Vivold Consulting

Confidential documents reveal OpenAI’s cost structure for Microsoft infrastructure

Key Insights

Leaked internal documents detail how much OpenAI pays Microsoft for compute and cloud infrastructure, revealing the economics behind large-scale AI training and deployment. The leak highlights the cost pressures behind frontier-model development.

Stay Updated

Get the latest insights delivered to your inbox

The true cost of AI training comes into focus


A trove of leaked financial documents reveals the scale of OpenAI’s payments to Microsoft, underscoring the enormous infrastructure burden behind frontier models.

What the documents show


- Compute spending is among OpenAI’s largest budget items.
- Training costs scale steeply with model size and iteration frequency.
- Cloud credits and pre-purchase agreements shape cost structures.

Strategic implications


- Highlights how cloud platforms influence the economics of model developers.
- Raises questions about long-term profitability of LLM companies.
- Suggests competitive advantages for vertically integrated players.

Why it matters


- Transparency helps investors understand the real cost curve of frontier AI.
- Will fuel debate over whether frontier-model economics are sustainable.
- Reflects increasing scrutiny on AI company financials and cloud dependencies.

Related Articles

Discovery Loop aims to automate science itself - and Google is funding the startup draining its own bench, as Hassabis exits the DeepMind CEO role

Jeff Dean, Google's chief scientist and 30th employee, is leaving after 27 years to found Discovery Loop, a public benefit corporation using AI to automate scientific research - taking co-founders Sanjay Ghemawat, Quoc Le (Google Brain), and Oriol Vinyals (DeepMind) with him. Google is a founding investor and cloud partner, supplying compute for at least the first year, with Radical Ventures and Khosla Ventures co-leading the seed. In the same announcement, Demis Hassabis steps down as DeepMind CEO to become chairman and Alphabet chief scientist, with Koray Kavukcuoglu taking over Gemini model development. Alphabet stock fell about 4%.

Abbott orders audits of every new project as ERCOT's queue hits 474GW, roughly 90% of it data centres

Governor Greg Abbott announced that all new Texas data-centre projects must be audited by the Public Utility Commission and grid operator ERCOT - a sharp turn for a state whose loose regulation and cheap power made it second only to Virginia for data centres. The trigger is a staggering queue: ERCOT's interconnection requests doubled from 233GW in January to 474GW, about 90% data centres, more than five times the grid's all-time peak demand. Audits will demand power and water use, noise mitigation, light controls, tax-incentive use, and ownership details - after a voluntary survey that most operators simply ignored.

Volta and Bitdeer will build a 133MW Nvidia Vera Rubin data centre in Norway - Anthropic's latest move in a compute land grab

Anthropic has reportedly signed a $10 billion, six-year compute deal with Volta, an AI cloud startup founded only earlier this year, per Bloomberg. Volta is partnering with crypto-mining firm Bitdeer to develop the data centre - located in Norway, delivering 133 megawatts, and running Nvidia's Vera Rubin architecture - and is a member of Nvidia's Cloud Partner programme. It caps an aggressive capacity spree that also includes recent compute deals with SpaceX and Amazon, as Anthropic races rivals for the scarcest input in the industry.