Marvell Technology and Snowflake both rode the AI wave in fiscal 2026, but they arrived at radically different destinations. Marvell posted $2.7 billion of net income on $8.2 billion of revenue, a 42% jump from the prior year that flipped a net loss into a 32.6% margin. Snowflake grew revenue 29% to $4.7 billion yet lost $1.3 billion, leaving a negative 28.4% margin. For investors picking between the picks and the shovels of the AI gold rush, the choice is now between a hardware vendor that prints money and a software platform that spends it.
The hardware moat has a concentration problem
Marvell’s numbers read like a textbook infrastructure play: free cash flow of $1.4 billion, a current ratio of 2.0x, and debt-to-equity of just 0.3x. The company designs custom silicon for Amazon’s Trainium chips and has deepened its partnership with NVIDIA, locking itself into the two most consequential build-outs in cloud computing. But 82% of fiscal 2026 revenue came from the top ten customers. That is not a customer base; it is a dependency. If any hyperscaler decides to insource more chip design, a trend the source explicitly flags, Marvell’s addressable market shrinks overnight. The Taiwan foundry exposure adds a geopolitical overlay that no balance sheet can hedge.
The software model runs on stock, not subscriptions
Snowflake’s consumption model sounds elegant until you look at the cash flow statement. Free cash flow hit $1.1 billion, but stock-based compensation consumed 130.9% of operating cash flow. That means the company paid out more in equity than its core operations generated in cash, a gap filled by issuing shares. With 12,000 customers and 790 Forbes Global 2000 logos, the platform has breadth, but the debt-to-equity ratio of 1.4x and a current ratio of 1.3x leave less cushion than Marvell’s. The source notes Snowflake runs on public clouds that are also its competitors, a structural conflict that no consumption metric resolves.
What to watch next
Marvell’s next inflection point is whether hyperscalers keep buying or start building. Snowflake’s is whether it can convert consumption growth into GAAP profitability before SBC dilution becomes a permanent feature of the cap table. Both stocks moved modestly on the day, Marvell down 2.38%, Snowflake up 1.81%, but the divergence in the fundamentals is the story. One company is selling the shovels at a 32% margin. The other is renting the mine and paying the miners in equity.