Astera Labs at Citi’s 2026 Global TMT Conference: AI build-out drives growth

Astera Labs guided Q3 revenue to $550 million at the midpoint, versus $392 million in the latest quarter, as Scorpio X enters volume production and is expected to become its largest product family by revenue. Management highlighted revenue growth to $1.2 billion over the past 12 months (+98%), EPS expansion from $0.10 at IPO to $0.80 in the latest quarter, and gross margins in the mid-70% range. The company expects hyperscaler AI capex to exceed $1 trillion next year and sees optical connectivity beginning to contribute revenue in 2027, although customer concentration remains elevated and third-party valuation analysis flags the shares as overvalued. ALAB shares were up 5.48% at $304.67.

Analysis

The investment debate is no longer whether ALAB can grow through the current ramp, but whether its architecture-level position converts into durable platform economics before competitors close the feature gap. The critical near-term issue is concentration: a pulled-forward lead-customer deployment can create a powerful beat-and-raise cycle over the next 1-3 months, but it also raises the probability that any platform timing change produces an outsized revenue air pocket. At a $52.8B equity value and after a 136% six-month run, execution must remain near-perfect; a gross-margin glide toward 70% would matter more to the multiple than incremental revenue upside.

ALAB’s scale-up gains are competitively disruptive to AVGO’s legacy PCIe-switch position, but AVGO has the customer relationships, manufacturing leverage and custom-silicon bundling capability to respond. MRVL is more exposed to the longer-duration optical-content narrative, yet ALAB’s switch-led strategy could commoditize standalone optical attachment economics if it successfully controls specification choices. NVDA is a qualified ecosystem beneficiary rather than a displaced incumbent: ALAB’s custom attach opportunities expand NVDA platform accessibility, while proprietary NVLink remains the limiting factor on how much scale-up TAM ALAB can capture.

Consensus is likely over-extrapolating the 2027-28 optical optionality into current valuation. Optical revenues remain dependent on hyperscaler architecture choices, qualification success and the timing of multi-rack scaling; these are design-win milestones, not booked demand. The more underappreciated upside is a diversified Scorpio X customer conversion cycle by year-end, which would reduce concentration risk and support a higher quality of revenue than a single-customer ramp.

Falsifiers: Scorpio X fails to convert several late-stage engagements into wins by year-end; quarterly gross margin moves below 72% before volume optics; or a lead customer’s AI-capex/platform timing weakens. Conversely, evidence of a second material Scorpio X deployment would justify revisiting the short valuation case.

Key Decisions for Investors

• Do not chase ALAB after the conference-driven move. Establish a 1-3 month watch position only on a 10-15% pullback or after confirmation of a second material Scorpio X customer; target 20-30% upside on multi-customer de-risking, with a 12% stop below entry tied to margin or design-win slippage.
• Pair trade: long ALAB / short MRVL in equal dollar risk through year-end, sized small. ALAB has nearer-term switch-ramp catalysts while MRVL’s optical exposure is more dependent on a later deployment curve; exit if MRVL announces a major hyperscaler optical design win or ALAB misses Scorpio X conversion expectations.
• For a valuation-aware expression, buy ALAB 6-9 month call spreads rather than outright stock after the next earnings volatility reset; use strikes around 10-15% out-of-the-money and cap premium at 3-4% of notional. The thesis requires a beat-and-raise plus customer-diversification catalyst, not merely sustained AI sentiment.
• Monitor AVGO’s next switching/custom-silicon commentary and hyperscaler capex guidance as competitive leading indicators. A meaningful AVGO Gen-6 share recovery or bundled scale-up design win would weaken the ALAB share-gain thesis before it appears in reported results.