Amazon and Nvidia are investing $80 billion in OpenAI, but the deals are structured as customer commitments rather than traditional equity investments, ensuring returns through compute infrastructure purchases.
OpenAI announced a massive $110 billion investment round on Friday, led by Amazon ($50 billion), Nvidia ($30 billion), and SoftBank ($30 billion) at a $730 billion pre-money valuation. However, the structure of these deals reveals a complex web of financial engineering designed to benefit the investors while driving up OpenAI's valuation.
The Fine Print on Amazon's $50 Billion
Amazon's investment comes with significant strings attached. Only $15 billion of the $50 billion investment is upfront cash. The remaining $35 billion is contingent on OpenAI meeting specific conditions:
- Renting out two gigawatts of Amazon's Trainium AI accelerators
- Deploying OpenAI models and services in AWS
- Becoming the "exclusive third-party cloud distribution provider for OpenAI Frontier"
The deal also extends OpenAI's existing $38 billion partnership with Amazon, which was originally specific to Nvidia GB200 and GB300 systems, to a total of $100 billion over eight years.
Nvidia's $30 Billion Infrastructure Play
Nvidia's investment follows a similar pattern but focuses on its own hardware. The expanded partnership includes:
- Three gigawatts of inference capacity
- Two gigawatts of training capacity
- Built on Nvidia's Vera Rubin systems
These systems, announced at CES in January, are expected to begin shipping in the second half of 2026. At an estimated $8.4 million per system and 250 kW per rack, five gigawatts of Vera Rubin accelerators would cost roughly $300 billion when accounting for all datacenter infrastructure.
SoftBank's More Traditional Investment
Unlike Amazon and Nvidia, SoftBank's $30 billion investment appears to be a more conventional equity stake. The funding will be paid out in three tranches of $10 billion between April and October 2026. This investment from Masayoshi Son's firm provides OpenAI with working capital while Altman and his team continue their pursuit of artificial general intelligence.
The Economics of Circular Dealing
Both Amazon's and Nvidia's investments are structured to guarantee returns on their capital. Rather than traditional equity investments that dilute ownership, these deals function as pre-paid compute infrastructure purchases at a discount. This approach:
- Ensures a return on every dollar invested
- Doesn't dilute the investors' revenues
- Drives up OpenAI's valuation through large headline numbers
This type of circular dealing has become increasingly common in the AI boom. In October, AMD issued OpenAI a warrant for roughly 10 percent of its stock, contingent on deploying six gigawatts of AMD's Instinct accelerators. Meta received the same offer this week.
The Bigger Picture
Despite annualized recurring revenue reportedly exceeding $20 billion and more than 50 million paying subscribers, OpenAI isn't expected to achieve profitability until at least 2029. The company remains heavily reliant on outside funding to sustain its operations and ambitious growth plans.
These investments highlight the complex relationships forming in the AI industry, where hardware manufacturers, cloud providers, and AI developers are increasingly intertwined through financial arrangements that blur the lines between investment, partnership, and customer commitment.


Comments
Please log in or register to join the discussion