding for two decades; Sony Pictures’ share of box office fluctuates with slate. The test cleanly separates the good businesses from the bad, and Sony’s own portfolio actions are now moving in exactly the direction the test recommends.

Overall competitive verdict. Sony has two genuinely durable moats (PlayStation, Music), one real but narrowing one (image sensors), one emerging niche one (Crunchyroll), and two businesses with none (film studio, televisions) that are being restructured or exited. Weighted by profit, the moated businesses generated ¥910.3bn of the ¥1,531.1bn segment operating income in FYE Mar-26 — approximately 59% — and, on a normalised basis excluding one-time items, approximately ¥996bn of ¥1,680bn, also 59%. That is a majority, and the direction of travel is toward more.

5. Growth History and Forward Opportunities

The historical record

Consolidated sales on the continuing-operations basis grew from ¥12,034.9bn to ¥12,479.6bn, up 3.7%. Operating income excluding Financial Services grew approximately 23% in the year ended March 2025 and 13.4% in the year ended March 2026, taking the margin from roughly 9.2% to 10.6% to 11.6% across three years. That is a genuine margin expansion story, and it is the best evidence for the “portfolio is improving” claim.

Decomposing FYE Mar-26 growth by segment shows how narrow it was. Of ¥444.7bn of consolidated sales growth, I&SS contributed ¥352.5bn and Music ¥277.5bn — together more than the total, because ET&S subtracted ¥148.7bn and Pictures subtracted ¥6.7bn while G&NS was flat. Two segments carried the entire company. Similarly on profit: of ¥170.9bn of operating income growth, I&SS contributed ¥96.2bn and Music ¥89.7bn, G&NS ¥48.4bn, offset by ET&S at −¥32.3bn, Pictures at −¥12.4bn and All Other at −¥56.7bn.

Growth quality varies sharply by source:

• High quality: Network Services +13.9%, Music Publishing +10.5%, Recorded Music streaming +8.1%, Media Networks (Crunchyroll) +11.5%, Digital Software & Add-on Content +5.4%. These are recurring, high-margin, and driven by user and engagement growth rather than price or currency.
• Medium quality: I&SS +19.6%, driven by higher average selling prices, improved product mix and higher unit sales. Real, but cyclical, capital-intensive, and management has already guided it to reverse slightly next year.
• Lower quality: Recorded Music “Other” +21.0% (live events and merchandising — good businesses but more capital- and labour-intensive than streaming), Visual Media & Platform +33.1% (boosted by Demon Slayer: Kimetsu no Yaiba Infinity Castle, a genuine phenomenon but a hit, not a run-rate).
• Currency: the FX contribution was material and must be stripped. Average rates were ¥150.7/USD and ¥174.7/EUR. G&NS sales benefited by ¥87.3bn and operating income by ¥54.3bn from FX; ET&S sales by ¥7.1bn and income by ¥5.3bn. I&SS was hurt by FX — sales −¥15.0bn and income −¥12.5bn — as was Music (sales −¥16.9bn). Net across disclosed segments, FX added roughly ¥47bn to operating income, or about 27% of the ¥170.9bn of growth. Excluding it, operating income growth was closer to 9.7%.

That last calculation matters and is not in the headline. Roughly a quarter of Sony’s reported FYE Mar-26 operating income growth came from a weak yen.

Forward opportunities, ranked by conviction

1. The PlayStation annuity compounding through the hardware trough (high conviction). The mechanism is already visible in the FYE Mar-26 numbers and requires no new development. A 125m-account base with growing play time and 13.9% Network Services growth generates increasing profit regardless of console shipments, and management guides FY2026 G&NS operating income to ¥600bn — a figure that is flat versus FY2025 excluding one-times only because it absorbs “an increase in investments of the next-generation platform,” with management stating that “excluding these factors, we expect steady double-digit growth in the profit generated by our current business.”

2. Music compounding at high single digits (high conviction). Streaming growth of 8–14%, an expanding catalogue via the Pink Floyd and Queen acquisitions and the GIC joint venture, and structural pricing power. Management guides FY2026 Music operating income of ¥400bn, essentially flat excluding one-times, only because it laps the Demon Slayer contribution — an unusually clean explanation of a flat guide.

3. Crunchyroll and anime as a cross-group platform (medium-high conviction). Anime is one of the few genuinely growing global content categories, and Sony has an unusual position across it: Aniplex produces, Crunchyroll distributes, Music monetises soundtracks and merchandising, and PlayStation and Pictures adapt. Demon Slayer is the proof of concept — it appeared in the FYE Mar-26 results of the Music segment and the Pictures segment. The strategic partnership with Bandai Namco and the increase of the Peanuts Holdings stake to 80% extend the same IP-ownership logic.

4. Physical AI sensors — automotive and robotics (medium conviction, high optionality). The TSMC MOU explicitly seeks to “explore emerging new opportunities in physical AI applications, such as automotive and robotics.” If machine vision becomes a volume market on anything like the scale of smartphone cameras, Sony is the default supplier. This is genuine optionality and it is not in any near-term number. It is also the loading the factor model already detects — Sony carries a +0.31 beta to a Robotics & AI industry factor.

5. The next PlayStation generation (medium conviction, uncertain timing). A PS6 launching into a 125m-account base is a powerful catalyst, but management has explicitly not decided timing or price and is openly wrestling with a memory-inflated bill of materials, floating “new ways of selling the product” and “changing business models.” That candour is welcome and the uncertainty is real.

6. I&SS returning to growth in the next mid-range plan (medium conviction). Management expects a “renewed acceleration towards larger-sized sensors” in the next plan period and is using FY2026 to “establish the infrastructure necessary to support this growth.” This is a deferral of growth, not a cancellation, but the near-term guide is down.

What growth is not available

Sony will not grow through television, and it should not grow through the film studio. Management appears to agree on both counts. Growth from acquisitions of operating businesses should be viewed with active suspicion given the record (Section 7).

Verdict on growth: genuine but narrow, high-quality in mix, and about to pause. The two segments carrying the company are both guided flat-to-down next year. The recurring-revenue engines underneath are compounding at high single to low double digits and should continue to. An investor buying Sony today is buying a company whose growth is on hold for a year while its earnings quality improves.

6. Financial Quality

The single most important thing on the financial statements

Sony reported a net loss attributable to stockholders of ¥326.9bn for the year ended 31 March 2026. This did not happen in any economic sense, and understanding why is prerequisite to valuing the company.

On execution of the partial spin-off, IFRS required Sony to transfer ¥1,377,795 million of accumulated other comprehensive income directly related to the disposal group into the income statement as a loss. That figure comprises a loss of ¥1,640,079 million relating to fair-value changes on debt instruments measured at fair value through OCI held in the Financial Services business, partly offset by income of ¥263,298 million relating to insurance finance income. In plain terms: Sony Life owned a very large Japanese bond portfolio; rising rates had produced large unrealised losses; those losses had already been charged against Sony’s book equity through the OCI reserve; the demerger triggered recycling of that reserve through the P&L.

The arithmetic confirms it is a wash. Retained earnings fell from ¥6,678.2bn to ¥5,294.9bn, a decline of ¥1,383.3bn. Other reserves rose from −¥566.4bn to +¥1,229.4bn, an increase of ¥1,795.8bn. Equity attributable to owners was ¥8,179.7bn before and ¥8,119.0bn after — down 0.7%. No value was destroyed by the entry.

The consequence is a screening artefact of some size. Trailing reported EPS is negative, so P/E is undefined. I verified this on two independent vendors: the AZI valuation index returns pe_ratio: null and pe_percentile: null; ROIC.ai returns pe_ratio: null for the fiscal year. Any quantitative process keyed to reported earnings currently classifies Sony as loss-making. That is a mechanical, temporary and completely explicable mispricing input, and it will resolve when the loss rolls out of the trailing twelve months.

Earnings quality on the continuing business

Source: Form 20-F, fiscal year ended 31 March 2026.

Three things in that table deserve attention.

First, the tax rate. It rose 660 basis points, and the 20-F explains why: the prior year benefited from two one-time reductions — ¥48.4bn from a repayment of capital from a subsidiary and ¥35.3bn from the dissolution of a subsidiary. Roughly ¥84bn of the prior year’s net income was a tax artefact. Anyone anchoring on FYE Mar-25’s ¥1,067.4bn continuing-operations net income as the base is overstating it. 25.8% is the cleaner run-rate.

Second, financial income swung by ¥91.7bn from a net income of ¥66.6bn to a net expense of ¥25.1bn, driven principally by lower unrealised gains on Sony’s Spotify shareholding: ¥64,764m (FYE Mar-24), ¥69,019m (Mar-25), ¥9,919m (Mar-26). Sony’s non-operating income has been meaningfully flattered by a mark on a listed equity for two years. That is genuine income but it is not operating income and it is not repeatable.

Third, and encouragingly, gross margin expanded 150 basis points with cost of sales growing only 1.5% against 3.7% sales growth. That is real operating leverage and it is not a currency effect.

Normalising the operating result

This is the analytical crux of the forward case. Management guides FY2026 operating income to ¥1,600bn and frames it as growth. Against normalised FYE Mar-26 of approximately ¥1,597bn it is flat. Management effectively concedes this segment by segment: G&NS “essentially flat year-on-year” excluding one-times; Music “at the same level as the previous fiscal year” excluding one-times; I&SS “essentially flat compared to the previous fiscal year if restructuring costs are excluded.” The company is telling you, if you read the segment commentary rather than the headline, that FY2026 is a flat year.

Cash generation

Free cash flow of ¥1,487.9bn on a ¥20.3tn market capitalisation is a 7.3% free-cash-flow yield. Cash conversion is sound: operating cash flow of ¥1,945.6bn against operating income of ¥1,447.5bn is a ratio of 1.34x, and capital expenditure fell 26% to ¥457.7bn as the sensor capacity build moderated.

Two caveats. Free cash flow here is measured after property, plant and equipment only; Sony also invests heavily in content assets (film, television, music catalogue), some of which flows through operating cash flow as production spending and some of which appears in acquisitions. And management guides FY2026 operating cash flow to ¥1,500bn — 23% below the FYE Mar-26 actual, despite guiding higher operating income. That divergence was not explained on the call and is recorded as an open question in Section 13. Taking the guide at face value and assuming capital expenditure of ¥450–500bn implies FY2026 free cash flow nearer ¥1.0–1.05tn, a 5.0–5.2% yield — still respectable, materially less exciting.

Balance sheet

This is a materially different and far stronger balance sheet. Sony now holds ¥1,166.9bn of net cash. Its inaugural US-registered bond in roughly three decades priced in June 2026 at approximately +70bp (5-year) and +90bp (10-year) over Treasuries — spreads consistent with a solidly investment-grade credit, and confirmation from the debt market that the post-demerger entity is well capitalised.

The honest counterweight is intangibility. Goodwill of ¥1,673.9bn and other intangibles of ¥3,218.2bn total ¥4,892.1bn, or 31.2% of total assets. Those are music catalogues, film libraries, Crunchyroll and Bungie — real, income-producing assets, but carried at acquisition-based values and subject to judgement. Sony’s own risk factors acknowledge the exposure, and the ¥120.1bn Bungie and ¥27.1bn Pixomondo write-downs in a single year demonstrate the risk is not theoretical. Tangible book value per share is roughly ¥538, against a share price of ¥3,409 — a 6.3x price-to-tangible-book. Anyone underwriting Sony on asset value is underwriting the intangibles.

Returns on capital

Reported ROE of −4.8% is meaningless for the same reason the net loss is. On continuing operations, return on average equity attributable to owners was approximately 12.7%. The more useful measure post-demerger is ROIC: NOPAT of approximately ¥1,074bn (operating income ¥1,447.5bn at the 25.8% effective rate) against invested capital of ¥7,974.4bn (equity ¥8,513.6bn plus debt ¥1,669.7bn less cash ¥2,208.9bn) gives approximately 13.5%. That comfortably exceeds any reasonable cost of capital and represents a genuine step-up from the pre-demerger structure, where a ¥17.5tn insurance investment portfolio sat in invested capital producing no continuing-operations profit.

Verdict on financial quality: good and improving, with a headline that says the opposite. Margins are expanding, cash conversion is strong, the balance sheet is now conservative with net cash, and ROIC has stepped up to a level that creates value. The offsets are real: a third of the balance sheet is intangible and has just proved impairable, non-operating income has been flattered by a Spotify mark, roughly a quarter of last year’s profit growth was currency, the tax rate is normalising upward, and next year’s cash flow is guided sharply lower without explanation. Do economics improve with scale? On this evidence, yes — but the improvement is coming from mix and capital discipline, not from volume.

7. Capital Allocation

Capital allocation is where the analysis of Sony divides most sharply, and both halves must be stated honestly.

The structural record: excellent

The Financial Services partial spin-off is the best capital-allocation act by Sony management in a decade. The process ran roughly two and a half years — assessment announced May 2023, METI approval of the Corporate Restructuring Plan under the Act on Strengthening Industrial Competitiveness in February 2024, SFGI listed on the TSE Prime Market in September 2025 as a condition precedent, execution 1 October 2025. Sony retained 16.40% and distributed the remainder to shareholders in kind.

What it achieved: removed ¥19.6tn of assets and liabilities unrelated to the operating businesses; eliminated an interest-rate-driven OCI exposure that had been quietly eroding book value; turned net debt into ¥1,166.9bn of net cash; lifted the tangible common equity ratio from 13.2% to 33.6%; raised group ROIC by removing non-earning invested capital; and — importantly — did not sell the asset into a poor market. Shareholders retained 100% of the economics through the in-kind distribution plus the 16.40% stake. The structure was chosen to avoid a forced sale at a depressed valuation, which is precisely the discipline one wants.

The TSMC pivot and the TCL exit follow the same logic applied to operating assets: reduce capital intensity where the advantage is narrowing (sensors) and exit entirely where there is none (televisions). Both were done through partnerships rather than distressed disposals.

The operating M&A record: poor

Bungie. Sony acquired Bungie in 2022 for approximately $3.6bn. In FYE Mar-26 Sony recorded a ¥120.1bn impairment against Bungie’s intangible and other assets, impairing the full amount of Bungie-related fixed assets except goodwill, with the CFO stating that “earnings from Bungie’s title portfolio did not reach our expectations, so we downwardly revised our business plan.” Trade press values the total Bungie-related write-down at close to $800m and — critically — notes this is the second Bungie impairment. Two write-downs on one acquisition inside four years is not bad luck. Management’s forward framing is that Marathon has an 82 Metacritic score and 90%+ positive Steam reviews with solid retention, but the CFO simultaneously made clear that Bungie is not the title Sony is relying on for first-party growth — that burden falls on Housemarque’s Saros (released April 2026) and Insomniac’s Marvel’s Wolverine (September 2026).

Sony Honda Mobility. A joint venture into automotive — the most capital-intensive, lowest-return large industry in the world — has been abandoned following Honda’s reassessment of its EV strategy. Sony recorded a ¥44.9bn additional equity-method loss in the March 2026 quarter and has guided ¥30bn more into FY2026. Management states the allocation of burden among the parties is settled and SHM “will not be claiming for any more damages to Honda.” Totoki’s defence — that the team learned about software-defined vehicles and that “the people who have gone through this experience should be actively leveraged within our group” — is the standard rationalisation of a failed venture. It is not worthless, but it is not a return on capital either.

Pixomondo was acquired and wound down with ¥27.1bn of impairment and shutdown costs. Sony Semiconductor Israel was sold at a ¥19.9bn loss.

Against that, the music-catalogue acquisitions (Pink Floyd, Queen) and the Peanuts Holdings increase to 80% look sound — buying non-substitutable rights with predictable royalty streams at a point in the cycle when the asset class is well understood is a very different activity from buying a game studio, and Totoki’s answer on catalogue valuations was thoughtful and specific. The GIC joint venture is an intelligent structure: it brings sovereign-wealth capital to music-rights acquisition, allowing Sony to scale the position without consuming its own balance sheet.

The pattern is clear: this management is excellent at structural portfolio surgery and mediocre-to-poor at buying operating businesses. That distinction matters enormously for the forward case, because roughly ¥800bn of the ¥1.8tn mid-range-plan strategic-investment frame remains undeployed. Where that ¥800bn goes is, in my judgement, the largest single controllable risk to the thesis.

Returns to shareholders: improving materially

Buybacks have more than doubled in two years. The FY2026 ¥500bn facility is roughly 2.5% of the current market capitalisation, and — importantly — it is being executed into weakness: 37,076,600 shares repurchased for ¥127.48bn through 30 June 2026, including 18,006,700 shares for ¥60.22bn between 1 and 23 June, a window in which the ADR touched its 52-week low. Buying your own stock aggressively at a 30% drawdown is the correct behaviour and is meaningfully rarer than management teams claim.

The dividend was raised ¥10 to ¥35 per share for FY2026 — a 40% increase — though on a ¥3,409 share price that is still only a ~1.0% yield. Sony is, correctly, a buyback-first returner of capital.

The mid-range-plan framing supports the trajectory: three-year cumulative operating cash flow excluding Financial Services has been revised upward from an original ¥4.5tn plan to ¥4.8tn and now to ¥5.7tn, with the CFO stating that the additional capital will be allocated “primarily to higher shareholder returns,” and that Sony now has the capacity to fund both strategic investment and returns. Mid-range-plan targets of ≥10% average annual operating income growth and a ≥10% three-year cumulative operating margin are on track for 16% and 11.7% respectively.

Financing and incentives

The inaugural US-registered bond offering in June 2026 — roughly $1bn across 5-year and 10-year senior fixed-rate tranches at approximately +70bp and +90bp — is worth noting precisely because Sony does not need the money. The plausible rationale is currency-matched funding for dollar-denominated assets (music catalogues, Crunchyroll, the US studio) and re-establishing a USD credit curve for future flexibility. Benign, cheap, and mildly confirming of the balance-sheet story.

Incentives are equity-linked but the alignment is weak in absolute terms. As of 31 March 2026, Chairman Kenichiro Yoshida beneficially owned 2,510,115 shares including 682,000 stock acquisition rights at ¥1,288 (expiring October 2028) and 600,000 at ¥2,278 (expiring November 2032); President and CEO Hiroki Totoki 1,480,250; CFO Lin Tao 198,503. Both of Yoshida’s strike prices are deep in the money against ¥3,409 — management has been paid for a share price that tripled from the 2016–18 base, which is at least defensible pay-for-performance. But every named individual’s stake rounds to less than 0.1% of the ~6.0bn shares outstanding, and the outside directors hold between zero and 24,500 shares each, with three holding none. This is a Japanese-governance pattern rather than a scandal, but an investor should not mistake it for the owner-operator alignment a US filer with comparable disclosure would show. Correspondingly, the Form 4 record — frequent filings accompanied by Rule 144 notices, with no discretionary open-market purchase identified in the trailing window — carries no signal for this name. That is itself worth knowing: the insider channel, which is often the highest-signal disclosure for a US company, is simply uninformative here.

Verdict on capital allocation. Structurally excellent; operationally poor at M&A; strongly and improvingly shareholder-friendly on returns. Management has allocated capital intelligently where the decision was what to own, and badly where the decision was what to buy. The forward risk is that ~¥800bn of undeployed strategic-investment capacity meets the second skill rather than the first. The mitigating evidence is Totoki’s own framing on the call — that Sony has learned from investments where “we had varied results and went well, some well,” that it will apply “the lessons learned” and “the rational price,” and that free-cash-flow generation and shareholder return are the mission. I take that as a genuine, if unproven, commitment.

8. Changes and Headwinds — Last Two Years

Portfolio and structural changes

The headwinds, stated plainly

1. The memory-price shock. This is the dominant near-term headwind and it is entirely exogenous. AI data-centre demand has drawn memory supply away from consumer applications; prices reportedly roughly doubled in a quarter with a further sharp increase forecast. Sony is a large buyer for consoles, televisions and cameras. Management has quantified and bounded the FY2026 impact — approximately ¥30bn contained within ET&S, calendar-2026 console volumes secured with prices “to a certain extent” agreed — but expects memory to remain expensive into FY2027 and beyond, and is explicitly unable to commit to next-generation console timing or pricing as a result. Read through the capital-cycle lens, this is a rent transfer from downstream assemblers to upstream memory makers that will mean-revert when capital returns to memory supply. But the timing is outside Sony’s control.

2. The console hardware down-cycle. PS5 shipments fell to ~16m in FYE Mar-26 from 18.5m and a 20.8m peak, with the March quarter down ~46% to 1.5m — the worst since launch. Sony raised the price to $650 from $550. FY2026 G&NS sales are guided down 5.7% to ¥4,420bn. This is a real headwind that the annuity has so far more than offset, but it will not offset it forever if the installed base stops growing.

3. The “AI content glut” narrative. This is the headwind that is hardest to quantify and, in my judgement, most likely to be wrong — but it is unquestionably moving the stock. Totoki named it explicitly: with AI, “content production become easier… so people will be — we will be taking — competing over the user’s time… so there might be anxiety that the entertainment business cannot grow as before.” The market is applying a lower multiple to entertainment assets on the theory that infinite cheap content destroys the scarcity value of professionally-produced content. Section 11 addresses why I think this is largely backwards for platform owners and rights holders.

4. Image sensors guided down. Management expects the shift to larger smartphone sensors to moderate, has embedded a slight year-on-year decrease in mobile sensor sales in FY2026 guidance, and is prioritising “efficiency… including through fixed cost control and yield improvements.” The segment that delivered 36.8% profit growth is guided to roughly flat.

5. Tariffs and geopolitics. The 20-F cites additional US tariffs as an ET&S headwind. Management’s answer on the reciprocal-tariff litigation was, candidly, that uncertainty has increased. Sony’s supply chain and manufacturing footprint span Japan, China and Taiwan; its largest end market is the US; and its semiconductor business is now a subject of Japanese industrial policy and a partner of a Taiwanese foundry.

6. Currency. A weak yen added roughly ¥47bn to FYE Mar-26 operating income across disclosed segments — about a quarter of the growth. It cuts both ways; a strengthening yen would reverse it. The factor model’s −0.20 USDollar loading confirms the market understands this.

7. Leadership. Hiroki Totoki succeeded Kenichiro Yoshida as President and CEO, with Yoshida remaining Chairman. Totoki, previously CFO and COO, is the architect of the financial-discipline agenda and the demerger. This is continuity rather than disruption, and on the evidence of the last two years it is favourable.

Verdict: on balance, the last two years strengthen the thesis materially. The structural changes — demerger, fab-light pivot, TV exit, capital-return step-up — are large, permanent and value-creating. The headwinds are predominantly cyclical (memory, console cycle, sensor cycle, currency) or narrative (AI content). The genuinely thesis-weakening items are the M&A write-offs, and they are sunk. A reasonable investor should weight permanent structural improvement above cyclical cost pressure — while noting that the cyclical pressure is what determines the next twelve months of reported numbers.

9. Risk Analysis

Risk commentary. The risk profile changed materially and favourably with the demerger. Pre-October 2025, Sony carried an insurance balance sheet whose interest-rate exposure was capable of moving book value by trillions of yen — the ¥1.64tn of unrealised bond losses recycled at spin-off is the measure of it. That risk is gone. What remains is a diversified operating company with net cash, whose principal risks are cyclical (1, 3, 8, 11), narrative (2), and self-inflicted (4, 5).

Risk 4 is the one I weight most heavily because it is the only one management fully controls and the track record is poor. Risk 2 is the one the market weights most heavily and is, in my judgement, mis-weighted. Risk 13 deserves more attention than it usually gets: Sony’s sensor manufacturing is concentrated in Kumamoto and Nagasaki, and Japan’s seismic risk is real — the Kumamoto region experienced major earthquakes in 2016.

The probability of permanent capital loss is genuinely low. A company with net cash, ¥1.5tn of annual free cash flow, five separate profit streams and a 13.5% ROIC does not go to zero absent fraud or catastrophe.

10. Valuation Discussion

This section discusses embedded expectations and scenarios only. It contains no price target and no recommendation.

Where the shares trade

At ¥3,409 (24 July 2026) and approximately 5.97bn shares outstanding after buybacks through 30 June, market capitalisation is approximately ¥20.3tn (~$125bn at ¥162.5/USD). Deducting net cash of ¥1,166.9bn gives an enterprise value of approximately ¥19.1tn (~$118bn).

Not credited in the enterprise value above: the retained 16.40% of SFGI, now a separately listed TSE Prime company, and Sony’s Spotify shareholding, which produced unrealised gains of ¥64.8bn, ¥69.0bn and ¥9.9bn over the last three years. Neither is separately valued in the 20-F, so neither is quantified here — but both bias the stated multiples upward. The operating business is modestly cheaper than the table shows.

The own-history percentile read, and why it must be handled carefully

The AZI valuation index places Sony’s price-to-book of 2.465 at the 86.4th percentile of its own approximately ten-year range, price-to-sales of 1.524 at the 58.7th percentile, and a composite at the 72.6th percentile. The P/E percentile is null because trailing EPS is negative.

Taken at face value this says Sony is expensive against its own history. I do not think it can be taken at face value, and the reason is specific. Book equity is essentially unchanged at ¥8,119.0bn, but its composition changed completely: the same book value now stands behind entertainment IP, music catalogues, image sensors and net cash rather than behind a ¥17.5tn fixed-income portfolio carried at fair value with its unrealised losses parked in OCI. A 2.5x multiple on a 13.5%-ROIC, net-cash operating company is a different fact from a 2.5x multiple on a leveraged conglomerate with an insurance balance sheet. The own-history comparison spans a structural break in what is being compared.

The price-to-sales percentile at 58.7 — mid-range — is the more usable of the two, and it is unaffected by the balance-sheet change. On that measure Sony is unremarkable versus its own history, not expensive. Per the standing discipline, these percentiles are own-history context only and are never used cross-sectionally or as a price reference.

Embedded expectations: what must the market believe?

At approximately 11.9x forward EV/EBIT, the market is underwriting a company whose operating profit does not grow in real terms from here. That is the honest translation, and it is a defensible position given that normalised FYE Mar-26 operating income of ~¥1,597bn against FY2026 guidance of ¥1,600bn is flat.

Working backwards, the current price is consistent with a set of beliefs that are individually reasonable:

• Memory costs are a persistent, not transitory, margin tax.
• PS5 hardware decline is structural and the next console cycle is uncertain in timing, cost and consumer reception.
• Image sensors have peaked for this cycle and the physical-AI opportunity is speculative.
• Music’s growth will decelerate as AI content expands supply.
• Management’s M&A record justifies a discount on the ~¥800bn of undeployed investment capacity.
• Sony is, in factor terms, a Japan-beta expression rather than a global entertainment compounder — and should be valued accordingly.

What the market is pricing correctly. All six of the above, in my view, contain truth. FY2026 is a flat year. Memory is expensive into FY2027 on management’s own statement. PS5 units are falling sharply. Sensors are guided down. Bungie was a serious error, twice.

What the market appears to be pricing incorrectly. Three things.

First, and most mechanically: trailing reported EPS is negative because of a non-cash OCI recycling, removing Sony from every earnings-based screen at the exact moment its balance sheet became clean. This is not a subtle mispricing argument; it is an arithmetic artefact with a known expiry date.

Second: the market is applying a hardware-cyclical multiple to a business whose profit engine has already moved. G&NS set a record operating profit in a year when console shipments fell 46% in the final quarter, because Network Services grew 13.9% and Digital Software grew 5.4%. Music grew 15.1% with a 19.4% clean margin. The annuity is carrying the cycle now, in reported numbers, not in projection.

Third: the TSMC joint venture structurally lowers the capital intensity of the segment the market most dislikes, and none of it is in FY2026 numbers. Sony states it will “improve the cash flow… reduce invested capital, and improve profitability” of I&SS. If it does, the ROIC of the group’s most capital-hungry business rises permanently.

Scenario analysis

All scenarios are stated on FY2027–28 operating income at a normalised 25.8% tax rate. These are explicitly assumption-driven illustrations, not forecasts, and no scenario constitutes a price target.

Two adjustments apply across all three. The buyback shrinks the share count by roughly 2.5% per year at the current facility, adding to per-share outcomes. And the retained 16.40% SFGI stake and Spotify holding are excluded entirely — both are positive, unquantified, and asymmetric to the upside.

The distribution is favourably skewed: approximately −28% in a bear case that assumes essentially everything goes wrong, against +15% in a base case and +54% in a bull case, with a net-cash balance sheet and a 5–7% free-cash-flow yield being paid to wait.

Sum-of-the-parts sanity check

A formal sum-of-the-parts is not warranted — Sony does not disclose segment invested capital or segment ROIC, so any allocation would be a construction rather than a measurement. But a directional check is useful. Music alone, at ¥447.0bn of operating income (¥412.3bn normalised) and a 19–21% margin with rights-based durability, would command a premium multiple as a standalone; applying a mid-to-high-teens EV/EBIT to it accounts for a very substantial fraction of Sony’s entire ¥19.1tn enterprise value. G&NS at a normalised ¥583bn is the largest profit pool. On any reasonable segment-multiple decomposition, the stub value implied for Pictures, ET&S and the corporate centre is low to negative — which is a familiar conglomerate-discount pattern and is precisely what the demerger and the TCL and TSMC transactions are designed to attack.

Notably, when a sell-side analyst asked about Pictures’ low ROIC on the May call, Totoki did not defend it. He acknowledged that where Sony’s returns are “inferior… we understand we have to take measures. For example, structural reform,” and noted Sony is “constantly carry out structural reform” in different segments without announcing it. That is worth remembering.

Valuation verdict. The shares embed low expectations relative to the asset quality and the balance sheet, but not implausibly low given a genuinely flat guided year and a real cyclical cost shock. The clearest analytical statement available is this: at ~11.9x forward EV/EBIT with net cash, a 5–7% free-cash-flow yield, a ~2.5% buyback and an unvalued stake in a listed financial company, the market is paying nothing for the post-demerger ROIC step-up, nothing for the fab-light capital-intensity reduction, and nothing for a PS6 cycle. Whether that constitutes an opportunity depends on whether the memory cycle mean-reverts and whether the AI-content thesis is right — which is the subject of Section 11.

11. Variant Perception

The consensus belief

The consensus, insofar as the 30.5% drawdown from November 2025 expresses one, is roughly this: Sony is a Japanese conglomerate whose two profit engines — consoles and image sensors — are both cyclically rolling over into a memory-cost squeeze, whose entertainment assets face structural devaluation from AI-generated content abundance, whose reported earnings are negative, and whose management has just written off a $3.6bn game studio for the second time. The demerger was a nice piece of financial engineering that is now in the price.

The factor evidence corroborates that this is genuinely how the market treats the name. Sony’s largest non-market factor loadings are Country: Japan +0.600 and Robotics & AI +0.310, with Quality −0.10 and Momentum −0.06. Its twenty closest factor-space neighbours are almost entirely Japan and international-developed ETFs — IPAC (0.909), JPXN (0.898), DXJ (0.879), EWJ (0.875), EWJV (0.865) — with Nomura Holdings ADR (0.909) the only individual common stock among them. Empirically, Sony’s daily return is better explained by “Japan” than by “entertainment” or “semiconductors.”

That last finding is, I think, the single most important piece of positioning evidence in this article, and it cuts both ways. It is a de-rating diagnosis: for a company arguing that 67% of its sales are entertainment and IP, being priced as Japan-beta is exactly what a company that has failed to convince the market of its transformation looks like. It is also a warning to the buyer: on a mark-to-market basis, an investor buying Sony for PlayStation and Sony Music is substantially buying Japan and yen exposure, whatever the fundamental thesis says.

The strongest bear case

The strongest bear case is not the memory cycle — that is cyclical and will mean-revert. It is the AI-content thesis applied to the whole entertainment portfolio, and it goes like this. Generative AI collapses the cost of producing games, music and video toward zero. Content supply explodes. Attention is fixed. Therefore the economic rent that has historically accrued to professionally-produced content — the rent that supports Sony Music’s 21% margins, PlayStation’s first-party premium and Crunchyroll’s library value — gets competed away. In this world Sony owns a large, depreciating library of assets whose scarcity value is evaporating, and the correct multiple for it is lower, permanently.

Sony’s own CEO articulated this as one of the two market concerns. It is a coherent thesis and it should not be dismissed.

The bear case’s second leg is the capital-allocation record. A management team that has impaired Bungie twice, written off Pixomondo, sold Sony Semiconductor Israel at a loss and abandoned an automotive joint venture is not a team that deserves the benefit of the doubt on ¥800bn of remaining strategic-investment capacity. And the third leg is simply the arithmetic: FY2026 operating income is guided flat on a clean basis, and paying 11.9x for a flat year in a cyclical business with a −0.10 quality loading is not obviously a bargain.

The strongest bull case

The bull case has four legs, and the first is not a forecast.

One: the reported loss is fictitious and the screens will fix themselves. ¥1,377.8bn of accumulated OCI was recycled through the income statement on a demerger. Equity was unchanged. Cash was unaffected. Trailing EPS is negative for accounting reasons alone, and every P/E-driven screen currently excludes or mis-classifies the company. This is an arithmetic fact with a known resolution date.

Two: the AI-content thesis is backwards for platform owners and rights holders, and Sony sits on the right side of both. If content creation becomes cheap and abundant, the scarce resources become distribution, curation and non-substitutable rights. Sony owns all three. Nishino’s articulation on the platform side is exactly right: “AI is lowering the barriers to creation… As a result, we expect to see a meaningful increase in the volume and diversity of the content available to the players. Our platform’s role will be critical in ensuring players find the right content in an increasingly crowded landscape.” A store that takes a percentage of every transaction benefits from more transactions, provided it keeps the audience. On the rights side, Totoki’s answer was equally specific and, in my judgement, correct: evergreen catalogue values are not falling; AI-composed music is a very low share of the chart; and “the evergreen catalogs are based on individual experience… listened to for a long period of time, and these listeners go to live music performances and that is something which AI cannot offer by itself.” The evidence supports him — Recorded Music “Other” (which includes live and merchandising) grew 21.0%, the fastest line in the segment. If AI commoditises recorded content, the value migrates to the things that cannot be synthesised: live performance, fandom, and ownership of the originals. Sony owns the originals.

Three: the annuity is already carrying the cycle, in reported numbers. This is the strongest single fact in the report. In a year when March-quarter PS5 shipments fell 46% and hardware revenue fell 12.1%, G&NS set a record operating income. Network Services grew 13.9%. That is not a projection; it happened. A business that sets profit records during a 46% collapse in its most visible unit metric is not a hardware cyclical.

Four: the capital intensity of the worst-loved segment is about to fall. The TSMC joint venture, if it converts from MOU to definitive agreement, moves Sony from an integrated device manufacturer to a fab-light designer with a controlling interest in a shared production vehicle. Sony states it will reduce invested capital and improve profitability and “increase the flexibility of our capital allocation across the Sony Group.” A permanent reduction in the capital intensity of a ¥2.15tn-revenue segment is a permanent increase in group ROIC, and none of it is in FY2026 numbers.

The three to five assumptions that actually matter

1. Does PlayStation’s Network Services and digital-content revenue keep compounding at high single digits or better through the hardware trough? Everything in G&NS depends on this. It grew 13.9% and 5.4% respectively in FYE Mar-26.
2. Does Music streaming growth hold at mid-to-high single digits or better? Sony guides the market to mid-to-high single digits; it delivered 9% and 14% on a dollar basis. This is the difference between an excellent asset and a merely good one.
3. Does the memory cycle mean-revert by FY2028? Management assumes elevated prices through FY2027. If it extends materially beyond, the next console cycle is genuinely compromised.
4. Does the TSMC JV convert to a definitive agreement with quantified capital relief? It is currently a non-binding MOU with undisclosed terms.
5. Where does the remaining ~¥800bn of strategic-investment capacity go? Music catalogue and anime IP would confirm the thesis. Another large studio or platform acquisition would seriously damage it.

What would falsify each side

Falsifying the bull case: Network Services growth decelerating below mid-single digits, or Music streaming growth falling below mid-single digits, in either of the next two quarterly prints. Either would break the “annuity carries the cycle” argument, which is the load-bearing element. A second falsifier: the TSMC MOU lapsing without a definitive agreement, which would leave I&SS as a capital-hungry IDM in a decelerating cycle. A third: deployment of a large slice of the strategic-investment frame into another operating acquisition of Bungie-like scale.

Falsifying the bear case: the August 2026 and November 2026 prints showing G&NS and Music tracking to or above guidance with the annuity lines still compounding, alongside memory-cost impacts contained within the guided ~¥30bn. Add a definitive TSMC agreement with disclosed capital-expenditure reduction and evidence that music catalogue transaction prices have not declined, and the “structural devaluation of entertainment” thesis loses its empirical footing.

My variant perception, stated plainly

The market is treating a change in accounting presentation as a change in business quality, and a cyclical cost shock as a structural one. Sony is a better business than it was two years ago on every measure that matters — mix, margin, balance sheet, return on invested capital, capital intensity, and simplicity of disclosure — and it is priced roughly 30% lower than it was in November 2025 with a reported net loss that did not economically occur.

The honest qualification is that the market is not being irrational, merely narrow. FY2026 really is a flat year. The memory tax is real and management cannot say when it ends. And the factor evidence says the market has not yet accepted the transformation narrative at all — it still trades Sony as Japan-beta with an AI kicker and a negative quality loading. A re-rating requires the market to change its mind, and there is no date on which it must.

12. Fact vs. Interpretation

13. Open Questions

1. Why is FY2026 operating cash flow guided to ¥1,500bn — 23% below the ¥1,945.6bn achieved — while operating income is guided higher? Working capital, content spend and tax are the candidates. Management did not address it. This is the most important unexplained number in the guidance and materially affects the forward free-cash-flow yield.
2. What is the market value of the retained 16.40% of SFGI? SFGI is listed on the TSE Prime Market, so this is knowable, but it is not disclosed in the 20-F and is excluded from every multiple in Section 10. It is a real asset that makes the operating business cheaper than the headline figures suggest.
3. What is the carrying and market value of the Spotify shareholding? The 20-F discloses only the annual revaluation gain (¥64.8bn, ¥69.0bn, ¥9.9bn over three years), not the position size.
4. What are the economics of the TSMC joint venture? Ownership split beyond “Sony majority and controlling,” capital commitment, transfer pricing, the size and conditionality of Japanese government support (a ¥80bn METI figure was raised by an analyst but not confirmed by management), and the timeline to a definitive agreement are all undisclosed.
5. What are the economics of the TCL joint venture? The definitive agreement covers “the enterprise value of the business in question and the consideration to be paid for the transfer” but neither figure is public. Roughly ¥20bn of FY2026 implementation cost is disclosed; the proceeds and the ongoing earnings treatment are not.
6. Are there genuine CMOS sensor yield problems? Trade commentary suggests yield issues creating uncertainty for a key customer. Neither the 20-F nor the earnings call corroborates this. If true it is material to the I&SS guide; if not it is noise.
7. What is executive compensation in quantum? The 20-F discloses share ownership but not the aggregate or individual cash-plus-equity pay figures a US proxy would provide. Japanese-language disclosure would be required.
8. Which segments earn what ROIC? Totoki stated on the call that segment ROIC “has been disclosed” and invited analysts to refer to it, but it does not appear in the 20-F. Given that he simultaneously conceded Pictures’ ROIC is low, this is the single most useful missing disclosure for judging where the next portfolio action falls.
9. When does the next PlayStation launch, at what price, and on what business model? Management explicitly has not decided, and is openly considering “new ways of selling the product” and “changing business models” because of memory costs.
10. How much of the ¥1.8tn strategic-investment frame remains, and what is it earmarked for? The CFO said “a little bit over ¥1 trillion” has been deployed. The remaining ~¥800bn is the largest controllable swing factor in the thesis.
11. Does the AZI adjusted price series account for the 1 October 2025 SFGI dividend-in-kind? No discontinuity is visible. Total-return comparisons spanning that date carry an unquantified break.
12. What is Sony’s actual exposure to its largest image-sensor customer? The 20-F does not quantify customer concentration for I&SS, though the segment commentary is organised around “our major customer.”

14. What Must Be True

For the bull case to work

For the bear case to work

15. Source Appendix

The full source appendix — with URLs, access dates and document sections — appears as Appendix B to this article. Primary sources relied upon are: Sony Group Corporation’s Form 20-F for the fiscal year ended 31 March 2026 (filed 18 June 2026); the Forms 6-K filed 3, 14 and 17 July 2026 and 26 June 2026; the F-3ASR, 424B5, 424B2 and FWP filed 18–25 June 2026; the Sony FY2025 Corporate Strategy and Earnings Announcement transcript of 8 May 2026; the Sony Semiconductor Solutions / TSMC joint press release of 8 May 2026; the AZI daily price series; the FactorsToday factor model; and ROIC.ai fundamental and transcript data used as a cross-check. Publicly available filings and disclosures of Sony’s peers, customers and counterparties — notably Spotify, Netflix, Disney, Warner Bros. Discovery, Electronic Arts, Take-Two, Microsoft, Apple and Micron — were used for peer and value-chain context.

Sections 1–15 contain no investment recommendation and no price target; the sole exception is the clearly labelled Claude's Take block at the head of this article, which is the author’s own subjective view. Nothing here is investment advice.

APPENDIX A — Standard Diligence Questionnaire

Sony Group Corporation (NYSE: SONY / TSE: 6758) — 26 July 2026

Supplemental to the analysis above. Answers are labelled Fact / Interpretation / Assumption / Open Question where the distinction matters. Where a question does not map cleanly to Sony’s business model, that is stated and the correct analogue is given.

General

What thoughtful questions have other investors asked about this company?

The questions asked on the 8 May 2026 call by BofA, JPMorgan, SMBC Nikko and Mizuho are a good proxy for what informed investors actually care about, and they were notably sharper than usual:

• BofA (Hirakawa) asked whether Sony would monetise by partnering with an AI player, explicitly referencing Disney’s approach. Totoki’s answer was strategically interesting: Sony is deliberately not picking a single AI partner because “if we specify a specific player to work with, it might be appealing to a certain extent. But on the other hand, this might confine our action.” [Interpretation] That is optionality preservation, and it is the right instinct for a rights owner negotiating with counterparties whose relative power is still being determined.
• JPMorgan (Ayada) asked the two best questions on the call: what is the long-term upside in gaming when MAU is struggling to grow, and does AI-generated music impair catalogue valuations. Both go directly to the load-bearing assumptions above.
• SMBC Nikko (Katsura) asked about capital allocation directly and, unusually bluntly, noted “your stock price is going down drastically” and that entertainment multiples are compressing industry-wide. He also asked about the TSMC JV’s national-security dimension and about I&SS profitability not having exceeded past peaks.
• Mizuho (Nakane) asked about R&D optimisation and, pointedly, about low-ROIC segments — naming Pictures. This produced Totoki’s most revealing answer of the call: an acknowledgement that where returns are “inferior… we understand we have to take measures. For example, structural reform,” and that Sony is “constantly carry out structural reform” without announcing it.
• Nikkei asked Totoki directly why the stock was falling. He answered with the two-part explanation quoted throughout this article: memory shortage and AI-driven content abundance.

[Interpretation] The recurring theme across all of these is that professional investors are not confused about Sony’s assets — they are unconvinced that the assets will compound. That is a different, and more tractable, problem than a business-quality problem.

Cyclicality and the Nature of Earnings

Are earnings at a cyclical high or low?

[Interpretation] Neither cleanly — the segments are out of phase, which is precisely why the conglomerate is hard to value. Image sensors are at a cyclical high (record ¥357.3bn operating income, +36.8%) and management has already guided the segment down for FY2026. Console hardware is at a cyclical low (March-quarter shipments −46% to ~1.5m, worst since launch) while console profits are at a record (¥463.3bn). Music is not cyclical in the usual sense and is on a secular growth path. Televisions are in secular decline. Pictures depends on slate timing.

Group operating income of ¥1,447.5bn is a reported record, but normalised for one-time items it was approximately ¥1,597bn [Interpretation], and management guides FY2026 to ¥1,600bn — flat. The honest answer is that group earnings are at a plateau, not a peak or a trough, with the mix improving underneath.

Driven by the external environment or internal actions?

[Fact] Both are separately identifiable, which is unusually convenient. External: currency contributed roughly ¥47bn of the ¥170.9bn of operating income growth [Interpretation] (G&NS +¥54.3bn, ET&S +¥5.3bn, I&SS −¥12.5bn); memory prices are entirely exogenous; smartphone demand drives sensors. Internal: gross margin expanded 150 basis points with cost of sales up only 1.5% against 3.7% sales growth; ET&S delivered “reductions in operating expenses”; I&SS took restructuring actions in the March quarter whose benefit is reflected in FY2026 guidance; and the demerger, TCL and TSMC transactions are entirely management-driven.

How stable are revenues?

[Interpretation] More stable than the reported figures suggest, and stabilising further. Approximately 30% of consolidated sales are genuinely recurring — PlayStation Network Services (¥763.1bn), Recorded Music streaming (¥852.7bn), Music Publishing (¥419.9bn), Crunchyroll subscriptions, ET&S Network Services (¥188.3bn), plus a conservative half of Digital Software & Add-on Content — with a further large tranche of repeatable catalogue and licensing income. The volatile components are console hardware (¥1,391.6bn, −12.1%), televisions (¥476.3bn, −20.3%) and theatrical film (¥495.7bn, −18.8%). All three volatile components are shrinking as a share of the whole, two of them deliberately.

Outlook for products and services?

[Fact] Management’s own FY2026 segment guidance: G&NS sales ¥4,420bn / operating income ¥600bn; Music ¥2,140bn / ¥400bn; Pictures ¥1,630bn / ¥145bn; ET&S ¥2,250bn / ¥150bn; I&SS ¥2,070bn / ¥400bn; consolidated ¥12.3tn / ¥1,600bn / ¥1,160bn net income / ¥1,500bn operating cash flow. [Interpretation] Read against normalised FY2025, this is a flat year in which growing annuity revenue offsets declining hardware, memory costs and joint-venture implementation expense.

How big will this market be — growing, shrinking, domestic or international?

[Fact] Image sensors: ~$25.6bn in 2025, forecast ~$27.4bn in 2026 and ~$39.9bn by 2031 — growing. Recorded music: management guides the market to mid-to-high single-digit growth — growing. Console gaming: hardware units declining, software and services growing. Televisions: declining in value terms with price deflation. Anime: growing globally, with Crunchyroll at 21m+ paid subscribers.

[Fact] Sony is overwhelmingly international. It reports in yen from a Japanese base but its end markets are global, its largest single market is the United States, and average FX rates in FYE Mar-26 were ¥150.7/USD and ¥174.7/EUR. Sony Pictures and Sony Music are substantially US-domiciled operations.

Business Quality and Competitive Moat

Is the industry getting more or less competitive?

[Interpretation] It depends entirely on which industry, and the divergence is the whole story.

• Recorded music: stable-to-less competitive. Three majors plus Merlin control ~72% of streams; catalogue consolidation continues; Sony bought Pink Floyd and Queen and is scaling further with sovereign-wealth capital.
• Console platforms: less competitive on the traditional axis — Microsoft has effectively abandoned exclusivity as a strategy — but more competitive for user attention against mobile, PC and, prospectively, an AI-enabled flood of independent content.
• Image sensors: stable and concentrated (top five ~83%), but Sony’s own move to a merchant foundry means the process advantage becomes more purchasable by competitors over time [Interpretation].
• Televisions: far more competitive, dominated by vertically integrated Chinese and Korean panel makers. Sony is exiting.
• Film and television production: more competitive, with distributors increasingly producing their own content — a risk Sony’s own 20-F names explicitly.

How profitable is the business (ROIC, ROE)?

[Fact] Reported return on equity is −4.8%, which is meaningless because the denominator’s numerator was hit by the non-cash spin-off entry. [Interpretation] On continuing operations, return on average equity attributable to owners was approximately 12.7%. Return on invested capital — NOPAT of ~¥1,074bn (operating income ¥1,447.5bn at the 25.8% effective tax rate) against invested capital of ¥7,974.4bn (equity ¥8,513.6bn + debt ¥1,669.7bn − cash ¥2,208.9bn) — is approximately 13.5%, a genuine step-up from the pre-demerger structure where a ¥17.5tn insurance portfolio sat in invested capital earning nothing attributable to continuing operations.

[Open Question] Segment ROIC. Totoki stated on the call that “as for ROIC, for each segment, we have the ROIC numbers. So this has been disclosed,” but it does not appear in the 20-F. Given that he simultaneously conceded Pictures’ ROIC is low, this is the most useful missing disclosure in the entire filing.

How profitable is the industry — how many competitors, what barriers to entry?

• Music: extremely profitable for rights owners (Sony 21.1% reported operating margin), structurally unprofitable for distributors (~68% of Spotify revenue flows to rights holders). Barriers: absolute for existing catalogue.
• Consoles: profitable for the three platform holders at the store and subscription layer, unprofitable at the hardware layer. Barriers: installed base, developer relationships, first-party studios — very high.
• Image sensors: Sony 16.6% segment margin (18.3% ex-restructuring [Interpretation]). Barriers: process know-how and capital — high but purchasable.
• Film: Sony 7.0% margin. Barriers: effectively none beyond capital.
• Televisions: thin to negative for non-integrated players. Barriers: none.

Can the business be easily understood?

[Interpretation] Now, yes — and that is new. Before 1 October 2025 Sony required an analyst to model an entertainment company and a Japanese life insurer simultaneously, with a ¥35.3tn balance sheet dominated by insurance assets. Post-demerger the balance sheet is ¥15.7tn, the segments are five clean operating businesses, and the disclosure by product category is genuinely good. The one remaining obstacle is the discontinued-operations accounting, which makes the current-year income statement unreadable at the headline level and requires the reader to work with continuing operations only.

Can it be undermined by foreign low-cost labour?

[Interpretation] Partially, and it already has been. Televisions and commodity consumer electronics were undermined by exactly this and are being exited. Image sensors are not primarily a labour-cost business — they are a capital and process business — but they are exposed to state-subsidised competition, which is a related threat and is why Sony’s own JV is premised in part on Japanese government support. Music, PlayStation platform economics, and Crunchyroll are effectively immune: the assets are copyrights and network effects, not manufacturing cost.

Do brands matter?

[Fact/Interpretation] Yes, and unusually so, but the brand that matters most is not “Sony.” PlayStation is one of the strongest consumer brands in the world and drives the platform’s demand-side advantage. BRAVIA and Alpha carry genuine premium in televisions and cameras respectively — though the television brand has not been sufficient to defend the economics. Crunchyroll is the default brand in Western anime distribution. In music, the brands that matter are the artists, not the label; Sony’s advantage is contractual ownership, not consumer-facing brand.

What is the nature of competition?

[Interpretation] In music, competition is for artist signings and catalogue acquisitions — bidding competition against two rivals with similar economics, which is why catalogue prices are the key variable. In gaming, competition is for exclusive content and for user attention, increasingly against non-console alternatives. In image sensors, competition is on specification, yield, price and design-win cycles at a handful of very large handset customers. In film, competition is for talent, slate and shelf space against better-capitalised streamers. In televisions, competition is pure price against vertically integrated manufacturers.

Customers’ switching costs?

[Fact/Interpretation] PlayStation: high and real — purchased digital entitlements, save data, trophies, friend graph and PS Plus state do not transfer. This is the group’s most valuable source of customer captivity. Music: not applicable in the consumer sense; the switching cost sits with the distributor, who cannot substitute away from the catalogue. Crunchyroll: moderate — library exclusivity and simulcast timing. Image sensors: high at the design-win level — a sensor is qualified into a specific handset with tuned image-processing pipelines, so switching mid-cycle is impractical, but the next cycle is fully contestable. Televisions and cameras: low; cameras have moderate switching costs through lens-mount lock-in, which is a genuine and under-appreciated advantage in the Alpha franchise.

Financial Condition and Balance Sheet

Assets not fully recognised on the balance sheet?

[Interpretation] Yes, and they are substantial. The most obvious is the owned music catalogue, which is carried at historic acquisition cost less amortisation rather than at the value of its royalty stream. Evergreen catalogues generate cash indefinitely and, on Totoki’s own account, are not declining in transaction value. Similarly, the Sony Pictures film and television library is carried at amortised cost. The PlayStation user base — 125 million monthly active accounts — is not on the balance sheet at all, and neither are the first-party studio franchises other than through goodwill. [Open Question] The retained 16.40% of SFGI is on the balance sheet as an equity-method investment at carrying value, not at the market value of a now-listed TSE Prime company. The Spotify shareholding is marked to fair value through profit or loss, so it is recognised, but its size is not disclosed.

Off-balance-sheet liabilities?

[Fact] Sony discloses lease liabilities separately from FYE Mar-26 (¥627.7bn of total capital leases), so leases are on balance sheet under IFRS 16. Pension liabilities of ¥165.0bn are recognised. [Interpretation] The genuine off-balance-sheet commitments in a business like Sony’s are content and talent commitments — film production obligations, artist advances, sports and licensing commitments — which are disclosed in the notes as commitments rather than as liabilities. [Open Question] These were not separately quantified for this report and merit a dedicated pass before any position of size.

How conservative is the accounting?

[Interpretation] Mixed, with one genuinely conservative act and one structural soft spot.

Conservative: Sony impaired the full amount of Bungie’s fixed assets except goodwill rather than taking a partial write-down, wrote off Pixomondo entirely, and took the sensor display-device impairment and the Sony Semiconductor Israel disposal loss in the same year. Taking ¥149.9bn of gross charges in a single year rather than spreading them is the behaviour of a management team clearing the decks, not managing earnings. The spin-off accounting itself is the opposite of flattering — Sony took a ¥1.38tn reported loss it could not avoid and did not attempt to obscure.

Soft spot: intangibles. Goodwill of ¥1,673.9bn and other intangibles of ¥3,218.2bn total ¥4,892.1bn, or 31.2% of total assets, against tangible book value per share of roughly ¥538 versus a ¥3,409 share price. Content and catalogue amortisation schedules involve significant judgement, and the year’s two impairments demonstrate the risk is live. Sony’s own risk factors flag it.

One caution for the reader: Sony’s “operating income” includes the share of profit or loss of equity-method investments (a ¥64.2bn loss in FYE Mar-26). This is a broader definition than most US filers use, and it is why third-party aggregators report a different EBIT (ROIC.ai computes ¥1,553.7bn against Sony’s ¥1,447.5bn). The company definition is used throughout this article.

How CapEx-hungry is the business?

[Fact] Purchases of property, plant and equipment were ¥605.8bn, ¥621.0bn and ¥457.7bn over the last three fiscal years — 3.7% of sales in FYE Mar-26, down from 5.2%. Of that, ¥227.4bn and ¥246.7bn in the last two years was specifically for image-sensor capacity. [Interpretation] The group therefore has a bimodal capital profile: I&SS is genuinely capital-hungry and consumes roughly half of group capital expenditure on 16% of sales, while Music, Pictures and G&NS are capital-light in fixed-asset terms (their “capital expenditure” is content and catalogue investment, which flows through operating cash flow and acquisitions instead). The mid-range plan set a ¥1.7tn three-year capital-expenditure target.

[Interpretation] The TSMC joint venture is a direct attack on this profile. Sony states it will “improve the cash flow… reduce invested capital, and improve profitability of the I&SS segment by lowering investment in production facilities and mitigating equipment procurement costs,” and “increase the flexibility of our capital allocation across the Sony Group.” If it converts, group capital intensity falls structurally.

Capital Allocation and Management

How much FCF does the business generate, how does management use it, what is the philosophy?

[Fact] Free cash flow was ¥767.4bn, ¥1,700.7bn and ¥1,487.9bn over three fiscal years — ¥245.02 per share in FYE Mar-26, a 7.3% yield on the current market capitalisation. Uses in FYE Mar-26: ¥522.1bn of buybacks, ¥135.0bn of dividends, ¥185.4bn on acquisitions of subsidiaries, ¥457.7bn of capital expenditure.

[Fact] The stated philosophy, from the CFO: three-year cumulative operating cash flow has been revised from ¥4.8tn to ¥5.7tn and “we plan to allocate the additional capital primarily to higher shareholder returns.” From the CEO: “if we can have free cash flow that generated, that is our mission to generate more free cash flow, having invested in the strategic areas and then to return to our shareholders.”

[Interpretation] The philosophy is credible because it is being executed. Buybacks tripled in three years and the FY2026 ¥500bn facility is running ahead of schedule and into the drawdown — ¥127.5bn deployed by 30 June, including ¥60.2bn in the three weeks when the stock touched its 52-week low. [Fact] Guided FY2026 operating cash flow of ¥1,500bn is 23% below the FYE Mar-26 actual, which management did not explain — [Open Question] and which materially lowers the forward free-cash-flow yield to approximately 5.1%.

Significant acquisitions recently?

[Fact] Last two years: additional equity in Peanuts Holdings taking Sony to 80% (acquiring WildBrain’s stake, producing a ¥34.7bn remeasurement gain); the Pink Floyd and Queen music catalogues; a music-rights acquisition joint venture with GIC; a strategic partnership with Bandai Namco Holdings; the TCL home-entertainment joint venture (definitive agreement March 2026, operational April 2027); and the TSMC image-sensor MOU. Divestitures and closures: Pixomondo wound down (¥27.1bn), Sony Semiconductor Israel equity interest sold (¥19.9bn loss), Sony Honda Mobility’s Afeela EV discontinued (¥44.9bn equity-method loss, ¥30bn more guided), and the Financial Services partial spin-off. Cash paid for acquisitions of subsidiaries was ¥283.4bn, ¥199.3bn, ¥294.4bn and ¥185.4bn over four years.

[Interpretation] The record separates cleanly into two categories. Structural portfolio actions — the demerger, TCL, TSMC — are excellent. Rights and IP acquisitions — music catalogue, Peanuts, the GIC structure — look sound, because buying non-substitutable royalty streams at a known cost of capital is a fundamentally different activity from buying an operating business. Operating-company acquisitions are poor: Bungie has been impaired twice against a ~$3.6bn purchase price, Pixomondo was bought and shut, Sony Semiconductor Israel was sold at a loss, and Sony Honda Mobility was abandoned. Roughly ¥800bn of the ¥1.8tn strategic-investment frame remains undeployed, and which of these two skills it meets is the single largest controllable variable in the investment case.

Buying back shares?

[Fact] Aggressively and increasingly. ¥203.0bn → ¥285.5bn → ¥522.1bn over three years, with a ¥500bn facility authorised for FY2026 (~2.5% of market capitalisation) of which ¥127.48bn / 37,076,600 shares was executed by 30 June 2026. Shares outstanding fell from 6,206.1m to 6,122.5m to 6,003.3m — a 3.3% three-year reduction, accelerating to 1.9% in the most recent year. Sony also cancelled ¥339.2bn of treasury stock in one of the periods presented, which is the correct follow-through on a repurchase.

Issuing large amounts of new shares to insiders?

[Fact] No. Sony introduced an RSU-based stock compensation plan in the fiscal year ended March 2023, and settles vesting through disposal of treasury shares rather than new issuance — the 6-K of 17 July 2026 documents exactly this mechanism. Legacy stock acquisition rights exist (Yoshida holds 682,000 at ¥1,288 and 600,000 at ¥2,278). [Interpretation] Dilution is immaterial: the diluted share count exceeds basic by only about 0.6% (6,110.5m versus 6,072.7m weighted average), and the buyback overwhelms it by a factor of roughly three.

Compensation policy of directors and management?

[Fact] Remuneration is set by a Compensation Committee a majority of whose members are outside directors; no member may vote on their own compensation. Compensation is equity-linked through restricted stock, RSUs and legacy stock acquisition rights. [Open Question] The 20-F discloses share ownership but not the quantum of individual or aggregate cash-plus-equity compensation in the form a US proxy would provide; that disclosure exists only in the Japanese-language filing.

Motivations of management?

[Interpretation] The revealed preference over the last two years is simplification and capital efficiency: a two-and-a-half-year demerger that removed ¥19.6tn of unrelated balance sheet, an exit from television manufacturing, a fab-light pivot in semiconductors, a tripling of buybacks, and a willingness to take ¥149.9bn of write-offs in a single year rather than defer them. Totoki is a former CFO and COO, and the agenda reads exactly like a former CFO’s.

[Interpretation] The alignment caveat is real and should be stated. Every named executive and director beneficially owns less than 0.1% of the company; three outside directors hold zero shares. Yoshida’s option strikes of ¥1,288 and ¥2,278 against ¥3,409 mean management has been well paid for a share price that tripled — defensible pay-for-performance — but the absolute ownership stakes do not create owner-operator alignment in the way a comparable US filer’s would. Correspondingly, the Form 4 record carries no signal: filings are frequent and paired with Rule 144 notices, consistent with routine vesting and planned disposition, with no discretionary open-market purchase identified in the trailing window. For a US company that absence would be informative; here it simply reflects a different compensation and ownership culture.

Valuation and Market Data

Is the stock an ADR, MLP, or K-1 issuer?

[Fact] It is an ADR. NYSE: SONY, with 1 American Depositary Share representing 1 ordinary share of Sony Group Corporation — an unusually clean 1:1 ratio. The primary listing is Tokyo Stock Exchange Prime (6758). It is not an MLP and issues no K-1. Sony is a foreign private issuer filing Form 20-F annually and Form 6-K for interim disclosure, so US investors receive materially less frequent and less granular financial disclosure than from a domestic filer — no quarterly 10-Q, no 8-K event reporting, no DEF 14A proxy. Japanese withholding tax applies to dividends; the 20-F states that dividends paid to US corporate holders of ADSs are not eligible for the dividends-received deduction. ADR depositary fees apply.

Dividend policy?

[Fact] Sony declared ¥25 per share for the fiscal year ended March 2026 and has raised it by ¥10 to ¥35 per share for FY2026, a 40% increase. Cash dividends paid were ¥98.6bn, ¥115.3bn and ¥135.0bn over three years. The FYE Mar-26 payout ratio on continuing-operations earnings was approximately 14%. [Fact] These figures exclude the dividend in kind of SFGI shares distributed in the partial spin-off, which the 20-F notes separately. At ¥3,409 the ¥35 dividend is a yield of approximately 1.0%.

[Interpretation] Sony is deliberately a buyback-first returner of capital: the ¥500bn repurchase facility is roughly 2.5% of market capitalisation against a 1.0% dividend yield. For a company with net cash and a 7.3% trailing free-cash-flow yield, that is the right emphasis — but income-oriented investors should understand that the yield is not the point here.

How profitable is the business?

[Fact] FYE Mar-26 continuing operations: gross margin 30.8% (up 150bp), operating margin 11.6% (up 100bp), net margin on continuing operations 8.3%, ROIC approximately 13.5% [Interpretation], ROE on continuing operations approximately 12.7% [Interpretation]. Segment operating margins ranged from 21.1% (Music) and 16.6% (I&SS) down to 9.9% (G&NS), 7.0% (ET&S) and 7.0% (Pictures).

Is net income diverging from cash from operations?

[Fact] Yes, and favourably. Operating cash flow of ¥1,945.6bn against continuing-operations attributable net income of ¥1,030.9bn is a ratio of 1.89x; against operating income of ¥1,447.5bn it is 1.34x. The gap is explained principally by depreciation and amortisation of ¥1,180.7bn — which for Sony includes substantial content and catalogue amortisation.

[Interpretation] This is the direction of divergence one wants, but with an important caveat that cuts against a naive EBITDA reading. Sony’s D&A is not primarily maintenance depreciation on factories; a large portion is amortisation of content assets — films, television series, music catalogues — that must be continually replenished with cash production and acquisition spending. EBITDA of ¥2,734.3bn and the resulting ~7.0x EV/EBITDA therefore materially flatter Sony’s valuation, and this article deliberately anchors on EV/EBIT, P/E and free-cash-flow yield instead. Anyone valuing Sony on EBITDA is double-counting the content spend.

[Fact] Against the reported net loss of ¥326.9bn, operating cash flow of ¥1,945.6bn represents a divergence of over ¥2.2tn — which is simply the arithmetic restatement of the central finding of this report: the loss was a non-cash accounting recycling of ¥1,377.8bn of accumulated other comprehensive income, and it never touched cash.

Risks and Downside

What factors would cause the stock to decline?

[Interpretation] In rough order of likelihood-weighted impact:

1. Network Services or Music streaming growth decelerating below mid-single digits — this would break the “annuity carries the cycle” argument, which is the load-bearing element of the case.
2. Memory prices remaining elevated beyond FY2027, forcing a delayed, expensive or economically compromised next-generation console.
3. A further large operating-M&A write-off from the ~¥800bn of undeployed strategic-investment capacity.
4. The TSMC MOU lapsing without a definitive agreement, leaving I&SS as a capital-hungry integrated manufacturer in a decelerating cycle.
5. Yen strength reversing the roughly ¥47bn of FX benefit in FYE Mar-26.
6. Goodwill or content-asset impairment against the ¥4,892.1bn of intangibles.
7. Continued multiple compression across entertainment assets on the AI-content-abundance thesis, independent of Sony’s own results.
8. Broad Japan-equity or yen weakness, given the empirical Country: Japan factor loading of +0.600.

Risk of a catastrophic loss?

[Interpretation] Low, but not zero, and the specific vectors are identifiable. Sony has a documented history of two major security incidents — the 2011 PlayStation Network breach and the 2014 Sony Pictures Entertainment attack — so cyber risk is demonstrated rather than theoretical for this issuer specifically. Physical concentration risk is real: image-sensor manufacturing is concentrated in Kumamoto and Nagasaki, in a seismically active country, and the Kumamoto region experienced major earthquakes in 2016. Geopolitical rupture involving Taiwan would simultaneously affect the TSMC partnership, the broader supply chain and the end markets. None of these is likely in any given year; all are severe if they occur.

Chance of a total loss?

[Interpretation] Very low. Sony holds ¥1,166.9bn of net cash, generated ¥1,487.9bn of free cash flow in the most recent year, operates five separate profit streams across three distinct industries, carries a tangible common equity ratio of 33.6%, and just priced senior unsecured debt at approximately +70bp and +90bp over Treasuries — spreads that reflect a solidly investment-grade credit. A company with that balance sheet and that cash generation does not go to zero absent fraud or catastrophe.

Recent News and Events

Has the business environment changed recently?

[Fact] Yes, in two large and opposite ways.

Structurally, favourably: the partial spin-off of Financial Services executed 1 October 2025 removed ¥19.6tn of assets and liabilities, converted ¥617.8bn of net debt into ¥1,166.9bn of net cash, and lifted the tangible common equity ratio from 13.2% to 33.6%. Sony retained 16.40% of SFGI, now separately listed on the TSE Prime Market.

Cyclically, unfavourably: an AI-datacentre-driven memory shortage has sharply raised the cost of a key input to consoles, televisions and cameras. Trade press reports memory prices roughly doubled in a quarter with a further sharp rise forecast. Management expects prices to remain high into FY2027, has already raised the PS5 to $650 from $550, will set FY2026 console volumes based on memory availability at reasonable prices, and has bounded the ET&S impact at approximately ¥30bn.

[Fact] In addition, Sony’s own CEO identified a narrative change: market anxiety that AI-driven content abundance will compete away entertainment’s share of user attention. He named this, alongside memory, as the reason for the share-price decline when asked directly by Nikkei on 8 May 2026.

Significant acquisitions?

[Fact] Covered above. The headline items of the last two years are the Peanuts Holdings increase to 80%, the Pink Floyd and Queen catalogues, the GIC music-rights joint venture, the Bandai Namco partnership, the TCL home-entertainment joint venture (definitive agreement March 2026), and the TSMC image-sensor MOU (8 May 2026).

Change in accounting policies?

[Fact] Yes — one large presentational change and two reclassifications.

1. From the date the SFGI distribution became highly probable, the Financial Services business was classified as a discontinued operation under IFRS 5, with revenue, expenses, other comprehensive income and cash flows separated from continuing operations, and its assets and liabilities classified as a disposal group held for distribution to owners. Prior-year figures were re-presented accordingly. This is why the comparative income statement in the 20-F shows continuing-operations sales of ¥12,034.9bn for FYE Mar-25 rather than the total previously reported.
2. Lease liabilities, previously included within “Current portion of long-term debt” and “Long-term debt,” are presented separately from FYE Mar-26 given increased materiality; ¥90,495m and ¥508,975m at 31 March 2025 were reclassified to lease liabilities of ¥599,470m.
3. Intangible assets acquired through business combinations, previously included within “Other,” are now presented separately.

[Interpretation] None of these is aggressive. All three improve transparency. The discontinued-operations presentation is mandatory under IFRS 5, and its consequence — an unreadable headline income statement — is a cost Sony had no ability to avoid.

Recent changes — new markets, facilities, management?

[Fact]

• Facilities: a newly constructed fab in Koshi City, Kumamoto is intended to house the TSMC joint venture’s development and production lines, with further capital investment contemplated at Sony’s existing Nagasaki plant, both premised in part on Japanese government support.
• New markets: the TSMC partnership explicitly seeks to “explore emerging new opportunities in physical AI applications, such as automotive and robotics” — a genuinely new end market for Sony’s sensor business.
• Exited markets: electric vehicles (Sony Honda Mobility’s Afeela discontinued), visual effects and virtual production (Pixomondo wound down), and — from April 2027 — televisions, B2B flat-panel displays, home theatre and home audio into the TCL joint venture.
• Management: Hiroki Totoki is President and CEO, with Kenichiro Yoshida remaining Chairman, and Lin Tao as CFO. Hideaki Nishino leads Sony Interactive Entertainment. [Interpretation] This is a continuity transition — Totoki was previously CFO and COO and is the architect of the current financial-discipline agenda — rather than a disruptive one.
• Financing: Sony completed its first US-registered investment-grade bond offering in roughly three decades in June 2026 — two senior fixed-rate tranches, approximately $1bn, at roughly +70bp (5-year) and +90bp (10-year) over comparable Treasuries. [Interpretation] A company with ¥1.17tn of net cash does not need the money; the plausible rationale is currency-matched funding for dollar-denominated assets and re-establishing a USD credit curve.
• Headcount: 94,900 at 31 March 2026, down 17,400, of which 14,300 was the departing Financial Services headcount and the balance came from ET&S restructuring in Japan and I&SS divestitures outside Japan.

APPENDIX B — Source Appendix

Sony Group Corporation (NYSE: SONY / TSE: 6758) — 26 July 2026

All sources accessed 26 July 2026 unless otherwise stated. Sources are ordered by source-quality priority: regulatory filings first, then company communications, then quantitative data services, then trade and financial press, then peer disclosures.

1. Primary regulatory filings — SEC EDGAR (CIK 0000313838)

Note on filing status: Sony is a foreign private issuer. It files Form 20-F annually and Form 6-K for interim disclosure. There is no Form 10-K, 10-Q, 8-K or DEF 14A. A conventional multi-year US-filer corpus sweep is not available for Sony: the 20-F is the sole audited annual document, and the 6-K stream substitutes for both 10-Q and 8-K.

2. Company communications

Verification caveat on source 15. The transcript is machine-generated and contains material numerical corruption — e.g. the FY2026 forecast rendered as “JPY 12.3 billion / operating income of JPY 1.6 billion” (trillions), Pictures FY2025 sales as “JPY 1.993 billion” against the 20-F’s ¥1,499.3bn, and consolidated sales as “JPY 12,796 billion” against the 20-F’s ¥12,479.6bn continuing-operations figure. Every number in this article comes from the Form 20-F. The transcript is quoted only for management’s framing and verbatim commentary.

3. Quantitative data services

Data-service caveats. (a) ROIC.ai now rejects bare tickers and requires EXCHANGE:SYMBOL, a ticker ID or a FIGI. (b) ROIC.ai list_earnings_calls ignored the identifier and returned a cross-ticker list; get_earnings_call_transcript with explicit year and quarter worked correctly. © ROIC.ai get_company_news returned an empty array for NYSE:SONY, so the recent-events timeline was built from SEC filings, company releases and trade press instead. (d) ROIC.ai’s computed operating income of ¥1,553.7bn does not equal Sony’s reported ¥1,447.5bn — Sony’s definition includes the share of profit/loss of equity-method investments (a ¥64.2bn loss); the filing governs. (e) FactorsToday loadings are in-sample ElasticNet estimates over a 756-day window with R² of 0.26–0.37 and are statistical facts about realised co-movement, not forecasts. (f) The AZI adjusted ADR series shows no discontinuity across the 1 October 2025 SFGI dividend-in-kind; total-return comparisons spanning that date carry an unquantified break.

4. Industry data and trade press

Trade-press caveat. Reporting of CMOS image-sensor yield issues creating uncertainty for a key customer appears in secondary commentary only. It is not corroborated in the Form 20-F or on the 8 May 2026 earnings call and is recorded above as an Open Question, not as a fact.

5. Analytical frameworks

6. Reconciliation and verification notes

1. All segment revenue and operating-income figures in this article and its appendices were taken from the Form 20-F “Operating Performance by Business Segment” tables, not from the earnings-call transcript, for the reasons set out under source 15.
2. Sony’s operating income of ¥1,447.5bn is the company-defined measure and includes the share of profit/loss of equity-method investments. It does not reconcile to ROIC.ai’s computed ¥1,553.7bn. The filing governs throughout.
3. The ¥1,377,795m OCI recycling — comprising a ¥1,640,079m loss on FVOCI debt instruments and ¥263,298m of insurance finance income — is taken verbatim from the 20-F discontinued-operations note and independently corroborated against the statement of changes in equity (retained earnings −¥1,383.3bn; other reserves +¥1,795.8bn; equity attributable to owners essentially unchanged at ¥8,179.7bn → ¥8,119.0bn).
4. Normalised operating income of ~¥1,597bn is a construction, not a disclosed figure. Each of the six constituent one-time items is individually disclosed in the 20-F and is itemised in the Section 6 table so the reader can rebuild or reject it.
5. The FX contribution to growth (~¥47bn) is summed from the segment-level foreign-exchange effects disclosed in the 20-F (G&NS operating income +¥54.3bn, ET&S +¥5.3bn, I&SS −¥12.5bn) and is therefore approximate — the 20-F does not disclose an FX effect for every segment’s operating income.
6. Market capitalisation and enterprise value are computed at the TSE closing price of ¥3,409 on 24 July 2026 against approximately 5.97bn shares (6,003.3m at 31 March 2026 less the 37.1m repurchased through 30 June 2026). The retained 16.40% SFGI stake and the Spotify shareholding are excluded from the enterprise-value bridge because neither is separately valued in the 20-F; both bias the stated multiples upward.
7. No third-party analyst price target, rating, or estimate was used as a valuation input anywhere in this article.
8. Position disclosure. Nothing in this article asserts or implies any position, long or short, in Sony Group Corporation.