Earlier this month, Fidji Simo laid out a sharper strategy for OpenAI: the startup’s everything era is over. Now, it’s about focus.
“We cannot miss this moment because we are distracted by side quests,” Simo said at an all-hands meeting, according to Business Insider’s must-read profile of OpenAI’s new CEO of applications.
One of the first “side quests” to go is the Sora AI video-generation app. The product was a breakout hit when it launched last year. So why shut it down?
The answer comes down to a growing constraint across the AI industry: compute.
“Given the frantic search for more compute across the industry, OpenAI is prioritizing its greatest growth engine — ChatGPT,” Bernard Golden, CEO of Navica, a Silicon Valley-based tech analysis, consulting, and investment firm.
Demand for AI compute has surged in recent months. Data from OpenRouter’s model platform shows usage more than tripled in 2.5 months, a sign of explosive growth.
OpenRouter
At the same time, supply isn’t keeping up. Building new data centers has become harder due to local opposition, energy constraints, and shortages of key components such as memory chips.
OpenAI has moved aggressively to secure capacity, committing hundreds of billions of dollars to data center and chip deals. Even so, it still doesn’t have enough compute to support everything it’s building.
The company hinted at this reality in a statement confirming the Sora shutdown: “As we focus and compute demand grows, the Sora research team continues to focus on world simulation research to advance robotics that will help people solve real-world, physical tasks.”
Every day, OpenAI must decide which projects get access to scarce compute resources. That dynamic isn’t unique — it’s playing out across Silicon Valley. Before the latest surge, OpenAI CFO Sarah Friar had spoken about these tradeoffs.
Simo’s background suggests how she’ll approach this challenge. At Instacart, she used advertising to drive profitability and take the company public. Before that, she helped Facebook build a mobile ads business during a critical transition period leading up to its IPO.
Now she faces a similar situation at OpenAI. An IPO is expected within the next year or two, and the company is still losing billions annually.
That puts pressure on identifying projects that consume large amounts of compute without generating revenue. After years of experimentation, OpenAI has plenty of those.
Sora is a prime example. The app allowed users to create AI-generated videos, an impressive and widely used tool. It didn’t make money, and video generation is among the most compute-intensive AI tasks.
“We have been quite amazed by how much our power users want to use Sora, and the economics are currently completely unsustainable,” Bill Peebles, the head of Sora, wrote on X at the end of October.
Shutting down the Sora app will free up significant compute resources. OpenAI can now redirect that capacity toward products with clearer revenue potential, such as enterprise offerings like Codex or advertising within ChatGPT.
In an environment defined by scarcity, focus is more important than ever, especially for a company like OpenAI that’s losing so much money.
Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at [email protected].
