In brief
- Microsoft allegedly used an exclusive Azure agreement from 2019 to control compute supply to OpenAI, keeping ChatGPT prices artificially high while developing competing AI products.
- When OpenAI began buying compute from Google in June 2025, token prices reportedly dropped 80 percent within weeks, which plaintiffs say is evidence of the restraint.
- The lawsuit seeks damages for overcharges from November 2022 to February 2025, warning Microsoft still holds contractual power to throttle OpenAI’s supply.
ChatGPT users have accused Microsoft of “mercilessly” choking OpenAI’s compute supply through an exclusive cloud agreement, artificially inflating AI subscription prices while simultaneously rushing its own competing products to market.
The lawsuit, filed Monday in San Francisco federal court, alleges Microsoft “secretly turned an investment into a stranglehold,” using its Azure cloud dominance to restrict the computational resources needed to run ChatGPT, keeping prices at levels reaching “100 to 200 times” competitors’ rates during a February 2025 AI price war.
Eleven ChatGPT Plus subscribers brought the case, alleging they overpaid for subscriptions and received degraded service from November 2022 through February 2025 due to Microsoft’s anticompetitive conduct.
The lawsuit calls the case “a sequel to U.S. v. Microsoft,” invoking the company’s 1990s antitrust battles and characterizing it as a “recidivist violator” that has “ported the same exclusionary playbook into AI.”
Microsoft’s 2019 deal gave exclusive rights to supply Azure compute to OpenAI’s commercial models, effectively giving the company contractual control over a “horizontal competitor’s supply chain” while it built its own generative AI products, according to the filing.
The plaintiffs define a new antitrust market, the Consumer Generative AI Market, encompassing subscription products like ChatGPT Plus, Claude Pro, Gemini Advanced, and DeepSeek Chat.
Microsoft reportedly holds a 49% stake in OpenAI’s for-profit arm and takes 20% of its paid-product revenue, the lawsuit alleges, effectively “profiting twice—first from compute sales, and again from the very AI product it constrains.”
When OpenAI began purchasing compute from Google Cloud in June 2025, ending Microsoft’s exclusivity, ChatGPT token prices dropped 80 percent within weeks, according to the complaint.
A ‘powerful experiment’
The complaint calls it a “powerful natural experiment” showing Microsoft’s “anticompetitive restraint,” noting that before the shift, ChatGPT users faced “poor quality, unreleased innovations, and slow response times.”
“On proof, the strongest evidence would be the exclusive agreement itself,” Navodaya Singh Rajpurohit, legal partner at Coinque Consulting, told Decrypt. “If that document shows Microsoft held or used control over OpenAI compute, that is primary proof. If it is not available, internal emails and capacity records can still carry the case.”
“Restraint of trade turns on control,” Rajpurohit said, adding the claim is strong if Microsoft actually exercised or clearly held that control, but weaker if it did not.
The complaint also alleges Microsoft “still retains the contractual ability to restrict OpenAI’s compute purchases,” a power that lingers “as a sword of Damocles over OpenAI, wielded by one of its principal competitors.”
Plaintiffs are seeking monetary damages and a permanent injunction barring Microsoft from exclusive compute deals with OpenAI, along with disclosure of internal communications on compute supply, pricing, and integration.
“The court can do more than fine the company,” Rajpurohit explained. “It can order changes to the OpenAI arrangement, remove exclusive terms, and set guardrails that prevent future choke points.”
The class-action arrives amid shifts in the Microsoft-OpenAI relationship. Japan’s SoftBank is reportedly negotiating to invest up to $25 billion in OpenAI, surpassing Microsoft’s $13 billion stake. Meanwhile, Microsoft has agreed under the Stargate project to end its exclusive cloud provider status.
Microsoft and OpenAI did not immediately respond to Decrypt’s requests for comment.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
in well organized HTML format with all tags properly closed. Create appropriate headings and subheadings to organize the content. Ensure the rewritten content is approximately 1500 words. Do not include the title and images. please do not add any introductory text in start and any Note in the end explaining about what you have done or how you done it .i am directly publishing the output as article so please only give me rewritten content. At the end of the content, include a “Conclusion” section and a well-formatted “FAQs” section.
