Marvell Gives Google $12.2 Billion Stake Option
Marvell Technology has expanded its partnership with Google to develop custom AI chips. As part of the agreement, Marvell gave Google a warrant to buy up to 58.97 million Marvell shares. The shares are priced at $206.58 each, putting the potential stake at about $12.2 billion.
The deal comes as Google increases its investment in custom silicon for artificial intelligence. Meanwhile, major technology companies are seeking alternatives to Nvidia’s expensive GPUs for specific AI workloads, particularly inference.
Google Expands Custom AI Chip Strategy
Under the expanded agreement, Marvell will develop several types of custom silicon for Google’s AI infrastructure. These include AI inference accelerators, storage controllers, networking components and memory-interface technologies.
The chips will support Google’s Tensor Processing Unit ecosystem. However, the agreement does not simply replace Google’s existing chip suppliers. Instead, it gives Google another major partner as demand for AI computing continues to rise.
Google’s warrant also includes performance conditions. Therefore, the full 58.97 million shares will not immediately become Google’s ownership. Most of the warrants will vest as Marvell reaches revenue targets tied to Google’s purchases through fiscal 2033.
If Google exercises the entire warrant, it could become Marvell’s fifth-largest investor. The structure consequently links Google’s future chip purchases with Marvell’s potential equity ownership.
Marvell Could Generate $120 Billion
The agreement could generate approximately $120 billion in revenue for Marvell through fiscal 2033 if Google meets the required targets. Consequently, the partnership represents a significant long-term commercial opportunity for the semiconductor company.
Marvell’s shares rose sharply after the announcement. Meanwhile, Broadcom shares fell as investors assessed the potential impact on Google’s custom AI chip supply chain.
Broadcom has been a major Google partner for custom TPU development. Nevertheless, Google continues to diversify its AI hardware supply chain as computing requirements grow. Broadcom also extended its TPU agreement with Google through 2031 earlier this year.
Marvell is increasingly positioning itself as a supplier of custom silicon for hyperscale AI infrastructure. Earlier this year, Nvidia invested $2 billion in Marvell as part of a broader partnership involving custom XPUs and networking technology.
AI Infrastructure Drives Chip Demand
The Marvell-Google agreement reflects a broader shift toward specialized AI hardware. Instead of relying entirely on general-purpose GPUs, cloud companies are developing chips optimized for their own workloads.
Google’s TPUs are central to this strategy. Moreover, the company has increasingly used custom accelerators to support AI training and inference across its infrastructure.
Inference is particularly important because AI applications now process enormous volumes of user requests. Therefore, specialized hardware can help companies improve performance and control infrastructure costs.
The deal also illustrates how closely linked the AI hardware ecosystem has become. Chip suppliers, cloud providers and AI companies are increasingly using long-term commercial agreements and equity arrangements to secure capacity and align incentives.
As AI infrastructure spending accelerates, Marvell’s expanded Google relationship could become one of the semiconductor industry’s most significant custom-chip partnerships.
