Nvidia is negotiating to become one of the anchor investors in Anthropic’s initial public offering, in a transaction that could reach historic proportions. The artificial intelligence company seeks to raise up to US$100 billion through its initial public offering and could achieve a valuation approaching US$2 trillion. Nvidia is considering contributing up to US$10 billion, although the terms are still under negotiation.
The transaction takes on special significance because Nvidia would not simply be a financial investor. Its accelerators constitute a fundamental part of the computational infrastructure used by Anthropic to train and execute its artificial intelligence models. Nvidia’s equity entry would therefore deepen an existing relationship between hardware manufacturer and computational capacity integrator.
Anthropic is also diversifying its processing sources. Amazon and Google are among its main backers and compute capacity providers, while the company continues expanding its infrastructure to sustain Claude’s growth.
Capital Begins to Follow the Chips
To develop increasingly larger models, talent and algorithms alone are not enough: infrastructure consisting of accelerators, memory, high-speed networks, data centers, cooling systems, and power supply is required.
Anthropic needs capital to expand that infrastructure; the infrastructure needs semiconductors; semiconductors generate compute capacity; and that capacity enables production of AI services capable of generating new revenue. Nvidia’s proposed investment fits precisely within that economic circuit.
The phenomenon also explains why major technology companies are participating simultaneously as suppliers, customers, and investors. Amazon, Google, and Microsoft have built relationships of this type around Anthropic, while Nvidia seeks to deepen its exposure to the growth in demand for artificial intelligence compute.
The consequence for the industry is significant: the AI market is no longer financing only software companies and is beginning to capitalize complete technological infrastructure chains. Each new advanced model requires a physical platform capable of training and executing it at scale.
An IPO That Will Price AI Infrastructure
Anthropic arrives at this potential IPO after extraordinary growth. Reuters reports that its annualized revenue reached approximately US$65 billion in July 2026, compared to a much lower figure at the close of 2025. The company projects revenue between US$190 billion and US$200 billion for 2028.
But there is an important difference between a private valuation and a public valuation. The approximately US$2 trillion currently appearing in conversations does not yet constitute a definitive market valuation. The final size of the offering, share price, and valuation will depend on the transaction terms and the regulatory process.
Precisely for this reason, Nvidia’s participation may prove relevant. A strategic investor of that size would provide an additional signal to the market about Anthropic’s importance within future AI infrastructure demand. The semiconductor company would be betting not only on the growth of a model developer, but on the growth of the amount of compute that the AI economy will require.
The New Financial Architecture of Artificial Intelligence
The move allows observation of a structure that normally disappears behind news about new models: capital, semiconductors, data centers, and energy are forming a single economic system.
A valuation approaching US$2 trillion cannot be interpreted exclusively from the software perspective. Behind it exists a physical infrastructure that must consume electricity, dissipate heat, transport data, and maintain continuous availability. Anthropic’s growth implies, by extension, growth of that infrastructure.
Nvidia’s potential investment shows how accelerator manufacturers can participate directly in financing the companies that drive demand for their own products. Advanced AI is thus creating an integrated technological-financial system, where model expansion depends on compute expansion and this, in turn, requires capital, energy, and industrial infrastructure.
SOURCE: https://www.reuters.com/
