ALGM › analyze
ALGM · Analyze
Price $35.42 · Mkt cap $6.60B · EV $6.50B · FY ends late March
Verdict: WATCH — conviction 5.0. A real business inflection bought at a price that already assumes it completes. The magnetic-sensor franchise is genuine and #1 by revenue share; the data-center story that drove the re-rating is the least defensible part of the portfolio. Every officer and director sold into $41–53 and none bought back at $35.
0. Framework placement
| Company quality | Market price | Verdict |
|---|---|---|
| Good (real #1 share, improving) | Expensive (~35x FY27) | Wait / watch — good company, bad entry |
1. Data corrections — read before any number below
Three vendor figures in the standard bundle are wrong or misleading for this name.
| Field | Vendor value | Corrected | Why |
|---|---|---|---|
PE(fwd) |
23.99 | ~34.7x | Implies EPS $1.48. FQ1 actual $0.23 + FQ2 guide mid $0.245 = $0.475 for 1H FY27; $1.48 needs $1.00 in 2H against a mid-single-digit sequential revenue guide. Consensus FY27 (FYE Mar-27) EPS is $1.02. The $1.48 is an FY28 number mislabelled. → pitfall-vendor-forward-eps-is-the-wrong-fiscal-year |
Buybacks −$853.9M (FY25) |
reads as capital return | Sanken stake repurchase | July–Aug 2024: 38.77M shares bought from Sanken at $23.16, funded by a $665.9M primary offering + ~$200M credit facility. Not a shareholder return — a recapitalisation. Ongoing buyback is $0. |
| "52w high $71.77" | −50% drawdown | −40% | $71.77 was an intraday tick. Highest weekly close was $59.00 (w/e 2026-06-15). ALGM is still +34% YTD and +58% off the Nov-2025 low of $22.41. |
One web claim also had to be killed: the widely-repeated "Greater China ≈ 60% of revenue" is wrong. The FY26 10-K gives Greater China $249.4M of $890.1M = 28%; management said ~25% in FQ1 FY27.
Unresolved data gap: FY28 consensus ($1.48) is paywalled and could not be independently confirmed. Confirm on a terminal before it is used to justify an entry.
2. Fundamentals — the trough is behind, the cash is not yet here
Reported trajectory (fiscal years end late March)
| FY23 | FY24 | FY25 | FY26 | TTM (thru Jun-26) | |
|---|---|---|---|---|---|
| Revenue | $973.7M | $1.05B | $725.0M | $890.1M | $945.9M |
| Gross profit | $546.1M | $574.5M | $321.5M | $412.0M | — |
| Operating income (rep.) | $203.3M | $209.5M | −$19.8M | $25.1M | $46.7M |
| Net income | $187.4M | $152.7M | −$73.0M | −$14.9M | $14.2M |
| Diluted EPS | $0.97 | $0.78 | −$0.39 | −$0.08 | $0.08 |
| FCF | $113.4M | $56.9M | $20.8M | $124.9M | $87.9M |
The shape is a textbook analog-semi cycle: peak FY24 → trough FY25 → recovery. FY25 was the bottom, not a structural break.
Sequential recovery — the bull's best evidence
| Quarter | Revenue | GAAP GM | GAAP op margin | FCF |
|---|---|---|---|---|
| Jun-25 | $203.4M | 44.9% | −1.3% | $51.0M |
| Sep-25 | $214.3M | 46.3% | 2.9% | $13.9M |
| Dec-25 | $229.2M | 46.7% | 4.2% | $41.3M |
| Mar-26 | $243.2M | 47.0% | 2.2% | $18.7M |
| Jun-26 (FQ1 FY27) | $259.2M | 48.5% | 9.8% | $14.0M |
Six consecutive quarters of sequential growth. Revenue +27% YoY, gross margin +360bp, operating margin from −1.3% to 9.8%. Distributor inventory is described as "flat… in a very healthy place" — destocking is over. FQ2 guide $265–275M (+26% YoY).
🚩 But the cash does not follow the P&L
TTM FCF is $87.9M against $51.2M of stock-based compensation. SBC is 58% of free cash flow. FQ1's $14.0M of FCF on $259.2M of revenue is a 5.4% FCF margin in the best quarter of the recovery. FCF/EV is 1.35%.
Share count is rising again: diluted 184.6M (Jun-25) → 187.8M (Jun-26), +1.7%. The FY25 recapitalisation retired Sanken stock; it did not begin a return programme. Per-share dilution has resumed.
Balance sheet — clean, and not the issue
Debt $305.9M vs cash $168.8M; Debt/Assets 21.6%; current ratio 3.70; interest expense falling ($6.4M → $4.4M quarterly). No solvency question here.
Capital allocation
No dividend, no buyback, no M&A since the FY24 acquisitions ($408.1M). FCF is going into a modest capex step-up ($38.2M FY26, rising to ~$8M/qtr) and working capital. Management is neither returning cash nor deploying it aggressively — a neutral, not a positive.
3. Moat — real, but narrower than the price implies
Quantitative base. Gross margin recovering hard (44.9% → 48.5% in four quarters) with a stated 55%+ target via copper wire bonding and fab efficiency. That trend is the moat showing up in the numbers. ROIC is not yet meaningful — GAAP operating income has only just crossed zero — which is itself the point: a moat you cannot see in returns on capital is a claim, not a finding.
Position. Yole Group puts Allegro at ~23–28% of magnetic sensor IC revenue — #1, with Infineon and Melexis each ~15%. TDK/Micronas and AKM lead on units via consumer design wins, not value. (Second-hand citation; an independent 2026 share table could not be sourced.)
Adversarial stress-test — "you are a well-funded rival." The honest answer splits by product line:
- Magnetic sensing (automotive): hard to attack in-cycle, easy at the redesign. AEC-Q100 qualification plus a 2–4 year design-in cycle means an incumbent part stays for the model life (5–8 years). But the socket re-competes at every platform refresh. This is a re-compete moat, not a lock-in moat — a meaningful distinction the bull case elides.
- Data center (current sensors, fan drivers, isolated gate drivers): easy to attack. ~1-year design cycles, and this is Monolithic Power, Infineon and TI's home turf. ALGM has the least incumbency precisely where the growth narrative now lives.
Revenue-stream map (FQ1 FY27). Automotive $165.3M (64%, +15%) · Industrial & Other $93.9M (36%, +59%). Data center is 17% of total sales, +32% sequential, current sensors +66% sequential at mid-50s gross margin (accretive). Guided to more than double in FY27.
Disruption forecast (5–10yr). Favourable: the content-per-vehicle arc from ~$40 in legacy ICE to "upwards of $100" in next-gen BEV runs through steer-by-wire and electro-mechanical braking — each architecture step adds sensors. Unfavourable: ALGM does not control the timing of those architecture changes, and Chinese OEM localisation pressure cuts both ways (China is 25–28% of revenue and the fastest-growing region). The humanoid-robotics claim of >$150 content per unit by 2030 is marketing; it should not enter a model.
Evergreen assessment: qualified yes. Magnetic sensing is a durable, growing content category and ALGM leads it. It is not evergreen at the socket level — every platform is a new fight.
4. Valuation
Non-dividend payer → DDM and DYT are N/A. Graham computes to $2.85 (or ~$9.65 on a normalised $0.80 mid-cycle EPS) against $35.42 — for a company crossing zero GAAP earnings at a cyclical inflection, Graham is not informative here. Ignore it; say so rather than dress it up as a signal.
The multiple that matters
| Basis | EPS | P/E at $35.42 |
|---|---|---|
| GAAP TTM | $0.08 | 443x |
| FY26 GAAP (actual) | −$0.08 | n/m |
| FY27 consensus (non-GAAP) | $1.02 | 34.7x |
| FY28 (vendor "forward", unconfirmed) | $1.48 | 23.9x |
The real forward multiple is ~35x FY27. The 24x that appears on the screen is an FY28 price for an FY27 business — a year and a half of doubling data-center revenue and the 55% gross-margin target both arriving, paid for today.
Cross-checks
- EV/EBITDA 48x (TTM EBITDA $100.9M). The June peak was struck at ~75x.
- FCF/EV 1.35% on TTM FCF; 1.9% on FY26's $124.9M.
- Bogle: 0% yield + cyclical-recovery earnings growth − a multiple that has further to fall if the 2H margin inflection slips. Wide and unreliable at an inflection; not load-bearing.
Fair value range
Anchoring on FY27 consensus EPS of $1.02 at 25–30x (a fair band for a recovering analog name with #1 share) gives $26–31. Granting the FY28 $1.48 at a de-rated 22–25x gives $33–37. Blended: $26–36.
At $35.42 the stock sits at the top of its own fair-value range. There is no margin of safety, and the conservative bias in the framework says that decides it.
5. Sentiment & intelligence
Why it fell — multiple reset, not earnings reset. 30 Jun: Mizuho $54→$67 and TD Cowen $55→$70, both on AI data-center demand; the blow-off top followed within days. 20 Jul: Barclays downgrades OW→EW, PT $48. ~24 Jul: new US tariffs of 10–12.5% on 60 partners. 30–31 Jul: beat-and-raise, stock sold anyway — TD Cowen cut $66→$48 (−27%), Mizuho $67→$58, objecting to guidance composition (mid-single-digit sequential in both segments, margin benefit deferred to 2H). 18 Aug: −9.6% on a broad chip rout, not ALGM-specific.
M&A. onsemi bid $35.10/share all-cash (~$6.9B) in March 2025 and withdrew 14 Apr 2025, citing the board's "reluctance to fully engage." Today's $35.42 is essentially that price. That cuts both ways: a floor reference, or a reminder that a strategic buyer's own diligence stopped there. No evidence of a live 2026 bid — the June spike traces to analyst PT hikes.
🚩 Sanken overhang. Post-2024 repurchase Sanken holds ~32.3% (confirmed 2025-09-26), with a board-observer right, against a 125M float. The 14-month lock-up expired ~Sept 2025 and no 2026 sale or 13D has been filed. A holder that has already demonstrated it wants out, now free to sell, against a float this size. Verify against the latest 13D/G and proxy before sizing.
🚩 Insider signal — the cleanest negative in the file. Zero open-market purchases in twelve
months; $5.4M sold, $0 bought. Nearly the entire officer and director bench sold, clustered
at $41–53 in May–June 2026, including CEO Doogue 30,089 shares at $47.23. The last genuine
open-market buys were Nov 2024 at $19.00–19.89 (Doogue 15,000) — the same people who bought
the bottom sold the top and have not returned at $35. (Grant lines — Doogue's 160,835 RSUs,
D'Antilio's 59,877 — are excluded per pitfall-yahoo-insider-purchases-counts-rsu-grants.)
Institutional. T. Rowe +3,160% (new ~3.7M shares), Invesco +56%, JPMorgan +47%, BlackRock +11% — aggressive new money during the run-up; FMR −6.7% (still largest at 13.7%), Capital Research −3.5%. Short interest 13% of float.
Consensus. 12 analysts, 12 Buy / 0 Hold / 0 Sell, mean target $54.82 (range $48–62). Every target sits ~35–75% above the price after a 40% drawdown — a stale book, not information.
6. Tensions surfaced
- Fundamentals vs Valuation. Fundamentals reports the best quarter in two years. Valuation reports 35x a consensus number that needs a 2H margin inflection management has promised but not delivered. Resolution: Valuation wins on the framework's conservative bias — the operating improvement is real and already in the price.
- Moat vs Sentiment. The moat analysis rates the auto franchise durable; the market is paying for the data-center leg, where the moat is weakest. Resolution: underwrite the two separately. The magnetic-sensor business justifies roughly a market multiple; the data-center business deserves a discount for competitive position, not the premium it received in June.
- Insider selling vs the recovery. Both are facts. The bench sold into the improvement they were reporting. That does not falsify the recovery — it argues the price ran past it, which is the same conclusion as (1).
7. Verdict — WATCH, conviction 5.0
The business is better than it has been in two years and the stock is still not cheap. At ~35x FY27 with TTM FCF of $87.9M against $51.2M of SBC, an unlocked 32% holder, 13% short interest, and universal insider selling into the top, the asymmetry is not there at $35.42.
Nothing here is broken enough to justify AVOID — it is a good business at a bad entry, which is precisely the "wait / watch" cell of the framework.
- Fair value $26–36 · price sits at the top of the range
- Entry $25–29 — a genuine discount to the FY27 lens, near the pre-spike March base and above the $22.41 low. With beta 1.91 and 43 daily moves >5% in the last year, this is a realistic wait, not a fantasy bid.
- Trim 30x forward (multiple, not a dollar — earnings are inflecting fast and any fixed dollar level would go stale within two quarters)
What would change the verdict
| Direction | Evidence |
|---|---|
| → ACCUMULATE | Price into $25–29 with the recovery intact; or FQ3 delivers the 2H gross-margin inflection toward 55% and FY28 consensus is independently confirmed; or an insider buys in the open market |
| → AVOID | Sanken files to sell; data-center growth stalls (it is the whole re-rating); gross margin flattens below 50% |
Key risks
- Sanken's ~32% unlocked stake against a 125M float — the largest single overhang.
- The forward multiple rests on an unconfirmed FY28 estimate. If FY28 consensus is lower than $1.48, the "cheap on forward" argument disappears entirely.
- The growth narrative sits in the weakest competitive position (data center vs MPWR/ Infineon/TI, ~1-year design cycles).
- Tariff exposure is unquantified — principal assembly/test is AMPI in the Philippines; the July 2026 tariff round was not addressed on the call.
- SBC at 58% of FCF with a rising share count.
Next binary event
2026-10-28 — FQ2 FY27. Guide is $265–275M / non-GAAP $0.23–0.26, and the 2H margin inflection must begin showing. Given the −5% to −11% single-day history, this print is decision-relevant on its own.
Sources
FY26 10-K (FYE 2026-03-27) · FQ1 FY27 8-K Ex-99.1 (2026-07-30) · FQ1 FY27 earnings call
transcript · onsemi withdrawal PR (2025-04-14) · Sanken repurchase PR (2024-07-29) · Yole Group
magnetic sensor share (second-hand) · SEC Form 4 filings (direction read individually) ·
Yahoo Finance MCP (statements, prices, recommendations) · .mcp/fin.py