Last week, OpenAI made one of the hardest decisions in its history: shutting down Sora, its video generator that caused a sensation just two years ago.
Why does this matter for you as a founder? Because it contains a lesson that took me selling a startup to fully understand.
The Context: $1.4M vs $1.9B
Let me give you the raw numbers:
| Metric | Value |
|---|---|
| Sora Revenue (since launch) | $1.4 million |
| ChatGPT Revenue (same period) | $1.9 billion |
| Sora Downloads Nov 2025 | 3.3 million |
| Sora Downloads Feb 2026 | 1.1 million |
| Drop | 67% |
Sora didn't fail because it was a bad product. It failed because it was a side quest — a secondary project consuming resources without generating sustained traction.
What Fidji Simo Said (And Why It Hit Home)
Fidji Simo is OpenAI's new CEO of Applications. In an all-hands meeting 10 days ago, she said something that made me stop:
"You can't afford to be distracted by side quests."
This comes from a company valued at $840 billion. If they can't afford to spread thin, why do you think your startup can?
While OpenAI was playing with video, robots, hardware, and a browser called Atlas, Anthropic chose one lane.
The Anthropic Strategy That's Winning
Saanya Ojha, an analyst I follow, summarized it perfectly:
"While OpenAI chose optionality, Anthropic chose focus. It didn't try to do everything. It ignored video. It skipped consumer tricks. It didn't chase social media. Instead, it picked one lane — developers and enterprises — and dug a trench."
The result: Anthropic has a 70% win rate in head-to-head matchups when enterprises buy AI for the first time.
Claude Code is capturing developers. Engineers are having "Claude benders" — intense coding sessions with Claude lasting hours.
My Experience With Focus (Or Lack Of It)
At my fintech startup Pago Fácil (sold for $23M in 2021), our focus was clear: payments for e-commerce. We never tried to build our own e-commerce platform or our own invoicing tool — those were obvious temptations we knew to avoid.
But my biggest time sink was different: trying to close corporate clients while still growing with SMBs.
Every time I chased a corporate, I stopped growing with SMBs. Literally. Corporates required endless meetings, customizations, months-long sales cycles. Meanwhile, SMBs — my real market — kept arriving but I wasn't paying attention.
That's exactly what happened to OpenAI: Anthropic chose ONE segment (developers and enterprise) and dug in. OpenAI tried to serve everyone — consumers with Sora, developers with Codex, enterprises with ChatGPT Enterprise — and ended up scattered.
The lesson isn't "don't grow." It's "choose your trench before you dig another."
The Timeline of Sora's Collapse
So you can see what a side quest looks like in slow motion:
- Feb 2024: Sora is revealed, causes a sensation. The hype is real.
- Sep 2025: They launch Sora 2 as a standalone app.
- Nov 2025: Peak downloads (3.3 million).
- Dec 2025: Announce Disney deal worth $1 billion for 200+ characters.
- Feb 2026: Downloads drop 67%.
- Mar 24, 2026: OpenAI announces shutdown. Disney confirms no investment.
From hype to death in two years. With $1 billion from Disney on the table that evaporated.
What OpenAI Is Doing Now
The reset is brutal but strategic:
- Superapp: Consolidating ChatGPT + Codex + Atlas into a single window.
- Focus on coding and enterprise: That's where the real money is. Codex already surpassed $1B ARR.
- Renamed the product team to "AGI Deployment." Not subtle.
- Sam Altman is delegating safety and security to focus on data centers and capital.
And they launched Spud, which according to Altman's internal communication is a "very strong model" that could "really accelerate the economy."
The Question You Should Be Asking
If OpenAI, with $840B in valuation, had to kill a product generating millions because it wasn't core... what side quests are you keeping alive?
I'm not saying never experiment. But there's a difference between:
- Strategic experimentation: Try something small, measure fast, decide in weeks.
- Side quest: Keep investing resources in something "that's already there" even though it doesn't move the needle.
Thomas Husson from Forrester put it clearly: Sora was "a black hole of resources" with "limited monetization."
The Real Cost of Scattered Focus
Henry Ajder, AI expert, summarized why they closed now:
"Given that OpenAI still isn't profitable and pressure from investors and rivals grows, this is money they probably decided they can't keep burning as initial interest fades."
Read that again: OpenAI is not profitable. With $1.9B in ChatGPT revenue. With $840B in valuation.
If they have to make hard decisions about where to focus resources, imagine how critical that is for an early-stage startup.
What To Do With This
- Do the exercise today: List all the projects/features you're working on. Which are your "ChatGPT" and which are your "Sora"?
- Kill something this week: Not next week. This week. The pain of closing something now is much smaller than the cost of keeping it alive 18 months longer.
- Choose your trench: What's the equivalent of "developers and enterprises" for your business? Dig there.
Focus isn't sexy. Saying no to bright opportunities hurts. But as the world's most valuable AI company just demonstrated: there's no alternative.
Want to think through how to prioritize in your startup? Join my community of founders at Cágala, Aprende, Repite — we can help each other figure it out.
📝 Originally published in Spanish at cristiantala.com
Top comments (0)