Google Corporation has secured the right to acquire up to 58.97 million shares of chipmaker Marvell Technology. This move is part of an expanded partnership to develop AI processors and was disclosed in an 8-K filing with the U.S. Securities and Exchange Commission.
The deal is a striking indicator of how the financial architecture of the AI industry is becoming increasingly complex. Chipmakers, hyperscalers, and model developers are moving from simple transactions to intricate cross-guarantees, options, and joint capital raising to build computing infrastructure.
An option tied to purchases
The collaboration between Google and Marvell covers key components of the TPU infrastructure — Google's proprietary AI accelerators. This includes inference processors, storage controllers, networking solutions, memory interfaces, and near-memory computing.
Of the total share package, only about 1.36 million shares will be unlocked in equal portions over the first year. The rest of the warrant is strictly tied to the volume of Google's orders. The shares are divided into 240 tranches, each of which activates after purchases generate an additional $500 million in revenue for Marvell. To fully unlock the option, sales volume must reach $120 billion.
Google can exercise the shares at a price of $206.58 until August 18, 2033. If the warrant is fully exercised, the company could enter the top five largest shareholders of Marvell. The market has already reacted: Marvell shares rose, while shares of competitor Broadcom fell by more than 5%.
A strategic maneuver against dependence on Nvidia
This step is part of Google's broader strategy to reduce its dependence on Nvidia accelerators and lower the cost of AI computing. Marvell will be involved not only in creating the TPUs themselves but also in developing components to combine massive arrays of chips into unified computing systems. The strengthening of its own infrastructure coincided with an internal restructuring of Google's AI division, where executives linked to Google Cloud gained more influence.
Similar schemes are becoming a trend. In October, AMD entered into a similar agreement with OpenAI, combining large-scale processor purchases with the opportunity for the ChatGPT developer to gain a stake in the chipmaker.
Chipmakers finance their own demand
An even more illustrative example is the recent deal between Nvidia and OpenAI. Nvidia agreed to provide a guarantee of up to $105 billion for OpenAI's data center in Ohio, which is being built by SB Energy (owned by SoftBank). In the event of an OpenAI default, Nvidia would compensate the difference between the guaranteed value and the amount from re-leasing or selling the infrastructure. Additionally, Nvidia is investing $1.5 billion in SB Energy and will become the exclusive supplier of computing infrastructure.
Such structures create unusual chains: a chipmaker finances the construction of a data center that then becomes a major buyer of its own accelerators. This raises questions about circular financing, although Nvidia's management rejects such accusations.
AI becomes a capital-intensive industry
Total spending by the largest technology companies on AI this year will exceed $730 billion. Against this backdrop, Nvidia, together with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, is creating platforms to attract more than $500 billion in third-party capital. The chipmaker itself is prepared to provide guarantees of up to $125 billion, calling AI chips "revenue-generating assets."
Private funds have already begun treating computing infrastructure as a separate asset class. For example, Apollo and Blackstone are participating in financing the $35 billion expansion of Anthropic's capacity using Broadcom equipment.
IMD Business School professor Amit Joshi notes that the competitive advantage in AI increasingly depends less on the model itself and more on who can finance, build, and operate infrastructure more cheaply. The Bank for International Settlements warns of macro-financial risks, particularly opaque agreements and the rehypothecation of assets.
My view: We are witnessing a fundamental shift: the AI industry is transforming from a software-based sector into a capital-intensive one, where financial engineering becomes as much a competitive advantage as algorithms. These cross-guarantees and options allow infrastructure to be built faster, but they create interdependencies that could become a source of systemic risk at the first serious downturn.