OpenAI's Losing Streak: How Anthropic Seized the AI Crown

OpenAI, the company that ignited the modern AI boom, has spent the past year watching a smaller, more focused rival lap it in the market that matters most. Anthropic — once the scrappy upstart — now boasts faster revenue growth, a valuation nearing $1 trillion, and a coding tool, Claude Code, that has become the default choice for professional software developers. Anthropic is accelerating toward a fall IPO and actively pitching investors on its lead over the ChatGPT maker, while OpenAI has pushed back its own public offering to potentially 2027.

The roots of OpenAI's stumble lie in a strategic misread. CEO Sam Altman bet the company's growth engine on ChatGPT subscriptions, assuming consumers would flock to an AI chatbot as part of their daily lives. That assumption was overtaken by reality: the real money was in selling to developers and the enterprises that employ them. Anthropic understood this earlier. While OpenAI poured resources into splashy consumer bets — a video generator called Sora, consumer hardware, custom chips — Anthropic quietly built Claude Code into a developer phenomenon. Engineers began going on what insiders called 'Claude-benders,' spending hours marveling at the tool's capabilities during the December 2025 holiday period while Altman vacationed in St. Barts.

OpenAI's own coding product, Codex, was supposed to be the answer — but it flopped. Developers found it slow and clunky. The company had trained its reasoning models to excel at high-school coding competitions rather than the messy, open-ended work of real software engineering. Anthropic explicitly took the opposite approach, prioritizing 'real-world tasks.' Internal OpenAI usage of Codex came in lower than expected, executives flagged the issue repeatedly, and when Codex launched publicly in May 2025, software engineers voted with their keyboards — for Claude. OpenAI scrambled to remake Codex to mimic its rival's product, and formed a team called 'coding ninja' to better align models with actual developer needs.

The fallout extended beyond product. Fidji Simo, the former Instacart CEO brought in to quarterback OpenAI's IPO push, warned colleagues before her departure that the company had been chasing 'side quests' and had fallen behind Anthropic in both research and product output. Employees openly challenged leadership, with one internal question reading: 'It feels like Anthropic, a much smaller company by headcount and market cap, is consistently setting the frame technically and culturally, and we are reacting. Why does this keep happening?' Meanwhile, major potential partners like Blackstone chose Anthropic over OpenAI for high-value enterprise deals. Google's rising consumer chatbot also began eating into ChatGPT's growth, causing the company to miss an internal target of one billion weekly active users by year-end — though it eventually crossed that milestone.

OpenAI is now mounting a counteroffensive. President Greg Brockman has been put in charge of overhauling the product lineup. The company hired former Slack CEO Denise Dresser as its first chief revenue officer, struck a distribution deal with Amazon, and released a new model — GPT 5.6 Sol — that won immediate praise from developers and reportedly forced Anthropic to broaden access to its own Fable model in response. OpenAI also launched a 'super app' combining Codex, ChatGPT, and a browser, claiming 10 million users across that product and the standalone Codex app. Altman himself acknowledged the rough patch publicly, posting on X that the past 12 months were 'mostly my fault' while promising the next 12 would be the company's best yet. Whether that confidence is warranted — or whether Anthropic's momentum is too entrenched — is the central question hanging over the AI industry's most consequential rivalry.