Summary
Class action lawsuit filed October 13, 2025 in the US District Court for the Northern District of California against Microsoft. ChatGPT Plus subscribers allege Microsoft used its exclusive Azure cloud computing agreement with OpenAI to restrict compute supply, artificially inflate ChatGPT prices by up to 200x versus competitors, and extract supracompetitive profits while advancing its own Copilot products. The case resolved when OpenAI gained access to Google’s cloud in June 2025 — prices immediately dropped 80%.
Key Points
- Microsoft invested $13B+ in OpenAI since 2019 and required OpenAI to use Azure exclusively for compute
- Plaintiffs allege Microsoft “mercilessly choked OpenAI’s compute supply” to inflate prices and delay product innovation
- ChatGPT API tokens were “100 to 200 times competitors’ prices on a per-token basis” during a 2025 price war
- DeepSeek disruption (early 2025): DeepSeek trained for $6M (vs. OpenAI’s billions) and released at fraction of cost; triggered global price war
- OpenAI could not cut prices because Microsoft’s exclusivity clause prevented it from getting alternative compute
- February 2025: DeepSeek cut “off-peak” pricing by 75%; OpenAI was still 136x the price of competition
- June 2025: OpenAI secured compute from Google, ending Microsoft exclusivity; prices immediately dropped 80%; long-delayed features launched
- As of July 2025: OpenAI held 82.65% of consumer generative AI market; HHI score of 6,916 (far above DOJ “highly concentrated” threshold)
- Relief sought: treble damages, restitution, injunction against future exclusivity terms, dismantling of Microsoft’s compute control over OpenAI
- UK parallel: competition lawyer filing £2B compensation claim against Microsoft over cloud overcharging
Newsletter Angles
- Compute infrastructure as the new chokepoint: this case is about who controls the physical layer of AI — GPUs, data centers, cloud agreements. Microsoft’s leverage over OpenAI via Azure is a new kind of structural power
- “AI Computational Barrier to Entry”: the complaint formally names this dynamic. Microsoft and Meta spent $9B on Nvidia H100 chips in 2023; Microsoft planned 1.8M GPUs by end of 2024. Compute concentration is market power
- The DeepSeek shock as regulatory proof: the price collapse when Microsoft’s grip loosened (80% drop) proves the artificial scarcity thesis — markets were working, but infrastructure control was preventing it
- This case could define antitrust doctrine for AI: if courts accept CGAI as a distinct concentrated market, the implications for AI platform regulation are enormous
Entities Mentioned
- OpenAI — victim of artificial scarcity; controlled via Azure exclusivity
- Meta — competitor; spent $9B on H100 chips alongside Microsoft in 2023
- Platform Antitrust — this lawsuit is the leading AI antitrust case
Concepts Mentioned
- Platform Antitrust — compute infrastructure as new antitrust frontier; AI market concentration at 82% for OpenAI
- Tech-State Conflict — antitrust law as regulatory check on AI company power
- Dynamic Pricing AI — the artificially high token pricing is the harm; compute scarcity is the mechanism
Quotes
“OpenAI’s flagship model was 136x the price charged by the competition” during the DeepSeek price war
“Microsoft mercilessly chok[ed] OpenAI’s compute supply to inflate prices, delay product innovation, and preserve its own market advantage”
“After OpenAI obtained compute from Google, ending its Microsoft exclusivity… OpenAI’s high-end model prices ‘immediately dropped 80%‘“
Notes
Three separate sources covered this lawsuit (PYMNTS, Medianama, Computing UK). The Medianama piece (this source) is the most substantive — it includes the full allegation structure, the DeepSeek disruption mechanism, and the broader market power analysis. The Computing UK piece adds the UK parallel claim. PYMNTS is too thin (mostly sidebar content).