ACM › analyze
ACM · Analyze
Price at analysis: $66.24 | 52wk range $60.35–$135.52 (−49% off high, −47% trailing 12mo) | Mkt cap $8.52B | Beta 0.92
0. Knowledge check
python .mcp/kb.py find ACM → no matches, nothing known on this name — researched fresh.
No live sweep in Knowledge/INDEX.md covers engineering & construction / infrastructure
consulting. Playbook pitfalls checked before trusting vendor numbers below: the ADR-share-count
and currency traps do not apply (ACM is a US domestic issuer, USD-reporting, single share
class) — but two other live pitfalls fired directly on this name and materially changed the
read (flagged inline): pitfall-vendor-forward-eps-is-the-wrong-fiscal-year and the general
mechanism behind pitfall-franchisor-revenue-includes-pass-through-reimbursements (a cost
pass-through inflating the GAAP revenue base — AECOM's own analogue, not yet in the Playbook
under this ticker). Both are worth a note; left for the orchestrator per scope lock.
1. What happened — why it's near a 52-week low
AECOM peaked near $135 in autumn 2025, then fell in three distinct legs:
- Nov 2025 — Q4/FY2025 print paired with an announcement that AECOM would review strategic alternatives (including a possible sale) for its Construction Management (CM) business, alongside a raised long-term target (segment adj. operating margin to 20% by FY2028, from 17%; adj. EPS CAGR to 15% for FY26–29). Stock fell ~$130 → ~$95–100 on the uncertainty despite the raised targets.
- Dec 2025 – Jun 2026 — Q1 (Feb) and Q2 (May) FY2026 prints were genuinely strong on
headline metrics: **record backlog every quarter, 22nd consecutive quarter with book-to-burn
1.0x, segment adj. operating margin at record highs (16.5–16.8%), guidance raised twice. The stock kept sliding anyway (~$97 → ~$84 → ~$69), which in hindsight reads as the market pricing in the unresolved legacy-project risk that had not yet been quantified. CEO, President and CFO all bought stock on the open market** at $70.63–$71.12 in May–Jun 2026 (confirmed via
get_holder_info insider_transactions— real cash purchases with dollar values, not RSU vesting or option exercises, distinguishing this from thepitfall-yahoo-insider-purchases- counts-rsu-grantstrap). - Aug 12, 2026 — Q3 FY2026, the confirming shock. A $337M pre-tax charge on a legacy (bid in Dec 2018) design-build P3 Construction Management project: subcontractor productivity delays in the final testing/commissioning phase pushed completion from Q1 FY2027 to Q2 FY2027. This sits in continuing operations (distinct from a small, already-wound-down "self-perform at-risk construction" line that has been in discontinued ops since FY2020 and shrank further this quarter, −$2.9M vs. −$43.9M a year ago). GAAP: revenue $3.586B (−14% YoY, still well above consensus on revenue), operating loss −$76M, net loss −$84M, diluted loss per share −$0.65 (continuing ops). NSR (AECOM's own non-GAAP net-service-revenue base, ex pass-through subcontractor costs) fell 16% YoY to $1.6B for the quarter.
Guidance cut: FY2026 adjusted EPS to $3.95–4.15 (from a pre-charge $5.90–6.10 — a $1.99/share hit) and FCF to ~$300M (from ~$400M), reflecting ~$500M of incremental cash outflow through 1H FY2027 to finish the two legacy projects. Share buybacks are paused near-term ("resume once leverage returns to historical levels" — management, Q3 call) even though the $1B authorization remains in place.
Management's framing: isolated. AECOM stopped bidding design-build P3 CM work "many years ago"; no similar project remains in the portfolio; the second legacy project is "on track" with a combined $600–650M claims position against both projects; underlying Design business grew (Americas design +6% adj. for workdays, International NSR +4%, Australia backlog +40%+, UK AMP8 accelerating, data-center/hyperscaler work "one of the fastest-growing" areas); total backlog hit a record $27.8B, +13% YoY, book-to-burn 1.6x.
Sell-side reaction: seven analysts cut price targets 12–17% in the week after (Argus $85→$75, Baird $73→$65, RBC $105→$90, Citi $97→$84, Truist $102→$85, KeyBanc $101→$79, Barclays $90→$73) but none downgraded the rating — consensus stayed "Strong Buy" (avg 1.46/5, 12 analysts), mean target $86.92 (+31% above spot), median $87.50, range $65–106. Short interest is modest (5.0% of float, 2.6 days to cover) — real skepticism exists but it is not a crowded short.
Open question — flag, don't resolve: sources disagree on whether the CM business is (a) still fully consolidated in continuing operations (which is what the Q3 charge implies — it hit continuing-ops operating income directly) or (b) formally held-for-sale/discontinued as the Nov 2025 release suggested "beginning with Q1 results." I could not get a clean primary-source read (SEC EDGAR blocked WebFetch; secondary sources conflict). Treat the CM segment's balance-sheet and P&L classification as unconfirmed and check the actual 10-Q at the next pass.
Read: sentiment/event-driven, not structural — with one real open risk. The charge is a bounded, dollar-quantified, one-time hit on a discontinued underwriting practice, landing on a company whose backlog, margin trajectory, dividend growth and insider buying all pointed the same direction throughout the decline. The genuine unresolved question is whether "isolated" holds through the second project's completion (Q1–Q2 FY2027) — a repeat would flip this from event-driven to structural.
2. Fundamentals
Vendor caveat up front: fin.py's TTM/derived figures for margin and per-share growth are
distorted by the just-booked charge; annual (FY-end Sept 30) figures below are cleaner and are
what the CAGRs use.
| Metric | FY2022 | FY2023 | FY2024 | FY2025 |
|---|---|---|---|---|
| Revenue (GAAP, incl. pass-through) | 13.15B | 14.38B | 16.11B | 16.14B |
| Net income | 310.6M | 55.3M | 402.3M | 561.8M |
| Diluted EPS | 2.18 | 0.39 | 2.95 | 4.21 |
| OCF | 713.6M | 696.0M | 827.5M | 821.6M |
| FCF | 576.6M | 590.4M | 707.9M | 684.9M |
| Buybacks | −473.0M | −379.3M | −478.5M | −388.4M |
| Dividends paid | −63.3M | −96.2M | −115.2M | −133.6M |
| Diluted shares | 142.70M | 140.11M | 136.45M | 133.31M |
CAGRs (3yr, FY22–25): Revenue +7.1% · Net income +21.8% (flattered by FY2023's own depressed, low-quality base — FY23 NI was itself charge-affected) · FCF +5.9% · OCF +4.8% · diluted shares −2.2% (buybacks a real per-share tailwind, though now paused near-term).
⚠️ The GAAP revenue base is ~2x too large for margin purposes — the same mechanism as
pitfall-franchisor-revenue-includes-pass-through-reimbursements, applied to an engineering &
construction firm instead of a hotel franchisor. AECOM's GAAP revenue includes subcontractor
and other pass-through costs at ~zero markup; its own disclosed metric, Net Service Revenue
(NSR), is the real economic base — guided at $7.3–7.7B for FY2026, roughly 46–48% of the
~$16B GAAP revenue run-rate. Consequences:
| Basis | Gross margin | Operating margin |
|---|---|---|
| GAAP revenue (what vendors show) | 6.7–7.5% | 5.7–6.6% (FY25 clean) / −1.8% TTM (charge-distorted) |
| NSR (the real base) | not disclosed directly | 16.5–18.0% segment adj., target 20% by FY2028 |
A margin screen or cross-sector comparison built on the GAAP-revenue figure understates AECOM's true profitability by roughly 2.5×, in the same direction and for the same structural reason as the lodging-franchisor case. This deserves a Playbook note (scope-locked here — flagging for the orchestrator).
Capital allocation (FY2025): buybacks $388M, dividends $134M, acquisitions $314M (up sharply from $27–74M/yr in FY22–24 — a larger bolt-on/advisory push), debt roughly flat (issued $3.40B, repaid $3.27B — mostly revolver churn, not net delevering or releverage). Now disrupted: buybacks are paused near-term post-charge; capital is being directed to finishing the legacy projects (~$500M through 1H FY2027) and to dividends/organic growth.
Debt & leverage: D/Assets 26.5% (fin.py) · net debt/EBITDA 0.94x, total debt/EBITDA 2.69x, EBITDA/interest 6.5x (roic.ai, FY2025) — genuinely comfortable coverage despite an optically alarming GAAP D/E of ~139%, which is an artifact of a thin GAAP equity base (heavy buybacks + $3.70B goodwill on $2.49B total equity), not distress. $2B undrawn revolver capacity. Interest expense is guided up $30–35M YoY in FY2027 on higher balances and the project cash burn.
Per-share (FY2025, clean): Revenue/share $121 · FCF/share $5.14 · BVPS $17.04 (fin.py) / $18.91 (roic.ai) · Tangible BVPS is NEGATIVE, −$10.56 (roic.ai) — essentially all book value is goodwill/intangibles from past M&A (URS 2014, Hunt, and the FY25 step-up in acquisition spend). This matters directly for Graham's model below.
FY2026 will look worse per-share than the trend implies: guided FCF ~$300M (~$2.3/share) vs. the $5.14 TTM figure — a real, temporary compression from the ~$500M legacy-project cash drag, not a change to the underlying FCF-conversion algorithm (which management reaffirms ex-charge).
Dividend-grower overlay (light weight — yield is only 1.9%, this is a growth name that also
pays, not an income name) - Quarterly dividend: $0.15 (FY22) → $0.18 → $0.22 → $0.26 → $0.31 (FY26, declared Jul 2026 — after the operational trouble was known internally). ≈20% dividend CAGR over 4 years. - Payout ratio 42% of GAAP TTM EPS (fin.py) is inflated by the charge-depressed denominator; on FCF/share the payout is closer to ~24% — well covered, room to keep growing even through this rough patch, which the July raise itself corroborates.
3. Moat & Competitive Advantage
Quantitative base. ROIC 14.2% FY2025, up from 12.5% FY2024 (roic.ai) — real, improving capital efficiency, consistent with the decade-long pivot from lower-margin at-risk construction toward higher-margin Design/Advisory work. The NSR-based margin trend (16.5% → 16.8% → 18.0% ex-charge, targeting 20% by FY2028) is rising, not compressing — the standard "margin compression = moat erosion" read would be wrong here if applied to the GAAP-revenue-denominated figure; on the correct (NSR) denominator, the moat signal is intact and improving.
Adversarial stress-test. "You are a well-funded rival — Jacobs, WSP, Stantec, Arcadis, or a PE-backed roll-up. How easily do you attack this?" Pure engineering/design work is a genuinely competitive, RFP-driven, fee-percentage market — a rival can and does bid the same contracts, and there is no network effect protecting AECOM here. What does protect the incumbent: (1) switching costs — program-management mandates on decades-long infrastructure programs (transit authorities, DOTs, water utilities, defense installations) are costly and risky for a client to re-compete mid-stream; (2) scale as an entry barrier — bonding capacity, security clearances, and a multi-decade track record are required to even qualify for megaproject bids (a $10B rail program is simply not biddable by a small firm), and AECOM is consistently ranked top-2/3 globally by ENR revenue. This is a moderate, not exceptional, moat — real barriers to entry on the megaproject tier, but genuine price competition on everything below it.
Disruption vectors.
1. AI/productivity deflation in professional-services labor — the exact mechanism recorded in
pattern-headcount-revenue-divergence-tests-ai-deflation (IT services/BPO being repriced on
whether AI productivity gains are kept by the provider or given back to clients). AECOM is a
51,000-employee, labor-based engineering/design firm and is exposed to the same dynamic over a
5–10yr horizon. I could not run the test today — AECOM doesn't disclose headcount growth
cleanly against NSR growth in what was pulled, so this is a genuinely open question, not a
dismissed one. Worth checking at the next print.
2. Public infrastructure capex cycle / political risk — less than 50% of IIJA funding spent in
core US markets is a multi-year tailwind, but a durable one only as long as funding priorities
don't shift; Middle East hospitality/tourism work is currently a real, if likely temporary,
headwind from regional conflict.
3. The now-proven tail risk of legacy at-risk/design-build contracts — a structural feature of
the E&C industry (multi-year projects underwritten years before problems surface), not unique to
AECOM. Management's policy change (no more design-build P3 CM bids) reduces the forward risk
but does not retroactively de-risk projects already on the books — the second legacy project is
the live test of whether that risk is truly bounded.
Evergreen assessment. Infrastructure spending (transportation, water, environmental, defense, data centers) is a permanent, recurring government/private function — genuinely evergreen at the industry level. At the company level, AECOM's specific position is durable but moat-lite: a well-run, top-tier share of a competitive market, not a dominant franchise. Rate it a good compounder with real but moderate barriers, not an unassailable moat.
4. Valuation
⚠️ Vendor forward-multiple trap — fired directly on this name. fin.py's snapshot showed
PE(fwd) 10.36 (implied forward EPS $6.40) alongside PE(ttm) 22.92 (implied TTM EPS $2.89).
Pulling the raw Yahoo fields resolves it exactly per pitfall-vendor-forward-eps-is-the-wrong-
fiscal-year:
| Yahoo field | Value | What it actually is |
|---|---|---|
trailingEps / trailingPE |
$2.89 / 22.92x | Stale TTM — doesn't fully reflect the Q3 charge yet |
epsCurrentYear / priceEpsCurrentYear |
$4.00 / 16.56x | The true current-FY (FY2026) read — matches management's own post-charge guidance midpoint ($3.95–4.15) almost exactly |
forwardEps / forwardPE |
$6.40 / 10.36x | FY2027 (next fiscal year), not "cheap now" — a normalized, post-recovery estimate |
The number to trust for "how expensive is it right now" is 16.6x current-FY earnings, not 10.4x. The 10.4x figure is real and useful, but it describes next year's normalized earnings power, contingent on the charge staying isolated — not today's multiple. Both directions of this pitfall are visible on the same ticker in the same pull, which is itself the value of the check.
Graham's Intrinsic Value √(22.5 × EPS × BVPS): using clean FY2025 EPS $4.21 and BVPS
$17.04 → ≈$40; using a blended current-year EPS ~$4.00–4.50 and BVPS $17–19 → range
~$39–45. Weight this low. Graham's formula assumes a hard-asset floor; AECOM's tangible
book value is negative (−$10.56/share) — almost the entire $17–19 BVPS is goodwill from past
M&A, not liquidatable asset backing. This is an asset-light people-and-backlog business, exactly
the type the framework says to de-weight Graham for.
DDM (light weight — 1.9% yield, this is a total-return name, not an income name): D1 ≈ $1.35 (10% dividend growth, conservative vs. the historical ~20%), r = 9% (mid-cap industrial, beta 0.92), g = 6% terminal → V = 1.35 / 0.03 ≈ $45. Consistent with the Graham floor.
DYT: current yield 1.9% vs. an estimated ~0.8–0.9% a year ago (price ~$120–130, div
~$1.04/yr then) — yield has roughly doubled, which reads "attractive" on a mechanical DYT basis.
But per pitfall-dyt-inverts-when-price-caused-the-yield, this rise is almost entirely
price-driven — the dividend grew ~19% while the price fell ~48%. DYT here is a question
("has the market overpunished the price relative to the dividend's own trajectory?"), not an
independent buy signal, and at 1.9% yield it was never the primary lens for this name anyway.
Bogle expected return (dividend yield + earnings growth ± P/E change), scenario range: - Bear (multiple stays compressed, growth undershoots to 8%): 1.9% + 8% + 0% ≈ 10%/yr - Base (growth ~12%, multiple flat): 1.9% + 12% ≈ 14%/yr - Bull (management's own 15% adj-EPS CAGR holds AND the FY2027 multiple normalizes from today's 10.4x toward a more typical mid-teens multiple over ~2 years as "isolated" gets proven out): 1.9% + 15% + a multi-year re-rate spread ≈ low-to-mid 20s%/yr for a couple of years
Earnings-power fair value (the primary lens for this name): applying a 13–16x multiple — a deliberate discount to AECOM's own pre-selloff historical forward band (it traded above 20x forward near its $135 peak), reflecting a real, not-yet-fully-resolved execution risk — to a haircut FY2027 consensus of $6.00–6.40 (vs. the raw $6.40 consensus) gives ≈$78–102.
Named tension, stated honestly: Graham/DDM (hard-asset + dividend math) say fair value is $40–50 — below today's $66 price. The earnings-power/backlog case says $78–102, and the sell-side consensus ($65–106, mean $86.92) sits closer to that end. ACM is not a statistically cheap value stock by hard-asset or dividend-discount math even after a 49% drawdown; it is a potential earnings-power mispricing, conditional on the charge staying isolated. Given the company's type (asset-light professional-services compounder with a real but moderate moat, not a mature dividend payer), the framework weights Fundamentals + Moat + Sentiment (the earnings-power case) over the static Graham/DDM floor — but the floor is real and should temper conviction, not be ignored.
Blended fair value range: $70–95, weighting the earnings-power case ~65%, the Graham/DDM floor ~15% (pulling the low end down), and the analyst consensus as a corroborating sanity check (~20%, not an independent model).
Trim convention — set on FY2027 consensus explicitly, not the raw vendor field. Trim 18x
fwd — "fwd" here must be read as FY2027 consensus EPS (~$6.40), i.e. Yahoo's forwardEps
field, cross-checked against the quarter-sum/company guide — never the vendor's epsCurrentYear
field (that one is the current-year multiple, 16.6x, and would badly overstate the trim level if
substituted in). 18x is chosen as a real but disciplined ceiling — well below the >20x the stock
carried at its euphoric $135 peak, appropriately conservative per the framework's "rather miss an
opportunity than overpay" bias.
5. Synthesis & Weighted Verdict
Tensions identified: 1. Graham/DDM (~$40–50) vs. earnings-power/consensus (~$78–106) — a genuine 2x spread depending on whether AECOM is read as an asset-backed value name (it is not, tangible book is negative) or a temporarily-mispriced compounder (the backlog/margin/insider-buying evidence supports this read, but it is a judgment call, not a certainty). 2. Headline GAAP metrics (revenue, margin, D/E) all read worse than the real economics (NSR-based margin, net leverage) — this cuts in AECOM's favor once corrected, the opposite of the usual direction these pass-through distortions run. 3. Backlog/margin/dividend/insider signals were all bullish throughout the Nov 2025 – Jun 2026 decline, yet the stock kept falling until the Aug 2026 charge confirmed a real (if bounded) problem — the market was arguably right to discount the stock ahead of the print, which is a reason for some humility about calling this simply "overreaction now that it's confirmed."
Weighting rationale: AECOM is a growth-oriented, asset-light professional-services compounder with a real but moderate moat (switching costs + megaproject scale barriers, not a dominant franchise) and a genuine, well-quantified, one-time charge rather than an ongoing structural problem — as long as the second legacy project stays "on track." Per the framework, this profile weights Fundamentals (corrected NSR-basis margins, improving ROIC, dividend growth sustained through the crisis) + Sentiment (insiders bought before the drop finished, consensus ratings held through target cuts) over the static Graham/DDM value floor, which is real but not the primary lens for a business this asset-light.
Verdict: ACCUMULATE, conviction 6.5/10. The setup — record backlog, improving true margins, sustained dividend growth, and open-market insider buying, all against a 49% drawdown driven by a single bounded, quantified, discontinued-practice charge — is the "quality on sale" pattern the screen flagged it for. Conviction is held at 6.5 rather than higher because: (a) the stock is not cheap on any hard-asset or income basis, only on a forward-earnings-power basis that assumes clean execution from here; (b) the CM segment's accounting classification (held-for-sale vs. consolidated) is unresolved in what I could source; (c) the AI-deflation risk to labor-based engineering services is a real, untested question over the medium term; and (d) buybacks — a real per-share tailwind historically — are paused indefinitely.
What would raise conviction: a clean Q4 FY2026 print (~Nov 17, 2026) with the second legacy project confirmed on track and no new charge; clarity on the CM strategic-alternatives process; any signal on buyback resumption timing.
What would break the thesis: a second charge on the remaining legacy project (would flip this from event-driven to structural — the "isolated" claim would no longer hold); a break in the backlog/book-to-burn trend (>1.0x for 22 consecutive quarters is the single best piece of evidence the core business is healthy); confirmation that AI-driven deflation is compressing NSR per employee the way it has hit IT-services peers.
6. Position sizing note
A portfolio-specific passage was removed from the public build.