OpenAI just inked a $38 billion deal with Amazon Web Services, giving it access to massive compute clusters—but the company still hasn’t proven it can generate sustainable revenue for its partners. Instead, it continues to Pac‑Man its way through billions in cloud commitments while leaving investors to wonder when, if ever, the economics will make sense.
Hot on the heels of its Microsoft entanglement, OpenAI has announced a multi‑year, $38 billion partnership with Amazon Web Services (AWS). The deal, revealed in an OpenAI press release, gives the company immediate access to AWS’s EC2 UltraServers, powered by hundreds of thousands of NVIDIA GPUs and scalable to “tens of millions of CPUs” by 2027.
On paper, this is a staggering expansion of OpenAI’s compute capacity. AWS CEO Matt Garman framed it as a natural fit: “As OpenAI continues to push the boundaries of what’s possible, AWS’s best‑in‑class infrastructure will serve as a backbone for their AI ambitions.” Sam Altman, OpenAI’s CEO, was equally grand: “Scaling frontier AI requires massive, reliable compute. Our partnership with AWS strengthens the broad compute ecosystem that will power this next era.”
But here’s the rub: OpenAI still hasn’t proven it can deliver solid revenue numbers for its partners. As we explored in the previous post, Microsoft has already absorbed billions in losses tied to its OpenAI stake, even as it trumpets Azure growth. Now Amazon is stepping into the same dance, offering up infrastructure at unprecedented scale while hoping that OpenAI’s magic eventually translates into real returns.
The pattern is becoming clear. OpenAI doesn’t just raise money, it consumes it. Microsoft’s $13 billion commitment, now Amazon’s $38 billion deal, and countless other partnerships all feed into the same cycle: OpenAI secures compute at industrial scale, burns through cash, and promises that the future will justify the present. It’s a strategy that looks less like sustainable business and more like Pac‑Man gobbling up quarters in an arcade, except the quarters are billions in cloud credits.
For AWS, the upside is obvious in the short term: OpenAI’s workloads will drive massive demand for its infrastructure, and OpenAI’s models are already available on Amazon Bedrock, where thousands of customers, from Peloton to Thomson Reuters, are experimenting with generative AI. But the long‑term question remains: will OpenAI ever generate enough revenue to justify these colossal commitments, or are Amazon and Microsoft simply subsidizing its burn rate in exchange for bragging rights?
The dot‑com parallels are hard to ignore. Just as AOL Time Warner once promised a new era of synergy before collapsing under the weight of its own hype, OpenAI is building a future on the assumption that scale alone guarantees success. The AWS deal may buy OpenAI more time and more GPUs, but it doesn’t change the underlying math: billions in losses, billions more in commitments, and no clear path to profitability.
For now, OpenAI remains the belle of the AI ball, charming tech giants into underwriting its ambitions. But as the bills pile up, the question isn’t whether OpenAI can keep eating, it’s whether anyone will still want to feed it.
