Google Taps Marvell for Custom AI Chips
Full vesting is tied to $120 billion in Google purchases through fiscal 2033.
Topics
News
- Apple Tightens Mac Access Rules as AI Agents Gain Reach
- Trump Puts Intelligence Chief Jay Clayton in Charge of AI Task Force
- IBM Widens AI, Quantum Research With IIT Bombay, IISc
- OpenAI, Synopsys Build AI Model That Could Slash Chip Design Time
- Pentagon Plans Four-Star Command for AI, Drone Warfare
- California Draws the Line on AI Firings, Layoffs and Workplace Surveillance
Google is expanding its custom-chip supply chain, bringing Marvell Technology into the development of processors and supporting hardware for its artificial intelligence infrastructure while securing warrants to buy nearly 59 million Marvell shares.
Under the agreement, Google can buy up to 58.97 million shares at $206.58 each. Exercising the full allotment would cost about $12.18 billion and could make Google Marvell’s fifth-largest shareholder, according to Reuters.
Most of the warrants are tied to Google’s purchases. About 1.36 million shares will vest quarterly during the first year. The remaining 57.61 million will vest in 240 equal tranches through fiscal 2033, with one tranche released for every $500 million in qualifying revenue. Full vesting would therefore require $120 billion in revenue, but that amount is not guaranteed.
Marvell shares rose nearly 8% after the announcement, while Broadcom, Google’s main custom-chip partner, fell more than 5%. Alphabet shares were little changed.
The partnership covers custom silicon linked to Google’s Tensor Processing Unit ecosystem, including AI inference accelerators, storage and network controllers, memory-interface controllers and near-memory computing products.
Demand for custom AI chips has increased as technology companies look for alternatives to Nvidia’s graphics processors and seek hardware optimized for their own workloads, particularly for inference, when trained models generate responses and perform tasks.
For Google, expanding the number of suppliers could also reduce dependence on a single chip partner as its spending on data centers and AI infrastructure rises.
The structure also adds to a growing pattern of large AI customers taking financial interests in companies supplying computing infrastructure. AMD last year gave OpenAI the option to acquire a stake alongside a major chip-supply agreement, while Nvidia has increasingly become involved in financing projects that will ultimately purchase its hardware.
The Marvell deal further blurs the line between AI customers, suppliers and investors as the cost of building computing infrastructure continues to rise.
