Nvidia’s reported discussions to anchor Anthropic’s planned $100 billion IPO represent more than a massive capital injection; they signal the maturation of a closed capital loop. In this arrangement, the landlord—the provider of the essential compute infrastructure—is now financing the tenant’s exit. By positioning itself as an anchor investor, Nvidia is not merely selling hardware; it is ensuring that the capital it provides to Anthropic returns to its own balance sheet, effectively underwriting the very entity that consumes its most valuable output.
A Record-Breaking Ambition
The scale of this proposed raise is difficult to overstate. Anthropic is reportedly seeking up to $100 billion at a valuation near $2 trillion. To put this in perspective, the largest IPO in history, Saudi Aramco’s 2019 offering, raised $29.4 billion. Anthropic’s target is 3.4 times that size. While these are preliminary talks and not a done deal, the ambition reflects a valuation trajectory that has moved from approximately $350 billion in November 2025 to nearly $965 billion by May 2026, with a $2 trillion target now in sight within an 18-month window.
From Vendor to Partner
The relationship between Nvidia and Anthropic has evolved rapidly. As of November 2025, Nvidia had already committed up to $10 billion in investment alongside 1 gigawatt of compute capacity utilizing Grace Blackwell and Vera Rubin architectures. This transition from a compute vendor to a capital partner is a logical progression of the compute landlord thesis, which posits that value in the AI era accrues primarily to those who control the physical and digital infrastructure. When the landlord provides the capital for the tenant to go public, the dependency becomes absolute.
The Logic of Forward-Looking Multiples
The valuation logic underpinning this IPO relies on an unusually forward-looking approach. Bankers are applying enterprise value-to-revenue multiples to 2028 forecasts, which project revenue between $190 billion and $200 billion. This is a significant leap from the company’s revenue progression, which saw growth from $9 billion at the end of 2025 to $47 billion by May 2026. The market is essentially pricing in a 10-fold annual growth rate, a bet that requires sustained, massive-scale compute access to maintain.
Justifying the IP Value
The question is whether the underlying technology can support such a valuation. Anthropic’s recent distillation disclosure on September 10, which highlighted 200 million exchanges, serves as the primary evidence for the value of its intellectual property. This data suggests that the company is successfully converting compute power into usable, scalable intelligence. However, the part that gets hidden is the sheer cost of maintaining that scale, which is why the capital bifurcation thesis—where infrastructure mega-rounds are separated from vertical specialists—is so critical to understanding this market.
The Broader IPO Wave
This move must be viewed within the context of the broader IPO wave, where three major labs are pursuing three distinct exit strategies. The connection to OpenAI’s recent $122 billion raise is instructive; while Nvidia contributed $30 billion to that effort, the investment was heavily weighted toward compute access rather than liquid cash. Anthropic’s IPO, by contrast, seeks to formalize this relationship through a public market exit, potentially setting a new standard for how AI labs capitalize their operations.
What to Watch
As the preliminary talks continue, several factors will determine the success of this offering. Investors should monitor the split between cash and compute in the final investment terms, as this will reveal the true nature of Nvidia’s commitment. Furthermore, regulatory scrutiny will be intense, given the sheer size of the IPO and the potential for market concentration. The IPO roadshow reception and the timing—expected before the November 2026 US midterm elections—will provide a clear signal of whether the public markets are ready to absorb sovereign-scale capital for agentic infrastructure.