ALAB

Astera Labs is a genuinely exceptional business at this stage of the AI build-out. Revenue compounded from $79.9M (FY22) to $853M (FY25) — +115% YoY in 2025 alone — and the company crossed into real GAAP profitability for the first time, swinging from a -$83.4M loss in FY24 to +$219M net income (net margin 25.7%). Gross margin is a software-like 75.7%, operating cash flow ($319M) comfortably exceeds net income, and the balance sheet is pristine: $1.36B equity against just $168M total liabilities (0.12x liabilities/equity) with effectively no debt. ROE of 16.1% is understated by the cash-heavy, recently-IPO'd equity base. The MD&A confirms the growth is volume- and mix-driven (more Aries, Scorpio, Taurus shipments plus higher ASPs from the module/Scorpio mix), and operating expenses grew only 13% against 115% revenue growth — clear operating leverage. This is a first-to-market PCIe/CXL/Ethernet connectivity leader sitting at the heart of AI infrastructure.

The problem is entirely valuation. At $417.07 the stock trades at 341.9x trailing EPS and 83.3x trailing sales for a $71B market cap on $853M of revenue. Even granting hyper-growth and a long AI runway, that multiple discounts years of flawless execution. The stock jumped +11.31% today and the entry price embeds extraordinary expectations; any deceleration, a lost design win, or pricing pressure would compress the multiple violently. The 10-K is explicit about the fragility beneath the growth: 'a substantial portion of our revenue is driven by a limited number of our end customers,' meaning hyperscaler concentration is severe, and management flags that 'pricing for the current generation of our existing products often decreases over time,' which already shaved gross margin 70bps (76.4%→75.7%) as module mix grew. Add lengthy/expensive customer qualification cycles, reliance on a limited number of third-party manufacturing partners with no long-term supply contracts, and China trade-policy headwinds that have 'reduced the demand for our products.'