OpenAI Signs Pentagon Deal - 2.5 Million Users Quit in 48 Hours

What You'll Learn

• The full structure of OpenAI's U.S. Department of Defense contract and what it covers
• Why the user backlash was faster and larger than any previous AI controversy
• The market effects: Claude at #1 on App Store, 340% Anthropic subscription surge, what happened to OpenAI's revenue
• Four enterprise implications from this week's events
• How to build a vendor ethics framework before the next trigger event

On February 27, 2026, OpenAI announced a contract with the U.S. Department of Defense providing AI capabilities across classified and unclassified Pentagon networks. The scope covers intelligence analysis, logistics optimization, and reconnaissance planning support. Within hours, #QuitGPT was trending globally.

What followed was unlike anything the AI industry had seen before: 2.5 million account deletions in 48 hours, a 295% spike in iOS uninstalls, Claude rising to #1 on the App Store, and a 340% surge in Anthropic subscriptions. OpenAI's original charter had explicitly prohibited military weapons applications - language that was revised in a late 2024 charter update. Users who joined because of that mission felt the terms had changed.

This is a PromptHacker Premium article.

The full market analysis, enterprise implications, and contingency framework are available to Premium subscribers.

The Scale of the Migration

App uninstall tracking services documented the migration in real time. Over 48 hours from February 27 - 28:

• 2.5 million ChatGPT account deletions and deactivations
• 295% spike in ChatGPT iOS uninstalls vs. prior 7-day average
• Claude to #1 on the App Store within 24 hours - the first time Claude reached #1
• 340% surge in Anthropic paid subscriptions; waitlist temporarily activated
• Multiple OpenAI employee departures made public and covered by major tech publications

Enterprise and API revenue - where OpenAI's growth lives - was largely unaffected. The 2.5 million users who left were primarily consumer-tier subscribers. The enterprise tier, which requires multi-year contracts and IT integration, has far higher switching costs. The same week OpenAI faced its largest consumer exodus, its valuation hit $840B on institutional capital.

Four Enterprise Implications

1. AI vendor ethics decisions are a procurement variable. Legal and procurement teams need to evaluate AI vendors on ethics commitments the same way they evaluate data security and SLA terms. OpenAI's charter revision was public; most procurement teams were not tracking it. That oversight creates organizational exposure when the commitment acts.

2. Defense contracts introduce data-handling questions. If your organization uses OpenAI's enterprise API, the Pentagon contract creates new questions about the data-handling framework and isolation model. The initiative to ask them belongs to corporate legal and IT security.

3. Platform switching costs are lower than assumed. 2.5 million users who migrated in 48 hours demonstrated that AI tool switching is fast at the individual level. For organizations, the switching cost is real - but not as high as most IT teams have estimated.

4. "Values-aligned AI" is a competitive positioning claim. Anthropic's 340% surge was not driven by a product improvement. "We don't do military weapons" is now a positioning statement with measurable market impact. Expect this framing to intensify in enterprise AI sales conversations.

Action Steps

• Review which OpenAI products your organization uses and document what workflows depend on them. This is the baseline for any contingency or ethics review.
• Brief your legal and IT security teams on the Pentagon contract. Ask them to determine whether any contractual or regulatory obligations require a vendor review.
• Pull and read OpenAI's current enterprise terms of service, specifically the data handling and acceptable use sections. Flag any changes from prior versions.
• Build a vendor ethics trigger framework. Define what categories of decisions by any AI vendor would prompt you to evaluate alternatives - and document it now, before a trigger event occurs. This week's Pro Tip covers the full framework.
• Have a brief, direct conversation with your team. You do not need to take a policy position. You do need to demonstrate awareness of the week's events.