OpenAI boosts compute margins to 70% as AI costs squeeze
TL;DR
According to reporting from The Information relayed by Storyboard18, OpenAI has increased its internal compute margin on paid products to about 70% as of October 2025, up from roughly 52% at the end of 2024. The company is still unprofitable and is reportedly in early discussions to raise at least $10 billion from Amazon, potentially including use of Amazon chips.
About this summary
This article aggregates reporting from 2 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Race to AGI Analysis
The headline here isn’t just that OpenAI is getting more efficient; it’s that the company appears to be actively managing its unit economics while still chasing ever‑larger models. A jump in compute margin from the low‑50s to around 70 percent in under a year suggests aggressive optimisation of inference stacks, pricing, and product mix. That’s what you’d expect from a company trying to prove it can be a durable platform business, not just a science project propped up by mega‑funding.
If OpenAI can keep ratcheting up margins while sustaining model quality, it buys itself crucial room to invest in the next generation of training runs. But the rumoured talks with Amazon for another $10 billion and alternative chips are a reminder that even at improved margins, frontier‑scale AGI work remains brutally capital‑ and compute‑hungry. For competitors like Anthropic and Google, this raises the bar: it’s no longer enough to have a great model; you need a credible path to 60–70%+ gross margins on compute just to keep funding the next leap.
Who Should Care
Companies Mentioned
Related Deals
OpenAI is in early-stage talks to raise up to $100 billion in new funding that could lift its valuation to roughly $750–830 billion, according to multiple media reports citing unnamed sources.
Amazon→DOE signed nonbinding MOUs with 24 AI and compute organizations to apply advanced AI and high-performance computing to Genesis Mission scientific and energy projects.
Google Public Sector and Google DeepMind will provide Gemini-based AI platforms and tools to DOE’s Genesis Mission, giving all 17 U.S. national laboratories secure access to frontier models such as Gemini for Government and the AI co-scientist system.
Preliminary talks for a potential funding round of up to $100 billion that would value OpenAI around $750 billion.
Waymo is reportedly negotiating a funding round exceeding $15 billion at around a $100 billion valuation to expand its robotaxi operations.
