The Great AI Rotation: Claude Hits #1 as OpenAI's Pentagon Deal Backfires
For the first time since the AI boom began, Anthropic's Claude overtook ChatGPT to claim the No. 1 spot on Apple's US App Store — triggered by OpenAI's decision to sign a Pentagon contract that Anthropic publicly refused.
Person holding a smartphone displaying app icons in a dark setting
Key Points
•For the first time since the AI boom began, Anthropic's Claude overtook ChatGPT to claim the No. 1 spot on Apple's US App Store. The surge was triggered directly by OpenAI's decision to sign a Pentagon contract worth up to $200 million — a deal Anthropic publicly refused. According to Anthropic, every single day in the week following the contract announcement was an all-time record for Claude sign-ups, with free active users increasing over 60% since January and paid subscribers more than doubling. [1][2]
•The boycott movement, branded #QuitGPT, has documented over 2.5 million people pledging to leave ChatGPT. Confirmed paid cancellations hit approximately 1.5 million in the first week, representing an estimated $30 million in lost monthly recurring revenue for OpenAI. ChatGPT daily uninstalls spiked 295% on February 28, 2026 — the single largest recorded uninstall event for any major AI application. [3]
•But the story isn't just about ethics — it's about product. Developers are building multi-model stacks rather than relying on a single provider. Claude's 1-million token context window, superior instruction-following, and agentic coding capabilities through Claude Code have made it the go-to for power users. The era of one AI model for everything is ending, and OpenAI's self-inflicted wounds — ads in ChatGPT, defense contracts, and frequent API deprecations — are accelerating the shift. [4]
The contract that started it all
On February 27, 2026, the US Department of Defense disclosed that it had approached both OpenAI and Anthropic about providing AI services for military applications. Anthropic CEO Dario Amodei responded the same day with a statement that would ripple across the entire tech industry: "I cannot in good conscience accede to the Pentagon's request." Anthropic declined the contract, citing ethical red lines on autonomous weapons and mass surveillance. [3]
The next morning, OpenAI announced it had signed the deal. The company framed it as supporting national security and responsible AI deployment in defense applications, emphasizing that the contract covers non-lethal uses — logistics optimization, administrative automation, and cybersecurity tooling. [3]
The contrast was immediate and devastating. Within hours, quitgpt.org launched with the tagline "ChatGPT takes Trump's killer robot deal." The site directed users to cancel subscriptions and listed Claude, Gemini, and Perplexity as alternatives. [3]
•OpenAI still has over 900 million weekly active users, so the boycott isn't existential. But it is structurally important: AI model choice is now a values-based decision for a meaningful segment of users. For developers evaluating which API to build on and which company to depend on for the next five years, the QuitGPT movement is a signal that their users may have strong opinions about that choice. [1][3]
The numbers tell the story
The scale of the backlash has no precedent in AI industry history. Approximately 1.5 million active paid subscribers cancelled in the first week after February 28. At roughly $20 per month per subscription, that represents over $360 million in annualized lost revenue. [3] Mobile analytics firms recorded a 295% spike in ChatGPT daily uninstalls on that single day — the largest recorded uninstall event for any major AI application. [3]
The QuitGPT movement triggered the largest recorded uninstall event for any major AI application.
On the flip side, Claude's downloads surged 37% on Friday and 51% on Saturday of that week. By Monday, March 2, Claude sat at No. 1 among the most downloaded free apps on Apple's US App Store, with ChatGPT falling to second and Google's Gemini in fourth. [1][2]
The top four productivity apps on the App Store were suddenly all AI chatbots — Claude, ChatGPT, Gemini, and Grok — a snapshot of a market that barely existed two years ago. [2] At least one senior OpenAI executive resigned in the days following the announcement, citing the Pentagon deal as the reason. [3]
Who's actually leaving — and why it matters
Here's the part that should worry OpenAI more than the raw numbers: the people leaving aren't casual users. They're the power users who built their entire workflows around ChatGPT — software developers, researchers, writers, and knowledge workers. [3] This is the cohort that pays $20 per month. The cohort that evangelizes the product, writes the tutorials, recommends it to teams and clients. As the ABHS analysis put it: "OpenAI was not losing casual users. It was losing the users who evangelised the product." [3]
Celebrities amplified the message. Pop musician Katy Perry posted on X that she was "done," sharing a screenshot of Claude's pricing page with a red heart around the $20 Pro plan. Actor Mark Ruffalo reframed the monthly subscription as an indirect political contribution, telling his tens of millions of followers that paying for ChatGPT meant funding the company whose leadership was funding MAGA-aligned causes. [2][4]
On Reddit's ChatGPT subreddit, dozens of users posted screenshots documenting their switch — Anthropic receipts alongside OpenAI cancellation confirmations. "Cancel ChatGPT" became a refrain, and some users took a more personal tone, saying Sam Altman's move "crossed the line." [2]
It's not just ethics — it's product
Strip away the Pentagon controversy and you still find a compelling product story. The migration to Claude has been building for months, and the boycott simply accelerated a shift that was already underway. Developers increasingly cite Claude's technical advantages. The 1-million token context window handles massive codebases and document sets that ChatGPT struggles with. Claude's instruction-following is consistently rated as more direct — less of the verbose, over-cautious boilerplate that GPT-5.2 became notorious for. OpenAI CEO Sam Altman publicly acknowledged the company "screwed up" the writing quality in recent updates. [4]
Claude Code, Anthropic's developer-focused agent tool, has been gaining serious momentum. Leaders at tools like Cursor and GitHub have publicly said that for the hardest coding challenges, they use Claude. [4] And then there's the structural shift that makes all of this possible: switching costs have collapsed. According to developer analysis, it now takes roughly 15 lines of code to swap from OpenAI's API to a competitor like Claude. [4] Anthropic even built a feature called "import memory" specifically to ease migration — you don't invest engineering resources in a migration tool unless migration has become a mainstream activity.
Data from Open Router shows the same trend: usage is spreading across models. The era of one AI model for everything is ending fast.
OpenAI's self-inflicted wounds
The Pentagon contract didn't happen in isolation. OpenAI had been accumulating trust damage throughout early 2026. First, the company started testing advertisements inside ChatGPT. For a user base that was already paying $20 per month, seeing ads felt like a betrayal of the implicit contract between a premium product and its customers. [4]
Then there's what developers call "platform churn." OpenAI moves fast but breaks things fast too. The popular Assistants API has a published sunset date of August 2026, and every deprecation forces expensive, time-consuming engineering work just to stay compatible. Each of those forced migrations creates a natural pause where developers seriously evaluate alternatives. [4] The defense contract was the final straw for many, but the kindling had been accumulating for months.
Anthropic gained market share by declining business
Here's the irony that makes this story remarkable: Anthropic's biggest market share gain didn't come from a product launch, a price cut, or a marketing campaign. It came from saying no. By publicly refusing the Pentagon contract, Anthropic created a product differentiator that no advertising could replicate. Claude's brand became synonymous with principled AI development — not because of a PR strategy, but because of a $200 million contract it walked away from. [1][3]
"Since the start of the year, free active users have increased by over 60% and daily signups have quadrupled," an Anthropic spokesperson told Business Insider. "Claude's paid subscribers have also more than doubled this year across Pro and Max plans." [2] The company's Super Bowl ad — which mocked OpenAI's decision to test ads in ChatGPT — had already pushed Claude from 42nd to the top 10 on the App Store. The Pentagon controversy finished the job. [1]
The reality check
Let's keep perspective. OpenAI still has over 900 million weekly active users. Losing 1.5 million paid subscribers is financially significant — $360 million annualized is real money — but it isn't existential for a company valued at $840 billion. [2][3] The enterprise side of OpenAI's business, which includes major corporate contracts and API revenue, wasn't directly affected by the consumer boycott. And GPT-5.4 remains a seriously powerful model that leads in several benchmarks. [4]
But what the boycott revealed is structurally important for the entire industry. AI model choice is now a values-based decision for a meaningful segment of users. When AI tools become infrastructure — embedded in daily workflows, creative processes, and professional output — the ethical posture of the company providing that infrastructure starts to matter. [3] For the first time in the AI era, users actively chose based on ethics, not just features. That changes the competitive dynamics permanently, regardless of whether the boycott numbers hold over time.
What happens next
The AI market in 2026 is genuinely competitive in a way it wasn't eighteen months ago. Claude, Gemini, Perplexity, and Grok all offer capable alternatives to ChatGPT, and the switching costs between them have never been lower. The lesson for OpenAI is uncomfortable but clear: when your product is built on trust — when users give you their documents, their code, their creative work, their daily thinking — the decisions you make about where that power gets deployed matter as much as the model weights themselves.
And for everyone evaluating their AI stack right now, the takeaway from the great rotation is simple: don't marry a model. Marry your evaluations and your workflow. The models will keep changing. Your values probably won't.
On this page
Web · https://www.tomsguide.com/ai/700-000-users-are-ditching-chatgpt-heres-why-and-where-theyre-going
