CRH plc (NYSE: CRH) — Buying at 15x With a 10x Currency

Independent equity research · Report date: 2026-08-15 Price reference: $97.15 (2026-08-14) · CIK 0000849395 · FY ends 31 December · Reporting currency USD, US GAAP An update to my previous CRH report dated 2026-06-14.

⚡ Claude’s Take

This block is my own subjective opinion. It is general information, not investment advice, and you should not act on it without doing your own work. The detailed analysis that follows takes no position and carries no price target — this clearly-fenced block is the only place in this article where I express a view.

Verdict: HOLD — unchanged in direction, but the reasoning has inverted. Accumulate on further weakness; do not chase; still not a short. Fair-value zone lowered to ~$95–110 (≈9.5–11x 2026E adjusted EBITDA of $8.3bn; ≈16–18x 2026E EPS of $5.83), from $105–125 in June. At $97.15 the stock now sits at the bottom of that zone — which is the good news — but the zone itself came down, and the reason it came down is the reason not to be more enthusiastic.

In June I wrote that CRH was a genuinely elite asset at a full price, that the multiple cushion was spent, and that “the price you actually want is sub-$95.” The stock obliged: it printed $95.01 on 31 July and sits at $97.15, down 8.8% while Vulcan fell 1.6%. Two months later the price is right and the case for paying it is weaker. The core moat performed exactly as advertised — Americas aggregates priced +5% through a weather-hit, residential-soft quarter, which passes the single most important bull test I set. But on 22 June, eight days after the last report, CRH agreed to buy Arcosa for $8.5bn — the largest deal in its history — at a headline “11.5x 2026E EBITDA” that is quoted including $175m of year-three synergies. Strip them out, as one must, and the entry multiple is ~15x; Arcosa’s own reported TTM EBITDA of $551m against $8.5bn says 15.4x. CRH is paying ~15x using a currency the market values at ~9.8x, funded with debt, and it has paused the buyback to do it. On disclosed numbers the acquisition returns ~2.8% unlevered after tax today and ~4.3% with every promised synergy banked — against a cost of capital near 8–9%. That is roughly 150bps of group ROIC dilution layered onto a return that had already fallen from 13.4% to 12.1%. The prior report’s bear falsification test was “further ROIC erosion as goodwill-heavy M&A outruns returns.” It fired.

The framing is no longer quality-compounder-at-a-price; the factor tape has stopped supporting that label — the Quality loading is now ~0.02 and Momentum has collapsed to ~0.11, leaving a de-rated high-beta cyclical with a dividend tilt, alpha flipped from +6.2% to −5.1%, price below all three EMAs, 25% off the January high. Call it what it is: a very good business whose management just wrote a large cheque against its own depressed multiple, in a tape that has stopped giving it credit. What keeps this a HOLD rather than a downgrade is that you are being paid to wait: ~9.8x forward EV/EBITDA against Vulcan at 17.6x and Martin Marietta at 18.9x, a 40–44% discount for the owner of the deepest reserve base in the United States, with aggregates pricing compounding at 5%, a genuinely accelerating reindustrialization tailwind (CRH is within 25 miles of 85% of every announced US data center), and a five-year $580bn highway bill that cleared committee 62–2. Conviction: medium. Bullish flip: the FY2026 adjusted-ROIC disclosure holding at or above 12% and the buyback restarting post-close — that would mean the machine still works and the Arcosa cheque was affordable. Bearish flip: adjusted ROIC printing below 11%, or leverage staying above 2.4x into 2028 with the buyback still dark — at which point the pure-play discount is permanently deserved and this is a value trap with a dividend. Tag: the toll road just paid retail for another road.

🔄 Changes Since the 2026-06-14 Report

Nine weeks; an unusually dense stretch. What actually moved:

Verdict on the diff: the business confirmed the thesis and the capital allocation undermined it. The prior report identified exactly one thing that could break the compounding story — M&A outrunning returns — and within eight days management executed the largest possible version of that risk. Meanwhile the price fell far enough that the arithmetic is no longer punishing. Those two facts roughly cancel, which is why the call does not change even though almost everything underneath it did.

One correction to the prior report. It listed Nancy Buese as CFO. The 8-K disclosing her departure (event date 2026-05-11) was filed 2026-05-15, a month before that report went out, and was missed. Aylwyn Bryan has been CFO since 2026-05-12.

📈 Stock Price Action — Five-Year Event Map

CRH has done a full round trip inside five years: from $31.54 at the September 2022 low, through a listing-driven re-rating that took it to $131.38 on 9 January 2026, to $97.15 today — −25.2% off the high, with a 52-week range of $95.01–$131.38 and the low printed on 31 July 2026, the session after the Q2 release. The stock is below its 21-, 50- and 200-day exponential moving averages (99.89 / 102.63 / 107.43). The first four events below are the making of the compounder; the last two are the unwinding of its premium.

Cycle narrative. (1) Through 2022 CRH traded as what it then was — a Dublin-listed conglomerate priced at roughly 7x earnings while rates rose and Europe looked recessionary. (2) The decision to move the primary listing to New York, executed in September 2023, was the single most valuable capital-markets action in the company’s modern history; the stock nearly doubled off the low in anticipation. (3) Index inclusion and a run of clean prints carried the multiple from a European discount to a US-industrial rating — the entire “Act One” re-rating the June report described as already spent. (4) A strong Q2 2025 (the largest idiosyncratic up-move in the trailing year) launched a final leg to the all-time high on 9 January 2026. (5) From that peak the market rotated out: the 30 September 2026 IIJA expiry moved into view, new-build residential stayed weak, and on 22 June the Arcosa agreement introduced $8.5bn of debt-funded M&A and suspended the buyback. (6) The Q2 release on 30 July was the capitulation point — headline growth was fine but earnings quality was not (a $0.16 divestiture gain inside a 14% EPS increase), guidance was reaffirmed rather than raised, and management confirmed the buyback would stay dark. The stock fell 8.1% across two sessions to a 52-week low. Notably, CRH fell 8.8% over the full window while Vulcan fell 1.6% and Martin Marietta 5.1% — this was a CRH-specific de-rating, not a sector move.

1. Executive Summary

CRH plc is the world’s largest building-materials company (TTM revenue $38.6bn, TTM adjusted EBITDA ~$7.8bn) and the dominant North American producer of aggregates, asphalt and road solutions, with roughly three-quarters of profit generated in North America across three segments — Americas Materials Solutions, Americas Building Solutions and International Solutions. The durable scaffolding of the investment case is unchanged from the June report and is not re-litigated here: aggregates is a permit-protected local-monopoly business in which Marathon’s capital-cycle mean-reversion mechanism is structurally broken, CRH owns the deepest US reserve base (~23.8bn tons), and its vertically integrated “connected portfolio” wins whole infrastructure jobs that single-product competitors cannot bid.

What this update establishes is that the business and the capital allocation have diverged. The business did what the bull case required: Americas aggregates pricing rose 5% in Q2 2026 on volumes up 2%, through a quarter disrupted by weather in the US South and Southeast, with new-build residential subdued and costs inflating. That is the moat working, and it passes the primary falsification test the June report set. Reindustrialization demand accelerated genuinely — CRH is active on 200 data-center projects and sits within 25 miles of 85% of every announced US data center, with one East Texas job alone consuming 3 million tons of aggregates. The infrastructure funding picture improved: the five-year, $580bn BUILD America 250 Act cleared the House Transportation and Infrastructure Committee 62–2 on 22 May.

The capital allocation went the other way. On 22 June CRH agreed to acquire Arcosa for $8.5bn in cash — its largest transaction ever. The disclosed multiple of “11.5x 2026E adjusted EBITDA” is stated including $175m of year-three run-rate synergies; stripping them gives an entry multiple of ~15.1x, corroborated by Arcosa’s own reported TTM adjusted EBITDA of $551m ($8.5bn ÷ $551m = 15.4x). CRH’s own forward EV/EBITDA is ~9.8x. On Arcosa’s TTM operating income taxed at CRH’s guided rate, the deal earns an unlevered ~2.8% today and ~4.3% with all promised synergies — beneath any plausible cost of capital — implying roughly 150bps of group ROIC dilution against an adjusted ROIC that had already slipped from 13.4% (2024) to 12.1% (2025). Independent GAAP data confirm the direction (10.37% → 10.17%). To fund it, CRH paused the share buyback — the mechanism that retired 24% of the share count since 2018 — and pro forma leverage rises from 1.8x to 2.4x.

Q2 2026 itself was solid but lower-quality than the headline. Revenue rose 5.6% and adjusted EBITDA 7% with 30bps of margin expansion; reported EPS rose 14%, but management disclosed that $0.16 of that came from divestiture gains, putting underlying EPS growth nearer 5.7%. Americas Building Solutions posted an 8% EBITDA decline and a negative organic incremental margin on haulage inflation and residential weakness. Guidance was reaffirmed, not raised, and capex guidance was trimmed.

The market has responded. The stock fell 8.8% since the prior report to $97.15 — a 52-week low of $95.01 was set on 31 July — while Vulcan fell 1.6% and Martin Marietta 5.1%. The pure-play discount did not close; it widened modestly, firing the June report’s fourth bull falsification test. Forward EV/EBITDA is now ~9.8x and forward P/E ~16.7x, against ~10.4x and ~18.3x in June, on unchanged guidance. Factor data show the transition plainly: alpha has flipped from +6.2% to −5.1%, the Quality loading has collapsed to ~0.02 and Momentum to ~0.11.

This article takes no position and sets no price target. It frames valuation as embedded expectations: at ~9.8x forward EBITDA the market is underwriting mid-to-high-single-digit EBITDA growth, delivery of the Arcosa synergies, and no further returns deterioration — while pricing in no closure of the peer discount. That is a materially less demanding hurdle than in June, applied to an asset whose capital-allocation quality is one notch worse.

2. Business Overview

What CRH does, and how the shape changed this year. CRH manufactures the physical inputs of construction and the integrated solutions assembled from them: aggregates (crushed stone, sand and gravel — 380.7m tons globally, ~230m tons in North America), cementitious materials, asphalt and ready-mixed concrete at the base; road and paving construction services, building and infrastructure products (precast, water-management structures, utility infrastructure) and outdoor-living products on top. Revenue is transactional, not contractual — there is no subscription base. Stickiness comes from geography rather than paper: a contractor cannot economically source aggregates from anywhere but the nearest quarry, so the lock-in is physical.

Segment structure and Q2 2026 performance. The three-segment structure is unchanged, but the quality distribution within it moved this quarter.

Americas Materials Solutions remains the profit engine and had an unambiguously good quarter: revenue +10% (organic +$170m, acquisitions +$312m), EBITDA +12%, margin up 40bps. Within it, Essential Materials revenue rose 20% on aggregates pricing +5% (volumes +2%) and the Eco Material contribution; Road Solutions rose 6% on asphalt volumes +3% and pricing +6%. This is where the moat lives and it delivered.

Americas Building Solutions is the segment to watch. Revenue fell 2% and EBITDA fell 8%, with margin down 140bps. The composition matters more than the headline: divestitures removed $192m of revenue and $37m of EBITDA, but organic revenue rose $145m while organic EBITDA fell $22m — a negative organic incremental margin, the first in the group. Management attributes this to elevated haulage rates and subdued new-build residential demand, and says surcharges and cost actions are in place with the impact moderating into Q3 and Q4. The explanation is specific and testable. But this is a genuinely weaker read than the “lower cyclicality in parts” framing the June report gave the segment, and it is the one place where the connected-portfolio story is currently not paying.

International Solutions continued its quiet rehabilitation: revenue +5%, EBITDA +8%, margin +70bps, with aggregates volumes +10% and cement volumes +6%. Management expects a ninth consecutive year of positive International pricing. Central and Eastern Europe recovered from a weather-hit Q1, with defence infrastructure emerging as a new demand category; Ireland, Spain and the Nordics were strong; the UK (Tarmac, >50% infrastructure) is on track for another year of profit and margin growth. The “International is the drag” caricature remains incomplete.

Portfolio shape. Active recycling continued: $1.9bn of divestiture proceeds from three non-core disposals (construction accessories $0.7bn, lawn and garden $1.1bn, MoistureShield $0.1bn) against $1.4bn deployed into 17 acquisitions, the largest being Axius Water at $0.7bn (closed 29 May 2026), which moves CRH up the value chain into water treatment and quality. The direction of travel — out of low-growth manufactured products, into aggregates, cementitious, roads and water — is coherent and consistent with the four-platform strategy. Arcosa, when it closes in Q1 2027, will be the largest single step in that direction and will add roughly 35m tons of annual aggregates plus a utility-structures manufacturing business.

Verdict: unchanged in substance from June — a real, tangible, essential-products business with a dominant position in the best node of its industry — with one new qualification. The segment data now show the group’s quality is more concentrated in Americas Materials than the blended figures suggest: that segment grew EBITDA 12% at a 27.9% margin while Americas Building Solutions shrank EBITDA 8% at 21.8%. The profit centre is doing more of the work, and the diversification is contributing less.

3. Industry Dynamics

The structural analysis from the June report stands and is not re-derived: aggregates sells for ~$15–20/ton against freight of ~$0.15–0.30 per ton-mile, making anything beyond a 30–50 mile radius uneconomic; new quarries take over a decade to permit against intensifying local opposition; and the consequence is that high returns do not attract offsetting supply, because capital cannot manufacture a permit. In Marathon’s capital-cycle terms this is the rare industrial setting where the mean-reversion mechanism is structurally broken on the supply side. What follows is what changed.

The moat produced fresh evidence this quarter, and it is unusually clean. In a single quarter, in a single company, in the same geography: aggregates pricing +5%; cement pricing −1%. Aggregates is the permit-protected, freight-bound, locally monopolistic node. Cement is capital-intensive, energy-hungry, importable at coastal markets and genuinely contestable. The June report argued the moat was “barbell-shaped” — very strong in aggregates, thin in cement and downstream products. Q2 2026 is close to a natural experiment confirming it, and it also explains why the pure-play margin premium is partly deserved rather than purely a listing artefact. Management notes that mix-adjusted cementitious pricing including Eco Material was positive and that cement margins still expanded on 3% volume growth and efficiency — both true, and both beside the structural point.

The infrastructure funding picture improved materially but is not resolved. The IIJA authorization expires 30 September 2026. Since the June report, the successor bill has advanced meaningfully:

• The BUILD America 250 Act (H.R. 8870) was introduced 19 May 2026 and approved by the House Transportation and Infrastructure Committee on 22 May 2026 by 62–2 — about as strong a bipartisan signal as this process generates.
• It is a five-year (FY2027–2031) package of roughly $580bn, including approximately $474.4bn of contract authority — guaranteed, multi-year Highway Trust Fund money that gives states planning certainty — with roughly $106bn subject to annual appropriations.
• It addresses the revenue side for the first time in three decades, imposing annual federal registration fees on electric vehicles ($130/year) and plug-in hybrids ($35/year), escalating biennially from 2029 to caps of $150 and $50.

Three caveats keep this from being a resolution. First, the full House has not voted and the Senate has not released text — the Senate Environment and Public Works Committee has held hearings but advanced nothing. With six weeks to the deadline, the realistic base case is a short-term extension at current funding levels, not a signed multi-year bill by 30 September. Second, the revenue fix is token relative to the problem: the EV/PHEV fees raise under $10bn over five years against a Highway Trust Fund gap the June report sized at roughly $166bn. Management said so plainly — “This won’t go the whole way, but it’s an important first step.” Anyone treating this bill as removing the funding-gap risk is over-reading it. Third, contract authority in a bill is not spending in a quarter.

What genuinely de-risks the near term is that the money already appropriated has not been spent. Management expects to exit 2026 with ~40% of the IIJA still unspent (against ~57% unspent in early 2026 — a consistent burn). Randy Lake’s point on the call is the operationally correct one: “even if we don’t reach a multiyear bill in the fall, our starting point for that extension is coming off record levels,” and in past continuing resolutions states have reallocated toward repair and maintenance — which favours the largest road paver in the United States. State DOT budgets remain at records. So the volume risk from a lapse is a 2028 problem, not a 2027 problem.

Reindustrialization is now the strongest demand leg, and it accelerated. This is the most important demand change since June. CRH is active on 200 data centers; 85% of all announced US data centers sit within 25 miles of a CRH facility; and the connected portfolio monetises far more of each project than a quarry alone would. Management’s East Texas example is worth quoting for its specificity: an 85-acre building footprint consuming 3 million tons of aggregates, where CRH is “first in” on subterranean energy and water infrastructure, then supplies cementitious material for site stabilization, then aggregates, then ready-mix, then asphalt and paving for access roads and car parks — a two-to-three-year revenue stream from one customer relationship. CRH has built dedicated sales teams that engage hyperscalers directly at the design and specification stage rather than through general contractors. Add advanced manufacturing, semiconductor fabs and LNG plants, and Mintern’s framing — “the early phase of a generation of capital investment” — is supported by the backlog commentary (bidding volumes and secured work both up year-on-year).

Residential got worse, and the timeline slipped. Management now expects no new-build recovery in 2026 and “the back end of '27 at best.” Repair-and-remodel remains resilient. This matters more than its revenue share implies because ready-mix is the most directly exposed product line (volumes flat, pricing +2%) and because the stock’s factor loadings carry a large residential-construction exposure (see the Variant Perception section).

European cement carbon costs (EU ETS, CBAM phasing in from 2026) remain a real but contained headwind for a minority European business, immaterial to the ~75% North American operations. No change since June.

Verdict: structurally good industry, best-in-class at the aggregates node — and the evidence improved this quarter. The aggregates/cement pricing divergence is direct confirmation of the moat’s shape. The funding picture improved on the margin (a 62–2 committee vote, a five-year $580bn framework, and the first revenue-side action in thirty years) without being resolved, while the already-appropriated IIJA backlog insulates 2027 volumes almost regardless of the September outcome. Reindustrialization has become a genuine second engine. Residential is worse for longer. Net: the industry read is modestly better than in June.

4. Competitive Position

The moat, re-tested. CRH’s advantage remains a cost/supply advantage — privileged access to irreplaceable, permit-protected reserves near demand — reinforced by local economies of scale and customer captivity, the strongest archetype in Greenwald’s taxonomy. The reserve base (~23.8bn tons across 1,334 mining properties) is the deepest in the United States. Greenwald’s two tests both continue to pass, and one of them produced new evidence this quarter.

Market-share stability. Local quarry positions essentially never change hands competitively; they change hands by acquisition, because the only way into a local market is to buy the incumbent. The Arcosa transaction is itself a demonstration: Mintern was explicit that Dallas and Phoenix were “geographic white spaces” for CRH aggregates — two of the fastest-growing MSAs in America, where CRH already operated other parts of the connected portfolio but could not obtain stone. CRH did not permit its way in over a decade; it wrote a cheque. That is the test passing, expensively.

Return persistence. Aggregates pricing +5% in Q2 2026, through weather disruption, soft residential demand and cost inflation, with mid-single-digit pricing guided for the full year and a ninth consecutive year of positive International pricing. Price rose while the environment deteriorated. That is the financial fingerprint of an advantage competitors cannot erode.

Where the moat is thin, quantified this quarter. Cement pricing fell 1%. Americas Building Solutions — manufactured, shippable products — posted a negative organic incremental margin. Both are the contestable ends of the barbell, and both underperformed in the same quarter the protected end priced +5%. The honest conclusion is stronger than June’s: the moat is not merely “barbell-shaped” in theory, it is demonstrably so in the reported numbers, and the blended franchise is therefore correctly valued at some discount to a pure-aggregates peer. The dispute is only about the size of the discount.

The peer comparison, refreshed at current prices.

Two readings, both changed since June. First, the asset-base case is intact and if anything stronger — CRH is larger, owns more reserves, moves more tons, earns a higher return on capital than either pure-play, and has just added 35m tons in premium Sunbelt MSAs. Second, and this is the new and uncomfortable fact: the discount did not close, it widened. The June report set an explicit falsification test — “the EV/EBITDA gap to VMC/MLM widens despite in-line execution” would falsify the closeable-discount thesis. CRH delivered in line with its own guidance and fell 8.8% while Vulcan fell 1.6%. The test fired. Two consecutive reporting periods have now failed to produce any convergence.

The natural interpretation is not that the market is stupid. It is that investors pay for purity of exposure to the single best node, not for size, diversification or even reported ROIC — and that CRH’s Q2 numbers gave them fresh reason to (aggregates +5%, cement −1%, products segment shrinking). A second interpretation, which the capital-allocation section develops, is that the market re-rated CRH downward on the Arcosa announcement specifically, and the relative performance is the tell.

What Arcosa does and does not add to the competitive position. The aggregates half is genuinely good: 35m tons, 109 quarries and yards, positions in 13 of the 50 largest US MSAs across Texas, New Jersey, Arizona, Florida and Tennessee, with unit economics that are excellent in their own right — adjusted cash gross margin expanded 190bps to 48.1% in Q2 2026 with cash gross profit per ton up 5% on a 3% unit-cost reduction. On strategic fit this is close to ideal, and it is the only available route into Dallas and Phoenix.

But roughly half of what CRH is buying is not aggregates. Arcosa’s Engineered Structures segment — utility and transmission structures plus wind towers — generated $301.7m of Q2 revenue against Construction Products’ $357.0m. It is a steel-fabrication manufacturing business serving grid modernization. It is growing well (utility structures revenue +12%, record backlog of $648.1m, +49% year-to-date) and it is aimed at a real megatrend. It also has no permit moat, no freight-radius protection and no geological scarcity — it is a contestable manufacturing business exposed to steel costs and order cyclicality, with wind-tower revenue already down 19%. This materially weakens the strongest defence of the ~15x entry multiple, namely “listed aggregates trade at 17–19x so 15x is a bargain.” You are paying ~15x for a blend of roughly half moat-protected aggregates and half contestable manufacturing.

Verdict: durable competitive advantage, intact and freshly evidenced — with the market’s willingness to pay for it moving the wrong way. The moat passed its test this quarter more convincingly than in June. But the competitive position and the competitive valuation have decoupled: CRH’s operating advantage is being confirmed while its relative multiple erodes. The Arcosa deal strengthens the aggregates position in exactly the markets CRH could not otherwise enter, and simultaneously dilutes the average moat quality of the asset base by adding a large non-moated manufacturing business.

5. Growth History and Forward Opportunities

The historical record is unchanged and remains strong. Revenue compounded from $25.9bn (2020) to $37.4bn (2025), ~7.6% annually, with margin expanding throughout — gross margin 33.1% → 36.1%, adjusted EBITDA margin ~16.6% → 20.5%. Management expects 2026 to be the thirteenth consecutive year of margin expansion, which, if delivered, is a genuinely rare record. TTM revenue is now $38.6bn.

Current-year growth composition. H1 2026 revenue grew 7.0%, with Q2 at 5.6%. The build is roughly: mid-single-digit pricing in the protected core, low-single-digit organic volume, plus acquisitions net of an unusually heavy divestiture programme. CFO Bryan quantified the net portfolio effect precisely: ~$200m of net incremental EBITDA for the year from the combination of $1.4bn of acquisitions and $1.9bn of divestitures. That is a useful discipline check — the scope activity is roughly 2.5% of EBITDA, so the reported growth is mostly organic and mostly price.

The pricing component is the highest-quality piece and it held. Americas aggregates +5%, asphalt +6%, ready-mix +2%; International aggregates +2%, cement +4%, ready-mix +3%. Mid-single-digit pricing carries close to 100% incremental margin, which is why the group expanded margin 30bps despite an inflationary cost backdrop and a segment posting negative organic incrementals. Management guides to mid-single-digit pricing for the full year with “midyear pricing in place,” giving momentum into 2027.

The four growth platforms, with the water leg now materially advanced. The framework is unchanged — aggregates, cementitious, road solutions, water — but water made real progress: Axius Water closed 29 May 2026 for ~$0.7bn, moving CRH up the value chain from transmission and structures into nutrient removal and wastewater treatment. The addressable opportunity remains large and fragmented (a ~$100bn+ annual US ecosystem; a third of US water infrastructure over 50 years old; roughly 4 million lead service lines to replace). Water is still not a needle-mover against $38.6bn of revenue, and the open question from June stands: does it earn aggregates-like returns, or dilute group ROIC as a lower-margin services adjacency? Too early to tell.

Reindustrialization has become the most credible near-term growth vector. This is the genuine upgrade since June. The evidence is no longer anecdotal: 200 active data-center projects, 85% proximity coverage of announced US data centers, dedicated hyperscaler-facing sales teams engaging at design stage, and a disclosed single project consuming 3 million tons of aggregates over two to three years across five different CRH product lines. Randy Lake confirmed both bidding volumes and secured work are up year-on-year, with private reindustrialization taking an increasing share of backlog. Backlog gives six-to-nine months of visibility and “remains positive.” Crucially, this demand leg is funded privately — it is independent of the September reauthorization outcome, which makes it a genuine diversifier of the public-infrastructure concentration.

Arcosa as growth. It adds ~$2.7bn of revenue and ~$564m of standalone EBITDA — roughly 7% of both — plus 35m tons of aggregates in fast-growing Sunbelt MSAs and a utility-structures business with a record, fast-growing backlog aimed at grid modernization and electrification. As growth, it is real and reasonably high quality. As capital allocation, it is the problem, and the capital-allocation section treats it there. The distinction matters: this is not a bad asset bought for bad reasons; it is a good asset bought at a price that struggles to earn its cost of capital.

The 2030 targets, restated on the call. Management reaffirmed 7–9% annual revenue growth, a 22–24% adjusted EBITDA margin by 2030 (from ~20.5% in 2025) and average adjusted free-cash-flow conversion above 100%, backed by ~$40bn of five-year financial capacity allocated ~70% to growth and ~30% to shareholder returns. These are management targets, not forecasts, and the 30% shareholder-return leg is currently being honoured by dividend alone.

Verdict: high-quality growth at the operating level, with the capital-efficiency caveat now larger, not smaller. The organic pricing engine — the part that reflects the moat — is performing exactly as it should and is the reason margin expanded in a difficult quarter. Reindustrialization has emerged as a real, privately-funded second demand leg. But the growth is increasingly being bought, and the June caveat that “the returns on incremental capital are thinning” has been sharpened by an $8.5bn transaction that, on disclosed figures, returns less than half CRH’s cost of capital before synergies. Growth is not the problem. The price paid for it is.

6. Financial Quality

The quarter, and what the headline conceals.

The single most important quality-of-earnings point in this update: reported EPS growth of 13.9% is roughly half real. Other nonoperating income swung by $291m pretax, “primarily related to gains on divestitures,” and income tax expense rose $236m “mainly due to the impact of divestitures in the period,” pushing the effective tax rate to 30.6% from 24.2% (against ~26% guided). Management quantified the net effect itself, to its credit: EPS growth “including a $0.16 net gain on divestitures in the period.” Stripping it, underlying diluted EPS was ~$2.05 against $1.94 — +5.7%. The cleaner cross-check: excluding other nonoperating income from both periods, pretax income was $1,881m versus $1,765m, +6.6% — consistent with adjusted EBITDA +7% and operating income +7.4%. The business grew mid-single-digit earnings; the income statement reported low-double-digit. Anyone comparing “+13% net income” to a peer’s operating growth is comparing the wrong things.

H1 2026 and the operating-leverage question. H1 revenue rose 7.0% and gross profit 8.1% — but SG&A rose 9.4%, and a $48m impairment was booked in Q1 (nil in Q2; none in H1 2025). The result is H1 operating income up only 4.5%. None of this is alarming on its own: the impairment is small, and a single half’s SG&A growth can reflect acquisition mix and integration costs. But the direction is the opposite of the operating-leverage narrative, and against a management target of reaching a 22–24% EBITDA margin by 2030, cost growth outrunning revenue growth is the thing that would prevent it. Watch Q3.

Cash flow — and a trap worth disarming explicitly. Reported H1 operating cash flow fell to $513m from $719m, down 29%. Taken at face value that looks like a serious deterioration in cash conversion. It is mostly mechanical. The operating reconciliation deducts “gain on disposals from businesses and long-lived assets, net” of $(334)m in H1 2026 against $(12)m in H1 2025 — a $322m non-cash swing whose cash counterpart appears in investing, as the $1,676m of divestiture proceeds. Adjusting for it, underlying H1 operating cash flow is broadly flat. Receivables corroborate: DSO was 57.2 days at Q2 2026 versus 57.9 days a year earlier — collections are fine. What is genuinely worse is cash interest ($335m vs $251m) and cash taxes ($404m vs $304m), together ~$184m, both of which are real and both of which grow further after Arcosa.

H1 free cash flow (CFO less $1,240m of capex) was −$727m against −$581m a year earlier. CRH’s first half is structurally cash-negative on seasonal working-capital build for the construction season, so this is normal in sign; the full-year figure and the >100% conversion target are what matter.

Returns on capital — the crux, now confirmed on two independent bases. The June report flagged management’s adjusted ROIC falling 130bps to 12.1% in 2025 from 13.4% in 2024, and noted the possible objection that this is a non-GAAP definition. That objection can now be closed. Independently computed GAAP return on invested capital runs 7.72% (2021) → 8.81% (2022) → 9.40% (2023) → 10.37% (2024) → 10.17% (2025) — the same peak-and-slip, on a basis management does not control. Return on common equity was 16.4% in FY2025. With $13.15bn of goodwill (22% of assets) and $2.04bn of other intangibles, price to tangible book is ~7.3x. The GAAP ROIC spread over an estimated 8–9% WACC is thin to nil; on the adjusted basis it is comfortable but narrowing.

The FY2026 adjusted-ROIC figure is an annual disclosure and will not appear until the FY2026 10-K. It is the single most important number this company will publish, and the Capital Allocation section explains why the Arcosa transaction makes a further decline close to arithmetically unavoidable.

Balance sheet. Total debt $17,926m at 30 June ($2,516m current, $15,410m long-term) against $17,653m at year-end; cash and equivalents $3,025m plus $58m restricted; CRH-basis net debt $15.4bn versus $14.2bn at 2025-12-31. Undrawn committed facilities of $4.5bn run to May 2030, and a $5.8bn bridge was in place at the balance-sheet date for the Arcosa consideration. Equity attributable to CRH $24,035m; shares outstanding 665,895,636. At ~1.9x annualised EBITDA the current balance sheet is comfortable and investment-grade; pro forma for Arcosa management guides to 2.4x, which is still investment-grade but is the highest leverage CRH has carried in this cycle and removes the optionality that a 1.8x balance sheet provides.

Guidance. FY2026 reaffirmed — net income $3.9–4.1bn, adjusted EBITDA $8.1–8.5bn, diluted EPS $5.60–6.05, on ~$0.7bn net interest, ~26% tax rate and ~674m average diluted shares. Capex guidance was cut from $2.8–3.0bn to $2.7–2.9bn “to reflect project timing and lower maintenance spend.” A capex trim is mildly positive for near-term free cash flow and mildly concerning if it reflects demand caution; the stated reason is benign and the magnitude ($100m) is immaterial either way.

Verdict: operating quality high and holding; earnings quality lower this quarter; capital efficiency now confirmed as the binding constraint. Margins expanded, pricing held, the balance sheet is sound, and the cash-flow “deterioration” is an accounting artefact rather than a business problem — all of which is genuinely good and should not be lost in the criticism. But reported earnings growth was roughly double the underlying rate because of divestiture gains; SG&A outgrew revenue in H1; and the returns-on-capital slippage that the June report identified as the one blemish is now corroborated by GAAP data and about to be compounded by the largest acquisition in company history. Economics still improve with scale operationally. They are now demonstrably deteriorating on capital.

7. Capital Allocation

This is where the thesis moved, and it deserves the space.

The model has not changed; the scale of the bet has. CRH runs a programmatic bolt-on machine — over 1,200 acquisitions historically, ~$3–5bn a year, mostly tuck-ins that densify existing local networks. In H1 2026 that machine did its normal work: $1.4bn across 17 acquisitions, largest being Axius Water at $0.7bn, funded partly by $1.9bn of divestitures of non-core businesses. That part of the record remains disciplined and defensible, and CFO Bryan’s disclosure that the whole programme nets to ~$200m of incremental EBITDA for the year is exactly the kind of honest scope quantification that builds confidence.

Then there is Arcosa. On 21 June 2026 CRH agreed to acquire Arcosa, Inc. for $150.00 per share in cash, a total enterprise value of approximately $8.5bn — the largest transaction in the company’s history, roughly twice the size of any prior deal, expected to close in Q1 2027 subject to an Arcosa stockholder vote scheduled for 4 September 2026 and HSR/regulatory clearance.

The multiple, stripped. CRH’s press release describes “an acquisition multiple of 11.5x 2026E Adjusted EBITDA, including estimated annual run-rate cost synergies of $175 million by year three.” The word including is doing all the work. Reversing it:

• $8,500m ÷ 11.5 = $739m of post-synergy EBITDA
• less $175m of synergies = ~$564m of Arcosa standalone 2026E adjusted EBITDA
• $8,500m ÷ $564m = ~15.1x true entry multiple

An independent check, using Arcosa’s own reported figures rather than CRH’s arithmetic: Arcosa’s TTM adjusted EBITDA to 30 June 2026 was $551.3m. $8,500m ÷ $551.3m = 15.4x. The two methods agree at roughly 15x. The headline 11.5x is a post-synergy number credited three years forward; the price actually being paid is ~15x.

The returns test. Multiples are a shorthand; the question is what the capital earns. Arcosa’s TTM operating income was $322.7m. Taxed at CRH’s own guided 26% rate, that is NOPAT of ~$239m — an unlevered return on the $8.5bn purchase price of ~2.8%. Credit the entire $175m of year-three synergies as a pretax EBIT uplift, and NOPAT rises to ~$368m, or ~4.3%. Against a CRH cost of capital of roughly 8–9%:

On disclosed numbers, the acquisition does not earn its cost of capital even after every promised synergy is banked. It requires several further years of Arcosa growth on top of full synergy delivery to clear the hurdle. That is not a forecast; it is arithmetic from figures both companies have published.

The consequence for group returns. CRH’s average invested capital is roughly $36bn earning ~12.1% on management’s adjusted basis. Adding $8.5bn earning ~4.3% blends to approximately 10.6% — about 150bps of group ROIC dilution from this transaction alone, before any further bolt-ons. The estimate is directional: purchase accounting, Arcosa’s growth and the precise invested-capital definition all move it. The sign is not in doubt. And it lands on a return that had already fallen from 13.4% to 12.1%.

The counter-argument, stated fairly, because it is not weak. Vulcan trades at ~17.6x and Martin Marietta at ~18.9x TTM EBITDA. Against the listed price of pure-play aggregates, 15x for a business whose aggregates arm earns a 48.1% cash gross margin — with 190bps of margin expansion and 5% cash-profit-per-ton growth in the latest quarter — in Dallas, Phoenix and eleven other top-50 MSAs, is not obviously expensive. You cannot permit your way into Dallas; buying the incumbent is the only entry. And CRH’s synergy record is real enough that the market has never seriously challenged it.

Three things blunt that defence. First, roughly half of Arcosa is not aggregates — Engineered Structures generated $301.7m of Q2 revenue against Construction Products’ $357.0m, and it is contestable steel fabrication with no permit moat, wind-tower revenue already down 19%, and steel-cost exposure. Paying an aggregates multiple for a half-manufacturing business is a mix problem. Second, the relevant cost of capital is CRH’s own multiple, not the peer group’s. CRH is issuing debt against an equity the market values at ~9.8x forward EBITDA in order to buy at ~15x. Unless one believes CRH’s own multiple is simply wrong and will converge upward — a belief the last two quarters have actively punished — that is a transfer of value from CRH holders to Arcosa holders. Third, the synergy yield is thin: $175m on $8.5bn is a 2.1% synergy yield, phased $60m in year one and full run-rate only in year three.

The premium, re-struck. The release quotes “a 25% premium to Arcosa’s 60-day trading VWAP as of June 18, 2026” — a VWAP, not a last close. Arcosa closed at $135.84 on 18 June, the last unaffected session, having already risen 7.0% from $126.95 on 11 June on no disclosed news. Against the last unaffected close, the premium is +10.4%. This cuts in CRH’s favour on price — they paid a modest premium to the prevailing mark — but the presentational choice of a 60-day VWAP is worth recognising for what it is, and the pre-announcement run-up is a familiar pattern. Arcosa closed at $145.40 on 14 August against the $150 cash price, a 3.2% gross spread, so the arbitrage market assigns a high completion probability.

The buyback pause — the change most likely to be underweighted. From the earnings release: “in connection with the agreement to acquire Arcosa, CRH has not initiated a new tranche of its share buyback program. CRH will reevaluate… at a later date.” CFO Bryan on the call: “we have taken the decision to pause our share buyback program following the completion of the latest tranche.” The last tranche completed 28 July, bringing YTD repurchases to $0.7bn.

This matters more than it sounds. Since the programme began in 2018, CRH has returned ~$10bn and repurchased 24% of its shares — the share count fell from 785.1m in 2020 to 665.9m today. That was one of the three pillars of the June report’s B+/A− capital-allocation grade and a large part of how per-share value compounded. It is now switched off for an indefinite period, to fund a deal struck at ~15x. Put plainly: CRH has stopped buying its own equity at ~9.8x EBITDA in order to buy someone else’s at ~15x. Whatever the strategic merits, that is the wrong direction on price. The dividend continues — $0.39 per quarter declared 30 July, +5% year-on-year, ~2.0% yield — so the income leg is intact and management’s ~30% shareholder-return allocation is being honoured by dividend alone.

Financing and leverage. A $5.75bn bridge was signed on 22 June; on 17 July CRH termed out $2.5bn of it into a three-year term loan, cutting bridge commitments to $3.25bn — sensible, prompt liability management that reduces refinancing risk. Pro forma FY2026E net debt to adjusted EBITDA rises to 2.4x from 1.8x at FY2025, with a stated commitment to a strong investment-grade rating. That is not a stressed balance sheet. It is, however, the top of CRH’s recent range, and it consumes the optionality that a 1.8x balance sheet provided in a business whose volumes are cyclical.

Incentives and governance. The compensation architecture is unchanged and reasonably well-designed for a roll-up: annual bonus on diluted EPS, cash flow and RONA; long-term plan on relative TSR with a sustainability modifier. Two observations. RONA and relative TSR are the right anchors, but RONA is an internal net-assets measure rather than an explicit ROIC-versus-WACC hurdle — and a deal returning 2.8–4.3% unlevered is precisely the kind of transaction an explicit WACC hurdle would have flagged and a RONA/EPS framework may not. Say-on-Pay support slipped to 93.6% from 94.6%. Tony Will (former CF Industries CEO) joined the board effective 1 July, expanding it to 13 — a credible industrial operator addition. The 5% and 7% “A” preference share classes were cancelled by scheme of arrangement in May, simplifying the capital structure.

The CFO turnover. Nancy Buese, hired in May 2025 as an external, US-credentialed appointment, stepped down “by mutual agreement” on 11 May 2026 after roughly twelve months. Aylwyn Bryan, a 14-year CRH veteran most recently CFO of the Americas Division and Group Head of Finance from 2022, succeeded her the following day. Nothing in the record suggests an accounting or control problem: the 8-K carries standard no-disagreement language, Deloitte was ratified without incident at the AGM, and there have been no restatements. But two CFOs in thirteen months at a company running an $8.5bn acquisition and a $40bn deployment plan is a real execution risk, and the incoming CFO had been in the seat six weeks when the largest deal in company history was agreed. That is a fact worth holding; it is not evidence of wrongdoing and should not be read as such.

Insider behaviour. Form 4 activity from 1 May to 15 August 2026 is almost entirely routine RSU vesting — code M settlements paired with code F tax withholding on 13 May across the board and executive team. The only discretionary transaction is a single sale (Padraig O’Riordain, 1,492 shares at $104.445, ~$156k — immaterial). There were zero open-market purchases. No insider bought during a drawdown that took the stock 25% off its high to a 52-week low, and none bought after announcing the transaction management describes as transformational. This is weak-form evidence — CRH’s European heritage means US-style insider buying was never a strong feature, and the June report found only three small director purchases historically. But the June read of “neutral-to-slightly-positive” should be marked down to neutral-to-slightly-negative on the absence of conviction buying at the lows.

Verdict: the grade falls from B+/A− to B−. The multi-decade record of accretive consolidation is real, the divestiture discipline is genuine, the liability management was prompt, and the dividend keeps compounding. But in nine weeks management has: paid ~15x pre-synergy for its largest-ever acquisition, roughly half of which is not a moated business; committed to a transaction that on disclosed figures earns ~4.3% unlevered against an ~8–9% cost of capital even at full synergies; suspended the buyback that had retired 24% of the share count; and raised pro forma leverage by 60bps — all while its own equity trades at a 40%+ discount to pure-play peers and its return on capital is already declining on two independent measures. The strategic logic is coherent. The price and the funding choice are not defensible on the numbers disclosed, and management was not asked a single question about either on the earnings call.

8. Changes and Headwinds — Last Two Years

The two-year picture from June holds; this section records what moved in the last nine weeks and reweights the headwinds.

Corporate and strategic.

• Arcosa agreed (21/22 June 2026) — $8.5bn, all cash, closing Q1 2027; stockholder vote 4 September 2026. Treated fully in the Capital Allocation section.
• Buyback paused (announced 22 June; last tranche completed 28 July).
• CFO change — Buese out 11 May 2026, Bryan in 12 May 2026. Second CFO under Mintern, who became CEO on 1 January 2025.
• Board expanded 12 → 13 with Tony Will effective 1 July 2026.
• Preference shares cancelled by scheme of arrangement (approved 21 May 2026); balance-sheet preferred stock is now nil.
• 2026 AGM (7 May): all directors re-elected; Say-on-Pay 93.6%; Deloitte & Touche LLP ratified for FY2026.

Portfolio.

• Divestitures completed: construction accessories ($0.7bn), lawn and garden ($1.1bn), MoistureShield ($0.1bn) — $1.9bn total, generating the divestiture gains that flattered Q2 EPS by $0.16.
• Axius Water closed 29 May 2026 (~$0.7bn) — the water platform’s move up-value-chain into treatment.
• 17 acquisitions YTD for $1.4bn; net scope contribution ~$200m of EBITDA for the year.

Operating.

• Q2 revenue +5.6%, adjusted EBITDA +7%, margin +30bps; on track for a thirteenth consecutive year of margin expansion.
• Americas Building Solutions EBITDA −8%, margin −140bps, negative organic incremental margin on haulage inflation and residential weakness.
• Q1 2026 carried a $48m impairment; H1 SG&A grew 9.4% against 7.0% revenue growth.
• Weather disrupted the US South and Southeast in late May and June.
• Guidance reaffirmed; capex guidance cut $100m.

Policy.

• BUILD America 250 Act (H.R. 8870) cleared House T&I committee 62–2 on 22 May 2026 — five years, ~$580bn, ~$474.4bn contract authority, plus the first Highway Trust Fund revenue action in three decades (EV/PHEV registration fees). No House floor vote; no Senate text; IIJA expires 30 September 2026.

Headwinds, reweighted since June.

1. Capital-allocation returns dilution — promoted to the number-one risk. In June this was a watch item; the Arcosa transaction converted it into a committed, quantified $8.5bn exposure at a return below cost of capital.
2. The reauthorization cliff — demoted, but not removed. A 62–2 committee vote, a five-year framework and ~40% of IIJA money still unspent make a 2027 volume air-pocket unlikely. The residual risk is a lapse into 2028 with no multi-year bill and an unaddressed $166bn trust-fund gap.
3. Residential weakness — worse for longer. No new-build recovery expected before “the back end of '27 at best,” directly pressuring ready-mix and Americas Building Solutions.
4. Cost inflation, specifically haulage — a new, named pressure that produced a negative organic incremental margin in one segment. Management expects mitigation via surcharges through Q3–Q4; that is the testable claim.
5. Leverage and financing cost — 2.4x pro forma; cash interest already up $84m year-on-year in H1 and rising with the bridge and term loan.
6. Leadership continuity — two CFOs in thirteen months, integrating the largest acquisition in company history.
7. Valuation/expectations risk — partly discharged. The stock has de-rated ~9% since June and sits at the 83rd valuation percentile of its own history rather than the 87th. The cushion is thicker than it was, though still not generous on own-history terms.
8. European cement carbon costs (EU ETS, CBAM) — unchanged, contained, immaterial to the North American majority.

Verdict: net negative on the thesis for the first time in this coverage — but with the price already reflecting much of it. The operating changes are neutral-to-positive (pricing held, reindustrialization accelerated, funding progressed). The capital-allocation and governance changes are clearly negative and are the reason the risk ranking has been reordered. What prevents this from being a downgrade of the overall view is that the market moved first: the stock fell 8.8% against a 1.6% decline at Vulcan, which is a reasonable first approximation of the news.

9. Risk Analysis (Risk Matrix)

Aggregate read. The risk profile has reordered rather than worsened in aggregate. The macro risk that dominated the June matrix (the funding cliff) has genuinely improved and moves down; the capital-allocation risk that sat third has become the largest and is now a committed exposure rather than a trend to monitor. The moat is not at risk, the asset base is not at risk, and the balance sheet — though more levered — is sound. The probability of permanent capital impairment remains very low. What has risen is the probability of a decade of mediocre returns on a good asset base, which is a different and less dramatic risk than loss.

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation in this section. The analysis frames what the current price implies.

Where the multiple sits. At $97.15 with 665,895,636 shares, market capitalization is $64.7bn. Adding CRH-basis net debt of $15.4bn and minority interests of $1.5bn gives an enterprise value of approximately $81.6bn.

The de-rating is entirely price-driven: guidance did not move. The stock is ~9% cheaper on forward earnings than it was nine weeks ago for identical guided earnings.

A necessary data warning. The AZI valuation_index percentile series is used here for own-history direction only, because its underlying inputs do not tie to the filing. It reports book value per share of $27.05 against the 10-Q’s $36.09 (equity attributable of $24,035m ÷ 665.9m shares), producing a P/B of 3.69x against an actual 2.69x; and TTM EPS of $6.91 against a filing-derived $5.66, producing a P/E of 14.44x against an actual 17.2x. The errors are large and run in opposite directions. Every level in this article is computed from the 10-Q.

Versus its own history: less full, still not cheap. On the percentile series, CRH has moved from roughly the 87th to roughly the 83rd percentile of its ten-year valuation range (P/B 87.7, P/S 89.8, P/E 70.9). It has de-rated but remains in the upper fifth of its own history. The June structural caveat still applies and still matters: the 2023 NYSE listing was a one-time regime shift from a European-conglomerate multiple to a US-industrial multiple, so a ten-year percentile mechanically overstates expensiveness. The right read is “modestly less full than in June; not cheap on its own history.”

Versus pure-play peers: cheap, and the gap refuses to close.

The June report framed the ~40% discount as “part mispricing, part deserved,” and set a test: if the gap widened despite in-line execution, the closeable-discount thesis was falsified. CRH executed in line with its own guidance and underperformed both pure-plays. The test fired. Two consecutive reporting periods have produced no convergence.

Three readings compete, and honesty requires holding all three. (i) The market pays for purity of exposure to the best node rather than for size, diversification or reported ROIC — and Q2 handed it fresh justification, with aggregates +5% against cement −1% and a shrinking products segment. (ii) The market specifically penalised the Arcosa announcement and the buyback pause, and the relative underperformance is the tell. (iii) The discount is a genuine, persistent mispricing that will eventually close. The evidence of the last nine weeks supports (i) and (ii) over (iii).

Pro forma for Arcosa — and why nobody objected. Adding $8.5bn to EV and Arcosa’s ~$564m standalone EBITDA gives ~$90.1bn / ~$8.86bn = ~10.2x; crediting full synergies (~$739m) gives ~10.0x. The transaction is therefore roughly EV/EBITDA-multiple-neutral. That is precisely why it can be presented as sensible and why not one analyst on the Q2 call asked about the entry multiple or the return on capital. The dilution does not appear in the EBITDA multiple. It appears in return on capital (~150bps), in leverage (1.8x → 2.4x) and in the opportunity cost of the paused buyback. An investor screening on EV/EBITDA sees nothing wrong; an investor screening on ROIC sees the entire problem. That gap is the variant perception of this report.

Embedded expectations. At ~9.8x forward EV/EBITDA and ~16.7x forward earnings, the market is underwriting: mid-to-high-single-digit EBITDA growth; delivery of the Arcosa synergies; no further material deterioration in returns on capital; and — critically — no closure of the pure-play discount whatsoever. That last point is what has changed. In June, at 10.4x, some discount closure was arguably still embedded in the price. At 9.8x, after two periods of widening, it is not. The hurdle the business must clear to justify the current price is therefore materially lower than it was nine weeks ago, applied to a business whose operating performance is confirmed and whose capital allocation is one notch worse.

Scenario sketch (illustrative, not a target).

• Bear: reauthorization lapses without a Senate bill; residential weakness extends past 2027; haulage inflation persists; Arcosa synergies slip and FY2026 adjusted ROIC prints below 11%. EBITDA stalls near $8.1bn, the multiple compresses toward 8–9x as the peer discount is judged permanent, and the paused buyback removes the per-share offset. Meaningful downside from here.
• Base: a short-term extension in September followed by a multi-year bill in 2027; residential recovers late 2027; reindustrialization compounds; EBITDA reaches the $8.1–8.5bn guide and grows high-single-digits into 2027; ROIC stabilises near 11–12% post-Arcosa; the multiple holds at 9.5–10.5x. Returns track EBITDA growth plus the ~2.0% dividend, with no multiple help.
• Bull: the BUILD America 250 Act passes broadly intact; data-center and advanced-manufacturing demand compounds; Arcosa synergies beat and the buyback restarts in 2027; ROIC re-accelerates above 12.5%. EBITDA approaches $9.5bn+ with partial multiple re-rating toward 11–12x. Attractive upside — but it now requires the capital-allocation case to be rehabilitated, not merely the demand case to work.

Verdict: fairly valued, with better asymmetry than in June and a worse asset behind it. The market is pricing CRH as a good business that will execute operationally and will not be given credit for its scale, diversification or superior return on capital. That pricing is defensible given what the last two quarters showed. The embedded expectations are undemanding; the risk is no longer that the multiple is too high but that the return on the capital being deployed into it keeps falling — which would make today’s undemanding multiple the correct one indefinitely.

11. Variant Perception

Consensus view. CRH is a high-quality, US-centric building-materials compounder and a prime beneficiary of infrastructure and reindustrialization spending; the Arcosa acquisition is a strategically sound, scale-enhancing, multiple-neutral transaction that reinforces its #1 US aggregates position; and the stock remains conspicuously cheap versus Vulcan and Martin Marietta. Sell-side framing on the Q2 call was uniformly constructive — the first analyst opened with “congrats on the strong quarter,” and across eight questions not one addressed the Arcosa entry multiple, the return on invested capital, or the buyback pause. The questions covered guidance building blocks, Americas Building Solutions margins, the reauthorization, the M&A pipeline, asphalt pricing, ready-mix, cement pricing and data centers. The call ended early.

The strongest bull case. The moat is elite and just proved it — aggregates priced +5% through weather disruption, soft residential demand and cost inflation, while cement (the contestable node) priced −1%, which is exactly what the theory predicts. CRH earns a higher return on capital than either pure-play yet trades at roughly half their multiple. Roughly 40% of IIJA money remains unspent, insulating 2027, while a five-year $580bn successor cleared committee 62–2. Reindustrialization is a genuine, privately-funded second engine where CRH’s connected portfolio is uniquely advantaged — 200 active data centers, 85% within 25 miles of a facility, five product lines on a single job. Arcosa buys the only available entry into Dallas and Phoenix at a discount to listed aggregates multiples, from a seller whose aggregates arm earns a 48.1% cash gross margin. Thirteen consecutive years of margin expansion. At 9.8x forward EBITDA and a 2.0% yield, you are paid to wait.

The strongest bear case. The compounding quality is deteriorating in a now-measurable way. Adjusted ROIC fell 130bps to 12.1% in 2025 and GAAP ROIC confirms it independently (10.37% → 10.17%); management responded by committing $8.5bn — the largest cheque in company history — at ~15x pre-synergy for an asset that returns ~2.8% unlevered today and ~4.3% with every promised synergy delivered, against an ~8–9% cost of capital, implying a further ~150bps of ROIC dilution. Roughly half of the acquired business is contestable steel fabrication, not moated aggregates. To pay for it, the buyback that retired 24% of the share count is off, leverage rises to 2.4x, and the finance function has changed hands twice in thirteen months. Reported earnings quality slipped — 14% EPS growth was really ~5.7% ex-divestiture-gains — one segment posted a negative organic incremental margin, H1 SG&A outgrew revenue, and guidance was reaffirmed rather than raised. No insider bought a single share through a 25% drawdown. The peer discount has now failed to close for two consecutive periods and widened in the latest one, which is what a permanently deserved discount looks like.

The 3–5 assumptions that matter most.

1. Return on invested capital post-Arcosa. Does adjusted ROIC stabilise near 11–12%, or slip below 11%? Falsification: the FY2026 10-K adjusted-ROIC disclosure, then FY2027 pro forma for Arcosa.
2. Arcosa synergy delivery. $60m in year one, $175m run-rate by year three, against a 2.1% synergy yield on price. Falsification: the first post-close disclosure in 2027; a slip in year-one delivery would confirm the deal is value-destructive at the price paid.
3. Whether the pure-play discount is closeable or permanent. Falsification: it has now failed twice; a third period of widening on in-line execution settles the question in the bear’s favour.
4. Reauthorization outcome. Multi-year bill versus indefinite continuing resolutions into 2028. Falsification: Senate text and a floor vote before or shortly after 30 September 2026.
5. Buyback restart. Does capital return resume post-close in 2027, or does leverage keep it dark? Falsification: the FY2026 results call and the 2027 capital-allocation guidance.

Factor-positioning read — and a correction to the June framing. The June report described CRH as a “quality-compounder-at-a-price” with “quality and dividend tilts” and momentum that had rolled over. The factor data no longer support the quality half of that. In the Base model the Quality loading is +0.021 — effectively zero — and Momentum has collapsed to +0.106 (+0.043 once sectors are stripped). What remains is Market 1.169, DividendYield +0.393, OilPrice −0.375 and SmallSize +0.196; adding sectors and industries brings Home Construction +0.390, Materials +0.382, Industrials +0.292 and Infrastructure +0.207. The honest label is now a de-rated, high-beta, dividend-tilted cyclical with heavy residential-construction sensitivity and short exposure to energy costs — not a quality compounder. Alpha has flipped from +6.2% in June to −5.1%; one-year return is −12.8%; six-month and three-month Sharpe ratios are −1.07 and −0.94; the price sits below all three EMAs. These are statistical estimates from an in-sample model, not verdicts on the business — but the direction is unambiguous and it corroborates the fundamental read rather than contradicting it. This is not a falling knife (the three-year record is +20.8% annualised, the dividend is rising, the balance sheet is investment-grade); it is a de-rating that has not yet found its floor, in a name the market has stopped classifying as high-quality.

My variant take. In June I argued the real debate was the quality of the compounding (ROIC), not the multiple, and that the market was under-focused on the 2025 ROIC dip. Nine weeks later that thesis has been confirmed in the strongest available way: management committed $8.5bn at a return below its cost of capital, and the sell-side asked eight questions without touching it. The market’s reaction — an 8.8% decline against 1.6% at Vulcan — shows the price mechanism registered the problem even though the analytical discussion did not. That gap between what the tape knows and what the commentary says is the variant perception. The consensus frame (“cheap versus peers, strategically sound deal, multiple-neutral”) is accurate on every individual claim and wrong in aggregate, because it evaluates the transaction on the one metric that hides the dilution. The correct question is not “is CRH cheap versus Vulcan?” — it has been, for three years, and the gap has not closed. It is “will a business that keeps deploying capital at 4% into a 9% cost of capital ever deserve a higher multiple?” If FY2026 ROIC holds near 12% and the buyback restarts, the answer is yes and today’s 9.8x is a gift. If it prints below 11% and the buyback stays dark, the discount is not a mispricing at all — it is the market correctly pricing a value-diluting compounder, and it will persist.

12. Fact vs. Interpretation Table

13. Open Questions

1. What will FY2026 adjusted ROIC print? The single most important number CRH will disclose. Above 12% and the machine still works; below 11% and the June bear case is confirmed. Only available with the FY2026 10-K.
2. Will the Arcosa synergies deliver on schedule? $60m in year one is the first testable checkpoint, in 2027. A 2.1% synergy yield on an $8.5bn price leaves no room for slippage.
3. When does the buyback restart, and at what leverage? Management says “at a later date.” Whether that is 2027 or 2029 determines whether per-share compounding resumes.
4. Was the CFO transition performance-related, strategic, or personal? The filing gives boilerplate. The answer bears on execution confidence through the largest integration in company history.
5. Why did no analyst ask about the entry multiple or return on capital? Either the sell-side is comfortable with the arithmetic and I am wrong, or the analytical frame in this sector is EV/EBITDA-anchored to a fault. Worth testing at Q3.
6. Does Americas Building Solutions recover in Q3–Q4 as management promises via surcharges, or is the negative organic incremental margin structural?
7. Reauthorization path: short-term extension then a 2027 bill, or an extended continuing-resolution regime into 2028?
8. Maintenance versus growth capex split — still not disclosed, still needed to assess true owner earnings and the return on growth capital. Management disclosed ~$800m of growth capex through Q2, a partial and welcome improvement.
9. Does the water platform (Axius) earn aggregates-like returns, or dilute group ROIC as a lower-margin services adjacency?
10. What happens to Arcosa’s Engineered Structures business under CRH ownership — retained as a growth platform, or divested as non-core within the “active portfolio management” framework?

14. What Must Be True (Bull and Bear, with Falsification Tests)

Scorecard on the June 2026 tests, before setting new ones:

Three of four bull tests are failing or falsified; both bear tests are confirmed. The one bull test that passed — pricing power — is the most important one for the business, which is why this is a HOLD and not a downgrade.

For the BULL case to be right from here, the following must be true:

• Aggregates pricing power persists. Falsification: two or more consecutive quarters of flat or negative mix-adjusted Americas aggregates pricing.
• FY2026 adjusted ROIC holds at or above ~12%, and the FY2027 figure stabilises post-Arcosa rather than stepping down toward 10%. Falsification: an FY2026 print below 11%, or any material goodwill impairment on the $13.15bn balance.
• Arcosa synergies deliver on the disclosed schedule — $60m in year one, $175m run-rate by year three — and the acquired aggregates outgrow the price paid. Falsification: a year-one shortfall, or a quiet retirement of the $175m figure post-close.
• The buyback restarts within roughly twelve months of closing, evidencing that leverage was a timing constraint rather than a structural one. Falsification: no restart through 2027 with leverage still above 2.2x.
• Reindustrialization compounds and offsets residential weakness. Falsification: the data-center-linked backlog flattening while residential stays soft into 2028.

For the BEAR case to be right, the following must be true:

• The pure-play discount is permanent, not a mispricing. Falsification: CRH re-rates toward peers on continued in-line execution — the opposite of what the last two periods produced.
• Returns on capital keep declining as goodwill-heavy M&A outruns them, eventually forcing impairments. Falsification: adjusted ROIC re-accelerating above 12.5% with no impairments, proving the deals are genuinely accretive.
• Americas Building Solutions weakness is structural, not a haulage-and-residential timing problem. Falsification: the segment returning to positive organic incremental margin in Q3 or Q4 2026, as management promises.
• Leverage constrains capital returns for years, leaving investors with a ~2% dividend and no per-share compounding. Falsification: net debt back below 2.0x within eighteen months of closing with the buyback resumed.
• The residential trough extends past 2027, keeping ready-mix and outdoor-living demand depressed. Falsification: housing starts inflecting and management pulling the recovery timeline forward from “back end of '27 at best.”

15. Source Appendix

See Appendix B below for the full citation list.

APPENDIX A — Standard Diligence Questionnaire

A companion set of diligence questions, answered against Q2 2026 results, the Arcosa transaction and the current filing record.

General

What thoughtful questions have other investors asked about this company?

The genuinely revealing observation this quarter is what investors did not ask. Across eight analyst questions on the Q2 2026 call, the topics were: 2026 guidance building blocks; Americas Building Solutions margin weakness; the reauthorization outlook; the M&A pipeline and divestiture programme; asphalt pricing and liquid-asphalt cost; ready-mix volumes; cement pricing; and data-center exposure. Not one question addressed the Arcosa entry multiple, the return on invested capital, the buyback pause, or the increase in leverage — nine weeks after the company committed $8.5bn, the largest cheque in its history. The call ended early. The first question opened with “congrats on the strong quarter.”

The questions that are being asked well: whether the connected-portfolio model genuinely converts into wallet share on megaprojects (management’s East Texas data-center answer — five product lines, 3 million tons, two-to-three years, first-on-site — is the best concrete evidence yet that it does); whether ABS margin weakness is timing or structural; and whether mid-single-digit aggregates pricing can persist into 2027.

The questions a sceptical investor should press at Q3: (1) what is the expected FY2026 adjusted ROIC, and what is the pro forma post-Arcosa figure? (2) what unlevered after-tax return does management underwrite on Arcosa, and over what horizon? (3) what leverage level triggers the buyback restart? (4) is Arcosa’s Engineered Structures segment a keeper or a future divestiture?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Neither extreme, and the components differ. Public infrastructure is at or near a funded cyclical high — record state DOT budgets with ~40% of IIJA money still unspent. Reindustrialization (data centers, advanced manufacturing, LNG, semiconductor fabs) is in an early upcycle that management describes as “the early phase of a generation of capital investment.” Residential new-build is at a cyclical low and management now expects no recovery before “the back end of '27 at best.” So group earnings sit mid-cycle with one leg near a peak, one accelerating from a low base, and one depressed. Margin, by contrast, is at an all-time high (2026 will be the thirteenth consecutive year of expansion if guidance holds), which limits the margin-recovery lever.

Driven by the external environment or internal actions? Both, and the split is measurable this quarter. External: public funding levels, weather (which disrupted the US South and Southeast in late May and June), residential affordability, haulage and energy cost inflation. Internal: pricing execution (+5% aggregates), cost and procurement discipline, and portfolio reshaping. The internal contribution is demonstrable — margin expanded 30bps despite an inflationary cost backdrop, weather disruption and a segment posting negative organic incrementals. CFO Bryan quantified the portfolio effect at ~$200m of net incremental EBITDA for the year, i.e. ~2.5% of EBITDA, which confirms the reported growth is mostly organic and mostly price.

How stable are revenues? Transactional, not contractual — there is no recurring or subscription base. Stability comes from three sources: end-market diversification (public infrastructure, non-residential/reindustrialization, residential), geographic diversification (~75% North America, remainder Europe and Australia), and the structural inelasticity of price. Aggregates pricing has risen through recessions for ~25 years; volume is what cycles. Backlog gives six-to-nine months of visibility in Americas Materials and management reports both bidding volumes and secured work up year-on-year.

Outlook for products/services? Aggregates: mid-single-digit pricing, low-single-digit volume. Cement: low-single-digit volume, broadly flat pricing (down 1% in H1 on geographic mix, with margin still expanding). Asphalt/paving: another year of margin progression, aided by the liquid-asphalt winter-fill advantage. Ready-mix: the most residential-exposed line, volumes flat. Products/outdoor living: pressured until residential recovers. International: ninth consecutive year of positive pricing expected.

How big will this market be — growing, shrinking, domestic or international? Growing, and increasingly domestic in profit terms. The near-term US demand pool is anchored by ~40% of unspent IIJA money plus record state budgets; the medium-term pool depends on the BUILD America 250 Act (~$580bn over FY2027–2031, ~$474.4bn contract authority, cleared House committee 62–2 but not yet passed by either chamber). Layered on that is privately-funded reindustrialization, which is genuinely additive and independent of federal appropriations. Water is a ~$100bn+ annual US ecosystem, highly fragmented, regulation-driven. International is lower-growth but improving, with defence infrastructure emerging as a new category in Central and Eastern Europe.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Less, and structurally so at the aggregates node. Consolidation continues to remove independent operators (Quikrete/Summit in 2025; CRH’s own ~1,200-plus historical acquisitions; now Arcosa). New supply cannot be permitted at scale — a decade-plus permitting timeline against intensifying local opposition. This is the rare industry where Marathon’s capital-cycle mechanism is broken on the supply side: high returns do not attract offsetting capacity because capital cannot manufacture a permit or a geological deposit near urban demand.

How profitable is the business (ROIC, ROE)? FY2025 GAAP return on invested capital 10.17%; return on common equity 16.36%; management’s adjusted ROIC 12.1%. Both measures peaked in 2024 and slipped in 2025 — GAAP from 10.37%, adjusted from 13.4%. That two independent bases agree removes the “it’s a non-GAAP definition” defence. The GAAP spread over an estimated 8–9% WACC is thin to nil; the adjusted spread is comfortable but narrowing. Post-Arcosa, roughly 150bps of further group dilution is close to arithmetically unavoidable.

How profitable is the industry — how many competitors, what barriers to entry? Highly profitable at the aggregates node and structurally protected. Pure-play peers earn 28.9% (Vulcan) and 30.9% (Martin Marietta) EBITDA margins against CRH’s ~20.2% blended. The barrier is a permit plus a geological endowment near demand, neither purchasable with capital alone. Principal North American competitors: Vulcan, Martin Marietta, Heidelberg Materials, Amrize (the 2025 Holcim North America spin), Cemex, Eagle Materials, Knife River, and Quikrete (private, owner of Summit Materials).

Can the business be easily understood? Yes at the unit level — rock is quarried, crushed, sold within ~30–50 miles, and the economics are governed by freight physics. Complexity comes from scale and reporting: 4,000 locations, three segments spanning six product families across two continents, an acquisition cadence of ~38–40 deals a year, and a non-GAAP adjusted-EBITDA presentation that requires reconciliation. The Q2 2026 quarter is a case study — a headline “net income +13%” that is really ~6% once divestiture gains are removed.

Can it be undermined by foreign low-cost labour? Not for aggregates — the product cannot be shipped economically and the production is local, capital-intensive and increasingly automated. Cement is partly exposed: it is globally traded and importable at coastal markets, which is a principal reason cement priced −1% while aggregates priced +5% this quarter. Arcosa’s Engineered Structures (steel fabrication for utility and wind structures) is the most import-exposed piece of the pro forma group.

Do brands matter? Marginally. Management cites “over 200 leading brands at a regional level” across the US, but the buying decision for a contractor is driven by landed cost, availability, technical specification and reliability — not brand equity. Where reputation genuinely matters is in integrated megaproject bidding, where speed, quality and single-source capability win work: CRH’s ability to be “first in” on a data-center site and supply five product lines over two to three years is a relationship and capability advantage, not a brand advantage.

What is the nature of competition? Local and oligopolistic in aggregates — competition for a job within a freight radius, where the nearest quarry usually wins on landed cost. National and bid-based in paving and infrastructure services. Genuinely competitive and shippable in manufactured products, which is where Americas Building Solutions posted a negative organic incremental margin this quarter.

Customers’ switching costs? Geographic rather than contractual. A customer can switch supplier freely on paper; the constraint is that there is often no economically viable alternative quarry within range. That is a stronger form of captivity than a contract, because it does not expire. In the integrated model, switching costs rise further: a customer using CRH for site stabilization, aggregates, ready-mix, drainage and paving on one project would have to re-tender five relationships to move.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Yes, and materially. Aggregate reserves — ~23.8bn tons across 1,334 mining properties — are carried at historical cost, much of it acquired decades ago, and are almost certainly worth a large multiple of book. Permits themselves have enormous economic value and no balance-sheet recognition; the Arcosa transaction is in part CRH paying cash for permits it could not obtain in Dallas and Phoenix. Long-standing local market positions and the liquid-asphalt storage network (roughly half of annual requirement, built over decades) are similarly unrecognised.

Off-balance-sheet liabilities? Nothing unusual. Operating leases are capitalised under current GAAP ($1,285m right-of-use assets; $258m current and $1,069m non-current liabilities). Pension obligations are small. Quarry reclamation and restoration obligations sit within other non-current liabilities and are a real long-term cash claim in an extractive business, though modest relative to cash generation. The $3.25bn bridge facility and $2.5bn term loan for Arcosa are committed but largely undrawn — a future on-balance-sheet obligation of ~$8.5bn at closing.

How conservative is the accounting? Mixed, and slightly less conservative than the June assessment. Positives: no restatements, a clean auditor ratification (Deloitte & Touche LLP, ratified 2026-05-07), small and non-distorting share-based compensation, and management’s own explicit disclosure of the $0.16 divestiture gain inside Q2 EPS — a genuinely helpful piece of candour. Negatives: $13.15bn of goodwill (22% of assets) with no material impairments despite a two-year decline in return on capital; recurring “gains on disposal of long-lived assets” running through operating income ($52m in Q2, $74m in H1); a $48m impairment in Q1 2026; and an adjusted-EBITDA presentation that requires the reader to do reconciliation work. The headline acquisition multiple being quoted including three-year-forward synergies is a presentational choice that flatters, even though the footnote is accurate.

How CapEx-hungry is the business? Substantially. FY2026 capex guidance is $2.7–2.9bn (trimmed from $2.8–3.0bn “to reflect project timing and lower maintenance spend”), roughly 7% of revenue, against H1 depreciation, depletion and amortization of $1,124m. Management disclosed ~$800m of growth capex through Q2 — a partial and welcome improvement on the maintenance/growth split that the June report flagged as missing — implying roughly half the total is maintenance. A quarrying and cement business is inherently capital-intensive; the offsetting feature is that the assets are extremely long-lived and the reserves deplete over 60–70+ years.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? FY2025 operating cash flow was $5.6bn against $2.7bn of capex, roughly $2.9bn of simple free cash flow. H1 2026 is seasonally negative (CFO $513m less $1,240m capex = −$727m) because of construction-season working-capital build; note that the reported H1 CFO decline from $719m is largely a non-cash artefact — the reconciliation deducts $334m of divestiture gains whose cash appears in investing. The stated philosophy is ~$40bn of five-year financial capacity allocated ~70% to growth and ~30% to shareholder returns, targeting >100% average adjusted free-cash-flow conversion by 2030.

Significant acquisitions recently? Yes — the largest in company history. Arcosa, $8.5bn enterprise value, agreed 21/22 June 2026, all cash, closing expected Q1 2027. Also in 2026: Axius Water ($0.7bn, closed 29 May) plus 16 other deals totalling $1.4bn year-to-date. Prior years: Eco Material ($2.1bn, 2025), Hunter cement/ready-mix (~$2.1bn, 2024), Adbri majority stake (2024). Offsetting divestitures of $1.9bn in 2026 (construction accessories, lawn and garden, MoistureShield).

The Arcosa price is the central capital-allocation question of this update. The disclosed “11.5x 2026E adjusted EBITDA” is stated including $175m of year-three run-rate synergies; the true entry multiple is ~15.1x (reversing CRH’s own arithmetic) or 15.4x (against Arcosa’s reported TTM adjusted EBITDA of $551.3m). On Arcosa’s TTM operating income of $322.7m taxed at CRH’s guided 26%, the unlevered return is ~2.8% as-is and ~4.3% with every promised synergy delivered, against a cost of capital near 8–9%. Roughly half of what is being bought — Engineered Structures, $301.7m of Q2 revenue — is contestable steel fabrication with no permit moat.

Buying back shares? Not currently. The buyback is paused. The last tranche completed 28 July 2026, taking year-to-date repurchases to $0.7bn; no new tranche has been initiated, explicitly “in connection with the agreement to acquire Arcosa,” to be reevaluated “at a later date.” This is a material change: since 2018 the programme returned ~$10bn and retired ~24% of the share count (785.1m shares in 2020 to 665.9m today), and it was one of the pillars of the prior capital-allocation grade. CRH has, in effect, stopped buying its own equity at ~9.8x forward EBITDA in order to buy someone else’s at ~15x.

Issuing large amounts of new shares to insiders? No. Share-based compensation is small and non-distorting. The 13 May 2026 RSU settlements across the board and executive team were routine and modest in size (CEO Mintern 15,194 shares settled, 7,932 withheld for tax). Share count continues to fall, not rise — 665,895,636 outstanding at 30 June against 668,630,350 at year-end.

Compensation policy of directors/management? Annual bonus driven by diluted EPS, cash flow and RONA (return on net assets); long-term incentive on relative TSR with a sustainability modifier. RONA and relative TSR are the right anchors for a roll-up and discourage pure empire-building. The gap is that RONA is an internal net-assets measure rather than an explicit ROIC-versus-WACC hurdle — and a transaction returning 2.8–4.3% unlevered is exactly the kind an explicit WACC hurdle would flag and an EPS/RONA framework may not, particularly since debt-funded acquisitions of profitable businesses are near-automatically EPS-accretive when the after-tax cost of debt sits below the asset’s earnings yield. Say-on-Pay support slipped to 93.6% at the 2026 AGM from 94.6% the prior year.

Motivations of management? Growth and scale, expressed through consolidation, with returns as a stated but secondary constraint. The language on the Q2 call is revealing: CRH describes itself as “the leading compounder of capital in our industry” and highlights that combined CRH-plus-Arcosa 2026E EBITDA would exceed “the next 4 U.S. peers combined.” Scale is presented as an end as well as a means. CEO Mintern (since January 2025) has now overseen two CFO appointments — Nancy Buese departed “by mutual agreement” on 11 May 2026 after roughly twelve months, succeeded by 14-year insider Aylwyn Bryan. Nothing suggests an accounting or control problem, but the finance function has turned over twice while the company committed its largest-ever cheque.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? CRH plc is an Irish-incorporated company whose ordinary shares trade directly on the NYSE under “CRH” following the September 2023 primary-listing move; it is an S&P 500 constituent and files US domestic forms (10-K, 10-Q, 8-K, DEF 14A) with the SEC. It is not an MLP and issues no K-1. Some data vendors still label it “CRH PLC ADR,” which is a legacy artefact. Nine series of guaranteed notes are separately NYSE-listed. Irish dividend withholding tax mechanics may apply depending on holder domicile — a jurisdictional matter for the investor, not a business issue.

Dividend policy? Consistent long-term growth. The quarterly dividend was raised 5% to $0.39 per share on 30 July 2026 (payable 16 September to holders of record 14 August), giving ~$1.56 annualised and a yield of roughly 2.0% at $97.15 — up from ~1.5% in June, purely because the price fell. H1 dividends paid were $521m. With the buyback paused, the dividend is currently the only channel of shareholder return.

How profitable is the business? FY2025: gross margin 36.1%, adjusted EBITDA margin 20.5%, operating margin 14.2%, net margin 10.0%, ROE 16.4%, GAAP ROIC 10.17%. Q2 2026 adjusted EBITDA margin 24.4% (seasonal peak). Profitable and improving at the operating line — 2026 would be a thirteenth consecutive year of margin expansion — while flat-to-declining on capital returns.

Is net income diverging from cash from operations? Yes this half, and the reason matters. H1 2026 net income was $1,331m against operating cash flow of just $513m — a striking gap. It is not a receivables or earnings-quality problem in the usual sense: DSO was 57.2 days at Q2 2026 versus 57.9 days a year earlier, so collections are fine. The gap is dominated by the reconciliation’s deduction of $334m of divestiture gains (versus $12m in H1 2025), whose cash counterpart appears in investing as $1,676m of divestiture proceeds, plus seasonal working-capital build. The genuine cash deteriorations are cash interest ($335m vs $251m) and cash taxes ($404m vs $304m). Over a full year CRH’s operating cash flow has consistently exceeded net income by roughly 1.5x, reflecting heavy D&A — the FY2026 figure, not the half, is the meaningful test.

Risks & Downside

What factors would cause the stock to decline? In rough order of current probability-weighted impact: (1) an FY2026 adjusted ROIC print below 11%, confirming that M&A is outrunning returns; (2) Arcosa synergy slippage or an integration stumble under a new CFO; (3) a reauthorization lapse extending into 2028 without a Senate bill; (4) residential weakness persisting past 2027, keeping ready-mix and outdoor-living depressed; (5) haulage and energy inflation continuing to outrun pricing, as it did in Americas Building Solutions this quarter; (6) a goodwill impairment on the $13.15bn balance; (7) a general construction downturn (beta 1.27, three-year max drawdown −27%); (8) further widening of the pure-play discount as the market concludes it is deserved.

Risk of a catastrophic loss? Low. The business is asset-heavy and diversified across products, end-markets and two continents, with investment-grade credit, $4.5bn of undrawn committed facilities to 2030, and reserves representing 60–70+ years of extraction. Pro forma leverage of 2.4x is the top of CRH’s recent range but is not stressed. The realistic bad outcome is not impairment of capital but a decade of mediocre returns on a good asset base — value dilution rather than value destruction.

Chance of a total loss? Effectively nil absent gross mismanagement or fraud. The underlying assets — permitted quarries near growing urban demand — retain value under almost any operating scenario, and are the very thing competitors cannot replicate.

Recent News & Events

Has the business environment changed recently? Yes, in three directions simultaneously. Better: infrastructure funding visibility improved (BUILD America 250 Act, ~$580bn over five years, cleared House Transportation and Infrastructure committee 62–2 on 22 May 2026, with ~$474.4bn of contract authority and the first Highway Trust Fund revenue action in three decades), and reindustrialization demand accelerated materially — 200 active data-center projects, 85% of announced US data centers within 25 miles of a CRH facility, one East Texas job consuming 3 million tons of aggregates across five product lines. Worse: residential new-build recovery pushed to “the back end of '27 at best”; haulage cost inflation produced a negative organic incremental margin in Americas Building Solutions. Unresolved: the IIJA expires 30 September 2026 with no House floor vote and no Senate text, so a short-term extension is the realistic base case.

Significant acquisitions? Arcosa ($8.5bn, agreed 21/22 June 2026, closing Q1 2027, stockholder vote 4 September 2026) and Axius Water ($0.7bn, closed 29 May 2026), plus 15 further bolt-ons. Offset by $1.9bn of divestitures. Financing: a $5.75bn bridge signed 22 June, partly termed out on 17 July into a three-year $2.5bn term loan, reducing bridge commitments to $3.25bn.

Change in accounting policies? None disclosed. Deloitte & Touche LLP was ratified as auditor for FY2026 at the 7 May AGM without incident. The 5% and 7% “A” cumulative preference share classes were cancelled by scheme of arrangement approved 21 May 2026, simplifying the capital structure — preferred stock is now nil on the balance sheet.

Recent changes — new markets, facilities, management? New markets: Dallas and Phoenix aggregates, via Arcosa — previously “geographic white spaces” (CEO Mintern’s phrase) that CRH could not permit its way into; plus water treatment and quality via Axius. Management: Aylwyn Bryan became CFO on 12 May 2026, succeeding Nancy Buese, who stepped down by mutual agreement on 11 May after roughly twelve months — the second CFO under a CEO who took office in January 2025. Board: expanded from 12 to 13 with W. Anthony (Tony) Will, former CF Industries chief executive, effective 1 July 2026. Capital returns: the share buyback was paused in connection with the Arcosa agreement; the dividend was raised 5%. Safety/regulatory: an MSHA Section 107(a) imminent-danger order was issued at the Okay Quarry, Oklahoma on 28 April 2026 (an employee standing on a handrail without fall protection); no injury occurred and the order was terminated — disclosed for completeness of the 8-K record, financially immaterial.

APPENDIX B — Source Appendix

Sources relied on for this update. The durable industry and moat analysis carried forward from my 2026-06-14 report, whose own source appendix remains the reference for those sections.

All URLs accessed 2026-08-15 unless otherwise noted.

A. Primary — CRH SEC filings

B. Primary — Arcosa, Inc. (NYSE: ACA) SEC filings

C. Management commentary — earnings call

D. Market, quantitative and factor data

E. Public policy sources

Verification note on the public-policy sources. An initial search result headline implied the full House had passed the BUILD America 250 Act. The underlying sources establish only House Transportation and Infrastructure Committee approval (62–2, 2026-05-22), with no floor vote and no Senate text as of this report’s date. This article states the corrected position. This correction is recorded here because the distinction is material to the infrastructure-funding argument.

F. Secondary — market reaction

G. Prior coverage

H. Data limitations and caveats

1. ROIC.ai news feed returned no results. get_company_news returned an empty array for CRH, a $65bn NYSE issuer — a recurrence of a documented tool failure. The recent-events timeline was therefore built from the 8-K record, the earnings release, the earnings-call transcript and targeted public sources rather than from a news feed.

2. AZI valuation_index input errors. As noted at item D.20, AZI’s book-value-per-share and TTM-EPS inputs for CRH do not tie to the Form 10-Q, and the two errors run in opposite directions. Percentiles are used for own-history direction only; every valuation level here is computed from the filing.

3. AZI price-series artefact. The daily series carries stale placeholder rows for 2026-07-21 and 2026-07-22 (zero volume, flat $99.95 close). These were excluded from the price-move attribution in the five-year event map.

4. FY2026 adjusted ROIC is not yet disclosed. It is an annual metric available only with the FY2026 Form 10-K. The prior report’s central falsification test therefore cannot be settled within this update; this article relies on the FY2025 figures, the independently computed GAAP series, and forward arithmetic from the Arcosa transaction’s disclosed economics.

5. Arcosa’s 2026E adjusted EBITDA is derived, not disclosed. Arcosa suspended guidance on 2026-08-05 because of the pending merger. The ~$564m standalone figure used in this article is reverse-engineered from CRH’s own disclosed 11.5x post-synergy multiple and cross-checked against Arcosa’s reported TTM adjusted EBITDA of $551.3m. Both derivations are shown explicitly in the Capital Allocation section.

6. Management claims labelled as such. The $175m Arcosa synergy figure, the “~600 basis points of margin improvement within the first 3 years post acquisition” track record, the ~$40bn five-year financial capacity, and the 2030 financial targets are management representations that cannot be verified from filings. They are labelled as assumptions in the Fact vs. Interpretation table and no conclusion here rests on them.

7. Cost-of-capital estimate. The ~8–9% WACC used in the Arcosa returns test is an estimate, not a disclosed figure. The conclusion — that the transaction returns ~2.8% unlevered as-is and ~4.3% at full synergies — is computed from disclosed figures and is insensitive to reasonable variation in the WACC assumption, since both returns fall well below any plausible range.