Date: 2026-02-25 Quarter: Q4 FY25 (ended Dec-2025) Market cap: ~$28.7B | EV/TTM Rev: ~33x | Non-GAAP EPS FY25: $1.84 Prior corpus coverage: None — first analysis of ALAB
I haven't covered Astera Labs in the newsletter before, but applying my framework here produces a clear conclusion: this is a genuinely excellent AI infrastructure picks-and-shovels business trading at a price that leaves no room for error. Q4 FY25 delivered $270.6M in revenue (+92% YoY, +17.5% QoQ), the fourth consecutive beat averaging 9% above the midpoint guidance — a deliberate sandbagging pattern I find credible and repeatable. Full-year FY25 revenue reached $852.5M, up 115% year-over-year, with non-GAAP operating margin expanding to 39.2% for the year and 40.2% in Q4. FCF for the year was $281.8M at a 33.1% margin, and the balance sheet holds $1.19B in cash and marketable securities with zero debt. This is not a company executing poorly. But three post-earnings disclosures created a legitimate re-rating event: gross margin guided to ~74% for Q1 FY26 (with Amazon warrant contra-revenue adding ~200bp additional drag from Q2), OpEx guiding up $16–20M sequentially as R&D accelerates, and a CFO departure mid-cycle. The stock dropped 20%+. My assessment is that the selloff reflected rational repricing of the gross margin trajectory, not panic. At ~33x TTM EV/S on a decelerating growth rate, ALAB is priced for the Scorpio X ramp to deliver growth reacceleration in H2 FY26. That is a bet I find intellectually interesting but not yet compelling at this valuation. I would not initiate a position here. I would watch for the Scorpio X material ramp signal and a gross margin stabilisation, at which point a pullback to the 20–22x TTM range would be attractive.
The headline growth is outstanding. Four consecutive quarters of +90–144% YoY growth, driven by three distinct forces that all still have runway: Aries PCIe retimers growing ~70% YoY (PCIe Gen 6 transition + smart cable module content uplift), Taurus Ethernet SCMs growing 4x+ YoY (400G→800G transition underway), and Scorpio switch fabric crossing 15% of FY25 revenue with X-Series just beginning to ship.
The sequential dollar adds through FY25 were remarkably stable: $18M, $33M, $38M, $40M per quarter. The Q1 FY26 guide midpoint of 291.5Mimplies+21M sequential — a sharp deceleration in absolute dollar terms. My read on this is that management is conservative by design. The 4-quarter history of 8–11% beats against deliberate low-ball guidance suggests actual Q1 lands closer to 315–320M, whichrestoresthesequentialaddto 47M. That is the most important number to verify in May.
What I find structurally compelling is the product diversification. In Q1 FY25, Scorpio+Taurus represented roughly 8% of revenue; by Q4 FY25, the two product families had jumped to ~30% combined. That is a genuine product mix inflection, not a marketing story. Three hyperscaler design wins on Scorpio P (up from one), and Scorpio X-Series starting initial customer shipments with material ramp targeted H2 FY26. If UA Link wins the scale-up protocol war (AWS Phranyon 4, AMD MI 500 both confirmed support for 2027 deployments), Scorpio X becomes the most important product in the portfolio.
The Amazon $6.5B cumulative warrant agreement through 2033 — implying roughly $800M per year in committed purchasing across smart fabric switches, signal conditioning, and optical — is a genuine demand floor, not a purchase order. The economic terms were not disclosed, which is the part of this that I find troubling. I will address that in the valuation section.
This is where the investment case gets complicated. ALAB's gross margin trajectory is clearly negative as the product mix shifts toward hardware-intensive products. The data is unambiguous:
| Quarter | Non-GAAP GM | Notes |
|---|---|---|
| Q1 FY24 | 78.1% | Aries-dominant mix |
| Q2 FY24 | 77.9% | |
| Q3 FY24 | 77.8% | |
| Q4 FY24 | 74.1% | Taurus SCM mix begins |
| Q1 FY25 | 74.9% | Partial recovery |
| Q2 FY25 | 76.0% | |
| Q3 FY25 | 76.5% | Mix improvement |
| Q4 FY25 | 75.7% | Scorpio/Taurus inflection |
| Q1 FY26e | ~74% | Guided; Amazon warrant adds ~200bp from Q2 |
The step-down from 77–78% in early FY24 to a 74–75% range in FY25–26 is structural, not cyclical. Scorpio (switch silicon + hardware) and Taurus (cable modules) carry lower margins than pure signal conditioning silicon like Aries PCIe retimers. As these products grow to dominate the mix, gross margins compress. Management guided Q1 FY26 to ~74%, and flagged approximately 200 basis points of additional contra-revenue headwind from the Amazon warrant starting Q2 FY26. That would push effective gross margins toward 72% or below in H2 FY26 unless product mix improves.
I want to be precise here because the non-GAAP/GAAP difference is negligible on gross margin (~10 bps). The operating margin story is different — non-GAAP op margin was 40.2% in Q4 vs. 24.8% GAAP, with the gap almost entirely SBC at ~$41M/quarter or 15.3% of revenue. That SBC load is declining (from 18.8% FY24) and appropriate for a post-IPO high-growth semiconductor company, but it is real dilution that I do not ignore.
The non-GAAP operating margin actually remained robust through Q4 at 40.2% despite the gross margin compression, because R&D as a percentage of revenue has been improving even as absolute dollars rise. Q4 R&D was $93.8M on $270.6M revenue = 34.7%. That said, Q1 FY26 non-GAAP OpEx is guided at $112–118M (midpoint $115M) on $291.5M revenue guide = 39.5% of revenue. Combined with 74% gross margin, that implies Q1 FY26 non-GAAP op margin of roughly 74% - 39.5% = ~34.5%. That is a meaningful step down from 40.2% in Q4 FY25. Not because the business is deteriorating, but because management is front-loading R&D investment (Israel design center, aiXscale/AQUI acquisitions, optical development) ahead of what they describe as an 18–24 month time-to-revenue cycle.
This is the kind of deliberate investment I find appropriate when management credibility is high. The question is whether the 18–24 month revenue payback is realistic. Given that Astera has consistently delivered on its product roadmap commitments (Scorpio P from 1 to 3 hyperscalers, Taurus 4x growth, Aries near-70% growth), I give management benefit of the doubt on the investment cycle. But it is a bet, and the bet gets harder to defend as valuations stay elevated.
FY25 FCF was $281.8M on $852.5M revenue — 33.1% margin. That is a strong number for a company still in hyper-growth. The $1.19B net cash position (zero debt) provides significant optionality. Capex was only $37.5M in FY25, so this is not a capital-intensive business in the traditional sense.
The quarterly FCF pattern warrants a note: Q1 FY25 was $6M (working capital drag), Q2 spiked to $133M (large AR collections), Q3 was $66M, Q4 was 77M.TheH2runrateof 71M/quarter, annualised to ~$284M, is the base from which to model FCF growth as revenue scales. On a $28.7B market cap, the FCF yield is approximately 1%, or ~102x FCF. Rich. But at 50%+ revenue growth, the 12-month forward FCF yield improves materially.
SBC at $160M in FY25 (18.8% of revenue) is the honest-cash dilution that investors who focus only on non-GAAP metrics miss. FCF already absorbs this — the OCF includes SBC as an add-back. But the GAAP net income of $219M on $28.7B market cap gives a trailing P/E of ~131x. I don't normally lead with GAAP P/E for high-growth companies, but the gap between GAAP and non-GAAP is wide enough here that investors should be aware.
Non-GAAP P/E: FY25 non-GAAP EPS 1.84on 158/share (approximating from 181M shares at 28.7B)= 86xtrailingnon − GAAPP/E.Forwardnon − GAAPP/EonestimatedFY26non − GAAPEPSof 2.50–2.75 (management implied by Q1 guide of $0.53–0.54, annualizing with growth and OpEx investment) is approximately 58–63x. PEG at 58x forward P/E with ~50–60% FY26 revenue growth expected: ~1.0–1.1x PEG. For a best-in-class AI connectivity platform with 40% non-GAAP operating margins, that PEG is defensible, but it leaves no discount.
The correct comparator set for ALAB is AI semiconductor infrastructure — specifically the picks-and-shovels layer where I have been concentrating exposure. The universe includes NVDA, CRDO, MRVL, and ALAB itself.
At ~33x TTM EV/S on 92% YoY growth: ALAB's EV/Rev-to-growth multiple is roughly 0.36x. Compare that to CRDO (Credo Technology) at approximately 22x TTM EV/S on 274% YoY growth = 0.08x. The premium ALAB carries versus CRDO is significant and requires justification. That justification exists in the form of platform breadth (5 product families vs. CRDO's more concentrated portfolio), established 40% non-GAAP operating margins, $1.19B net cash, and the Amazon $6.5B demand commitment. Platform companies deserve a premium over point-solution companies — I accept that.
But the specific quantum of premium gives me pause. For ALAB to trade at fair value on a forward basis relative to its growth cohort, the market needs to believe:
None of these are unreasonable assumptions. But they are priced in at current levels. The downside scenario — gross margins compress to 70% as Amazon's warrant terms become apparent, Scorpio X ramp lags by one to two quarters, or hyperscaler capex pacing slows — produces a re-rating that would be painful from 33x TTM EV/S.
The Amazon $6.5B warrant is both the bull and bear case embedded in the same data point. The commitment is structurally positive for demand visibility. But the pricing concessions that Amazon extracted in exchange for $6.5B in committed purchasing are likely the mechanism through which gross margins decline. A volume customer of Amazon's scale does not sign 9-year $6.5B commitments without extracting margin. The ~200bp contra-revenue drag is only the accounting recognition; the pricing dynamic in the underlying contracts matters more.
If I were managing my position sizing in the high growth portfolio, I would be holding ALAB at no more than 3–4% given valuation and concentration risk, rather than initiating new capital here. The EV/S at 33x is at the high end of what I find defensible for a 90% grower. At 22x (roughly where CRDO trades despite faster growth), ALAB would become compelling — that requires either a ~33% price decline or a period of earnings growth without multiple expansion.
Sandbagging pattern: 4 consecutive beats averaging 9% above guide midpoint. That is a deliberately undemanding guidance approach I respect. The Q3→Q4 beat narrowed to 8%, but the absolute dollar beats ($19–24M) remained consistent. This management team guides conservatively and delivers consistently.
The CFO transition is a yellow flag, not a red one. Mike Tate guided the company through IPO, FY24, and FY25, building a track record of financial credibility. Desmond Lynch (ex-Rambus CFO) steps in March 2, 2026 — his first earnings call is Q1 FY26 in approximately May. Rambus is a semiconductor IP company with similar margin complexity. The qualifications are appropriate. But until Lynch demonstrates the same sandbagging discipline on the Q1 call, I will mark this as a watch item.
The Israel Design Center expansion — guided R&D up $16M sequentially in Q4, with further step-up in Q1 FY26 — signals genuine confidence in the pipeline. Management's commentary that "the TAM is much bigger than we originally expected just 12–18 months ago" is the kind of statement that either precedes a major upside surprise or a cycle of over-investment. Given Astera's product execution history, I lean toward the former, but I want to see silicon delivery timelines before committing capital to that belief.
The Scorpio 20% of Q4 revenue target was an apparent miss — management deflected the direct analyst question and the language suggested 15–18% rather than 20%. That is a minor credibility dent. Not a trend, but I track management credibility carefully. One miss on an aspirational target is not disqualifying. A second would be.
The verdict: ALAB is a high-quality AI infrastructure platform company executing well in the picks-and-shovels layer I believe will accrue value before the application layer. The beat-and-raise cadence is intact, the product roadmap is deepening, the balance sheet is pristine, and the Amazon demand commitment is structurally validating.
But I would not be initiating a position at ~33x TTM EV/S. Gross margin compression is structural (72–74% heading into FY26 including warrant headwind, vs. 78% a year ago), OpEx is accelerating ahead of revenue, and the forward growth reacceleration that justifies this multiple depends on Scorpio X delivering material revenue in H2 FY26. The setup is a stock priced for perfection in a business model where the economics are measurably getting worse on a gross margin basis.
If I were reviewing this for the high growth portfolio, I would target a 2–3% position initiated on a pullback to 22–24x TTM EV/S — roughly $200–210/share (down ~25–30% from pre-earnings levels), at which point the risk/reward begins to compensate. At current prices, CRDO offers superior risk/reward within the same picks-and-shovels AI connectivity thesis.
Thesis status: Not in portfolio — new coverage. Watching for Scorpio X ramp confirmation and gross margin stabilisation as conditions for initiation. Target entry EV/S: 22–24x TTM (vs. ~33x current) Management credibility: Reliable — beat-and-raise cadence intact; CFO transition is a watch item