ALAB Astera Labs

Research status: Primary-source refreshed through the Sep. 9 Citi conference and Sep. 15 Leo X-Series launch. Valuation reference: $360.51 close on 24 Sep 2026, up ~27% from the August dossier reference.

The September evidence strengthens the platform-diversification thesis. Management said Scorpio P is now deployed across multiple customers and that an additional hyperscaler has been added; Scorpio X has seen strong customer traction and remains positioned at the center of scale-up fabrics. The new Leo X-Series smart memory-controller family broadens Astera's CXL/memory role for agentic-AI and general cloud workloads and is designed to pair with Scorpio X-Series to improve token economics and memory utilization.

The business is executing extremely well, but the stock rerating has consumed much of the upside that existed in August. At $360.51, the current price is now above the old probability-weighted 24-month range and near the upper end of the prior base case before Q3 results have proven the next step-up in earnings power.

PowerFund therefore keeps ALAB at investigate / do not chase. The September evidence raises confidence in the company, not necessarily expected return at today's price. The next material re-underwrite belongs after Q3, when we can test the guided Scorpio inflection, customer diversification, margins, tax normalization and diluted share count.

Investment case

Astera Labs is evolving from a retimer-led semiconductor supplier into a broader rack-scale AI connectivity platform spanning PCIe/CXL signal conditioning, fabric switches, memory controllers and Ethernet connectivity.

> The PowerFund thesis is that Astera can materially increase content per AI rack through Scorpio switching, Leo memory connectivity, Taurus Ethernet signal conditioning and Aries retimers while preserving ~70% gross margins. September 2026 strengthens the evidence that product and customer breadth are improving, but the current valuation requires the Q3 Scorpio ramp to convert into durable per-share earnings rather than merely technical leadership.

September evidence

• Management said Scorpio P is deployed across multiple customers and added another hyperscaler customer.
• Scorpio X has seen strong traction and remains central to scale-up architecture ambitions.
• Leo X-Series expands the CXL smart-memory opportunity for agentic AI / general cloud and pairs with Scorpio X for memory-centric scale-up systems.
• Taurus already extends to 3.2T smart retimers/redrivers for 224G Ethernet/UALink.

What changed

1. Customer-diversification confidence improves modestly.
2. Astera's content-per-rack opportunity broadens further across switch + memory + signal conditioning.
3. Product breadth increasingly reduces reliance on Aries alone.
4. Valuation risk has increased because the stock has rerated before Q3 proves the guided revenue step-up.

The Q2 financial baseline remains unchanged pending Q3 results.

• Q3 delivery against $540–560M revenue and ~72% gross-margin guidance.
• Scorpio becoming the largest product family while revenue broadens across hyperscalers/platforms.
• Leo X-Series design wins and evidence of meaningful CXL memory-controller revenue.
• Taurus 3.2T / 224G production wins across Ethernet/UALink fabrics.
• Customer concentration declining while margins remain ~70%+.
• GAAP operating leverage improving as SBC intensity normalizes.

• Valuation: the stock has rerated materially ahead of the Q3 proof point.
• Customer concentration: diversification is improving but still needs economic end-customer validation.
• Platform integration: accelerator/hyperscaler custom silicon can absorb connectivity functions.
• Competition: Broadcom, Marvell, Credo and internal silicon remain formidable.
• SBC/dilution: per-share economics may trail headline revenue growth.
• Architecture cycles: fast platform transitions can create sharp product-revenue swings.
• AI-factor concentration: ALAB would add to PowerFund's dominant deployed factor.

Warning — freeze additions

• Scorpio does not become / remain a major product family after the guided ramp.
• Revenue growth falls below ~30% before diversification is achieved.
• Gross margin <70% for two quarters or <68% once without strategic explanation.
• A major customer cuts purchases >25% without offsetting wins.
• GAAP operating leverage fails to improve despite revenue scale.

Reduce

• Customer concentration remains extreme into 2027 while growth slows.
• Integrated accelerator/rack architectures structurally reduce retimer/switch attach.
• New product families fail to broaden revenue beyond Aries/Scorpio.

Invalidate

> Invalidate if Astera fails to become a diversified fabric-infrastructure platform and remains a concentrated accessory supplier whose attach opportunity is integrated away by larger semiconductor/platform vendors.

Astera's differentiation is broadening from retimer specialization toward a unified rack-scale connectivity platform.

• Credo: strongest overlap in SerDes/AEC/retimer connectivity, now expanding into optics.
• Marvell/Broadcom: broader switching/custom silicon and deeper system integration.
• Hyperscaler/accelerator internal silicon: can remove attach opportunities but also increases need for interoperable merchant connectivity around heterogeneous systems.

The September Scorpio/Leo evidence improves confidence in Astera's ability to raise content per rack. The remaining proof is commercial diversification and per-share economics across multiple platforms rather than one hyperscaler architecture.

1. At Q3, map direct invoiced customers to economic end customers and quantify whether Scorpio concentration is actually declining.
2. Track Scorpio P/X revenue mix, customer count and platform breadth.
3. Quantify Leo X / CXL memory-controller contribution and identify material production deployments.
4. Normalize Q3 tax and SBC effects; track diluted share count against operating leverage.
5. Compare Astera content per rack with Credo/Marvell/Broadcom alternatives across PCIe/CXL/Ethernet/scale-up.
6. Re-underwrite 24/60-month value only after Q3 establishes the new earnings baseline; do not raise fair value merely because the stock price rerated.
7. Define staged entry bands from the post-Q3 expected-return distribution under the principle: scale capital with expected return.

Next full review: after Q3 2026 results or earlier on a material customer/platform disclosure.