Google Is Getting Paid in Marvell Stock Warrants for Buying Marvell's Chips

Marvell granted Google a warrant for 58,970,907 shares (about 7% of Marvell) at a fixed $206.58 exercise price tied to Google’s discretionary chip purchases ($500M per vesting block). Marvell stock jumped ~10% after the disclosure, while Alphabet reported a strong Q2 with revenue up 24% YoY to $119.8B and a $98.0B net gain (mostly unrealized equity gains), lifting EPS to $9.11 nearly 4x. The warrant structure effectively converts Alphabet’s AI chip spending into Marvell equity, with an estimated full-package paper upside of about $12.2B, but it is economically small versus Alphabet’s ~$4T market value.

Analysis

This is less about one vendor and more about a shift in procurement power. Hyperscalers are turning AI capex into an option on supplier equity, which means future chip deals may increasingly be priced as revenue plus embedded dilution rather than clean margin expansion. That dynamic is structurally negative for smaller semicap names with concentrated customer exposure: the customer captures more of the economics, while the supplier takes balance-sheet-like dilution risk without guaranteed volume.

The near-term winner is the buyer, not the seller. GOOG/GOOGL gets a low-cost call option on MRVL upside that can be marked through other income, but the real value is strategic: it strengthens Google's negotiating leverage with AVGO and any future TPU partners. Second-order, this pressures the entire custom-ASIC ecosystem to defend share with more aggressive pricing, better terms, or equity sweeteners, which could cap multiple expansion across MRVL-style names even if revenue remains solid.

The move is likely overdone on the stock reaction, underdone on the precedent. In 1-3 months, the key catalyst is whether MRVL can show that the Google relationship expands design-win breadth without margin dilution; if not, the market should re-rate the stock from "AI beneficiary" to "customer-financing story." Falsifiers: a large upward revision in MRVL gross margin/FCF guidance, or evidence the warrant is purely symbolic and non-replicable. Otherwise, this is a template for tougher commercial terms across AI silicon over the next 6-18 months.

Key Decisions for Investors

• Short MRVL into strength / buy 1-3 month puts if the post-disclosure rally keeps the stock near recent highs; the thesis is that the market is overcapitalizing headline AI validation while underpricing dilution and customer leverage. Falsify if management raises FY gross margin or FCF guide by enough to offset the economics of the warrant.
• Long GOOG vs short MRVL as a pair trade for 1-3 months: GOOG gets strategic optionality with little balance-sheet impact, while MRVL bears the economic cost of winning the account. Best entry is on any follow-through gap-up in MRVL that stretches the relative spread.
• Watch AVGO for a relative short or underweight versus GOOG/mega-cap software if the market starts extrapolating this structure to TPU renewals; the risk is that hyperscalers increasingly demand economics that compress supplier returns. Trigger: any sign that Google's TPU supply discussions include similar equity-linked terms.
• If staying long semis, prefer NVDA/AMD exposure over names with a single-hyperscaler concentration and custom-ASIC dependence. The equity-kicker precedent is most damaging where a customer can dictate terms and multi-source, not where product scarcity keeps supplier leverage intact.