Amazon is in early-stage negotiations to invest approximately $10 billion in OpenAI, a transaction that would value the artificial intelligence company at over $500 billion and represent the latest escalation in infrastructure-driven competition within the AI sector.
The potential collaboration comes just weeks after OpenAI’s $38 billion, seven-year cloud partnership with Amazon Web Services (AWS).
This Amazon OpenAI deal encompasses both direct capital injection and OpenAI’s adoption of Amazon’s proprietary Inferentia and Trainium AI chips for model training, which marks a shift in OpenAI’s hardware diversification strategy, as the ChatGPT-maker only made use of Nvidia chips for training its models prior to these negotiations.
While the talks remain preliminary and terms are fluid, the negotiations make note of the increasing reorientation of the AI industry’s power structure, as companies in the sector now operate using the circular financing structure.
The circular financing structure in AI is a funding model where a group of companies in the same industry invests in each other and then uses that capital to purchase products or services from their investors, effectively recycling money within a closed ecosystem. So far, this model has invited criticisms, with the most critical one being the fact that it could cause a burst in the AI bubble currently going on.
So far in the AI industry, this widely known funding trajectory is a sign of the industry’s capital requirements for frontier AI development. For example, OpenAI announced projections of over $1 trillion in infrastructure spending over the next decade, a figure exceeding the total venture capital invested in all U.S. startups annually.
It is why this humongous capital need justifies OpenAI having simultaneous partnerships with Microsoft, Amazon, Oracle, SoftBank, and other funding companies or partners, because capital injection cannot come from one place.
This Amazon OpenAI deal must also be understood within OpenAI’s deliberate pivot away from singular vendor dependence. In October, Microsoft-backed OpenAI entered a new chapter with its foremost partner, Microsoft, that allowed it to complete a restructuring from a capped-profit nonprofit subsidiary into a for-profit public benefit corporation (PBC).
Conversely, the Amazon OpenAI deal is a sign of the fast erosion of Nvidia’s monopoly position. While Nvidia’s 80% market share remains threatening, AWS’ Inferentia and Trainium chips adoption by OpenAI would represent the highest-profile validation of custom silicon’s viability for frontier AI model training.
This follows the recent Google’s TPU v5 deployments and Microsoft’s emerging neural processor initiatives, which are all challenges to Nvidia’s dominance that could reshape chip procurement decisions across the AI industry.
For the AI industry at large, the proposed Amazon OpenAI deal is a sign that there has been a recent shift in AI companies opening up their supply chains for custom chips for their respective model training. For now, the era of companies engaging in single-vendor dependence is graduating into competition-driven infrastructure procurement.
