The Gist
- OpenAI shifts focus to products. Despite leading in AI research, OpenAI is positioning itself primarily as a product company, relying on ChatGPT for revenue.
- Open-source AI challenges dominance. DeepSeek’s model performs at a fraction of the cost, pressuring OpenAI’s ability to maintain proprietary advantages.
- ChatGPT remains the AI app leader. OpenAI’s consumer product continues to dominate, boasting 300 million weekly users and significant revenue from mobile apps.
When OpenAI’s financial projections leaked late last year, the numbers seemed a bit strange. The company had developed the best AI foundational models, had a head start in reasoning research and aimed to develop artificial general intelligence. Yet for every projected year through at least 2029, it told investors that ChatGPT — not its underlying technology — would bring in most of its revenue.
Then last week, it started to make sense. When DeepSeek’s open source reasoning model all but equaled OpenAI’s at 3% of the cost, it exposed OpenAI’s model-building business. Despite investing billions of dollars in research, the company couldn’t maintain its domination over open source. And even if it jumps out ahead again, its margins will likely compress due to commoditization. All this leaves OpenAI right where it told investors it would be: A product company above all else.
“The business of training foundation models and selling API access is tough,” wrote AI researcher Andrew Ng. “In contrast, building applications on top of foundation models presents many great business opportunities.”
Related Article: DeepSeek Shakes Up AI Game: $1 Trillion at Stake
Table of Contents
- AI’s Four Business Models—And Why OpenAI’s Path Is Clear
- Not Even Close: OpenAI Leads AI App Game
- The Race for AI App Supremacy Is Just Beginning
- Core Questions Around OpenAI's App Game
AI’s Four Business Models—And Why OpenAI’s Path Is Clear
Of the four businesses in the AI boom, AI model building is the worst of all, as one industry source put it to me.
- The first business is hardware, and NVIDIA is dominating there.
- The second is datacenter, which Amazon, Google and Microsoft are dividing up via cloud hosting services.
- The third is model building, where margins eventually disappear. And the fourth is applications.
For OpenAI, the lane is clear. Building products on top of the technology — whether its own or others’ — is its best shot at a justifying the billions it’s raised. The other three businesses simply aren’t accessible or are too small for a company chasing a $300 billion valuation.
If this is a moment of clarification for OpenAI, it isn’t a bad thing. No company turns artificial intelligence into products better than it does. Its ability to build its GPT models into ChatGPT, which it initially conceived as a demo, kicked off this entire generative AI moment.
And ChatGPT is now booming. After stagnating a bit last year, the bot now has more than 300 million weekly users. And OpenAI’s product innovations — like voice chat — have propelled it forward. That’s a solid position to be in, far better than its fellow proprietary model builders, like Anthropic, whose consumer products haven’t yet achieved the same traction.
Not Even Close: OpenAI Leads AI App Game
New data shows OpenAI leads the AI app game by some margin. Its ChatGPT app has earned $529 million to date, about four times the amount of its nearest competitor, according to a report from Appfigures. The total estimated mobile AI app market is currently $2 billion, according to the report, and that market should grow as more improvements — like OpenAI’s new Operator product — roll out.
Now, it’s possible that OpenAI builds artificial general intelligence, or AGI, and its models will be worth far more than anyone can anticipate. Sam Altman winked at the notion in his response to DeepSeek when he said, “look forward to bringing you all AGI and beyond.”
But that milestone is likely at least a few years away, and OpenAI would have to be the only one providing such technology to fully capitalize on it. Listen to leaders at Anthropic and Google DeepMind, and it seems clear the industry will reach that point almost at once.
Related Article: Notes on DeepSeek: Generative AI Is All About the Applications Now
The Race for AI App Supremacy Is Just Beginning
For OpenAI then, the worst part of the DeepSeek frenzy was seeing the DeepSeek app climb to the top of the charts. That’s a feat the Claudes and the Perplexities of the world have yet to achieve. It was a direct assault: An upstart competitor coming for its real business, a chat app. That said, once the initial wave of DeepSeek interest subsides, OpenAI will still have a better product.
ChatGPT, for instance, has voice capabilities that sets it apart. But it’ll now have to keep churning out the hits to stay ahead, something every app business knows all too well.
Core Questions Around OpenAI's App Game
Editor's note: Here are some key insights into OpenAI's evolving business strategy and the AI App landscape.
Why is OpenAI shifting its focus from model building to products?
With open-source AI models advancing rapidly and reducing margins, OpenAI is positioning itself as a product-driven company, banking on ChatGPT’s continued success for revenue growth.
How does DeepSeek’s emergence impact OpenAI?
DeepSeek’s ability to deliver high-quality AI models at a fraction of the cost threatens OpenAI’s proprietary edge, forcing the company to double down on consumer applications rather than model-building dominance.
What gives OpenAI an advantage over competitors in the AI app space?
ChatGPT’s strong user base, ongoing product innovations like voice chat, and significant market share in AI apps position OpenAI as the frontrunner despite rising competition.
Learn how you can join our contributor community.