OpenAI killed Sora Tuesday after just six months, shutting down both the consumer app and developer API as the company refocuses ahead of its planned IPO. The move comes after Sora downloads collapsed from 3.3 million in November to 1.1 million by February—a 67% drop that made the resource-intensive video model an easy target for budget cuts. CFO Sarah Friar told CNBC the company needs to be "ready to be a public company," signaling an end to OpenAI's spray-and-pray approach under Sam Altman's leadership.
This represents a major strategic pivot from OpenAI's Y Combinator-style betting on everything—browsers, hardware, robots—toward two core areas: a ChatGPT "superapp" and enterprise coding tools. The shift makes business sense: Codex just crossed $1 billion in annualized revenue and is growing, while Sora burned through expensive GPU resources for a shrinking user base. Multiple sources describe OpenAI's previous "bottom-up" culture as spreading resources too thin, exactly what public market investors hate to see.
As I reported when covering Sora's initial struggles, the writing was on the wall when Disney pulled their $1 billion partnership deal last month. The video AI market remains brutally competitive, with Runway and Pika eating OpenAI's lunch while requiring far less computational overhead. OpenAI's bet that combining ChatGPT, Codex, and Atlas into a unified interface will create their long-promised "super assistant" is ambitious but unproven—previous agent features like Operator saw limited adoption.
For developers, this consolidation could actually be positive. Instead of managing separate APIs for different OpenAI products, the unified platform should simplify integration. But if you built anything on the Sora API, start planning your migration now—OpenAI's focus era means less experimentation and more pressure on remaining products to perform." "tags": ["openai", "sora", "ipo", "codex