Google Corporation has obtained the exclusive right to acquire up to 58.97 million shares of chip manufacturer Marvell Technology. This is part of an expanded partnership to develop AI processors, and judging by the documents filed with the SEC, we are witnessing not just a purchase, but a complex financial instrument designed to synchronize the interests of the two giants for a decade ahead.
The deal is indicative of the entire industry: the connections between hyperscalers, model developers, and chipmakers are no longer simple transactions. Now they are an intertwining of guarantees, options, and joint financing of capital-intensive AI infrastructure.
Warrant Terms: A Bet on Scale
The collaboration covers key components of TPU infrastructure — Google's proprietary accelerators, including inference processors, storage controllers, networking solutions, and memory interfaces. However, the main point of interest is the option structure. Only about 1.36 million shares will be unlocked evenly over the first year. The rest of the package is divided into 240 tranches, each of which activates only after Google's purchases bring Marvell an additional $500 million in revenue.
For full exercise of the warrant, cumulative sales volume must reach an astronomical $120 billion. The exercise price is set at $206.58 per share, and the option expires on August 18, 2033. If Google fully realizes its potential, it will enter the top five largest shareholders of Marvell. The market has already reacted: Marvell shares moved up, while competitor Broadcom's stock lost more than 5%.
This is not a displacement of Broadcom, but rather diversification. Google, by scaling up its own TPUs, is clearly seeking to reduce its dependence on Nvidia and gain more control over the entire supply chain — from silicon to chip interconnect systems.
Financial Circles of the AI Economy
We are seeing the formation of a new model where chip manufacturers effectively finance their own demand. Nvidia, for example, provided $105 billion in guarantees for OpenAI's data center in Ohio, additionally investing $1.5 billion in the SB Energy project. This creates an unusual chain: Nvidia supports the construction of a facility that will become a major buyer of its own accelerators. CEO Jensen Huang rejects accusations of circular financing, but such structures inevitably increase the interdependence of market participants.
The scale of investment already exceeds $730 billion this year, and Wall Street is actively joining the financing. Nvidia, together with Apollo, BlackRock, and KKR, plans to raise more than $500 billion in third-party capital. This transforms AI from a purely software story into a capital-intensive industry where competitive advantage increasingly depends on the ability to cheaply finance infrastructure construction.
My view: such schemes are an inevitable evolution of the market, but they harbor systemic risks. The Bank for International Settlements has already warned about the opacity of such agreements and the risks of asset rehypothecation. While the market is growing, these mechanisms work, but at the first serious downturn, we could see a chain reaction that hits all participants.