How OpenAI plans to accelerate monetization
Advertising, e-commerce, social networks... OpenAI tests new revenue streams
Two announcements this week highlight OpenAI’s determination to step up efforts to monetize ChatGPT’s massive audience, going beyond paid subscriptions and the planned rollout of ads in chatbot responses. The goal: diversify revenue streams as the company continues to post staggering losses.
On Monday, the AI pioneer introduced in-chat purchases, allowing users to complete transactions without leaving its chatbot, which now boasts 700 million weekly users. OpenAI will take a commission on each purchase, though the rate remains undisclosed. A day later, it unveiled a new mobile app designed for creating and sharing short AI-generated videos — a TikTok-style platform that could eventually unlock a lucrative advertising channel.
Free users, high costs
For now, OpenAI relies on two main sources of revenue. First, its APIs, which let developers integrate its models into their own services. Second, subscriptions, which offer advanced features and higher usage limits. The company surpassed 20 million paying customers this spring.
But free users generate no direct revenue while driving up inference costs — the computing power required to generate responses. “The rich are paying so the poor can use it for free,” OpenAI CEO Sam Altman quipped. According to The Information, the company expects to burn through $8.5 billion in cash this year, against projected revenues of $13 billion, 3.5 times higher than last year.
To monetize its free user base, OpenAI is turning first to online shopping. Chatbots are beginning to change consumer habits, gradually replacing searches once made on Google or Amazon. Their advantage lies in handling more specific queries while offering personalized recommendations.
A different model than Google
Spotting the opportunity, OpenAI began adding product suggestions in April, with “buy” buttons linking out to retailers. In the US, users can now complete purchases directly inside ChatGPT. The feature is currently limited to Etsy sellers but is expected to expand soon to Shopify’s million merchants.
By charging commissions, OpenAI is adopting a strategy distinct from Google Shopping, which depends on sponsored listings and auction systems. On ChatGPT, product recommendations remain organic, drawing in part on expert guides. The company has pledged not to give priority to items available via direct purchase, even though those generate revenue.
Still, the model has yet to prove itself: similar experiments by social networks largely failed to take off. For OpenAI, this may simply be the first step toward “agentic commerce,” where ChatGPT could autonomously make purchases on users’ behalf.
Ads are coming
Meanwhile, OpenAI’s pivot toward advertising seems increasingly inevitable. In June, Altman admitted he was “not totally opposed” to ads, provided they were introduced “with great care.” While formats are still undefined, he promised that answers will never be altered to favor advertisers.
To prepare for this shift, OpenAI has appointed Fidji Simo, a French executive and former Meta insider, as co-CEO. It also hired Kevin Weil, once Instagram’s head of product during the platform’s ad rollout, and Shivakumar Venkataraman, who spent six years overseeing Google’s ad engineering. According to The Verge, OpenAI is now actively searching for a head of advertising.
OpenAI’s new app, Sora, named after its proprietary video-generation model, has a dual aim. In the short term, it seeks to capture audiences directly, instead of watching user-created clips spread across rival platforms, as happened with its image model DALL·E 4o. OpenAI isn’t alone: Meta recently rolled out a similar feature on Instagram.
But Sora offers a key differentiator: users can generate videos featuring their own faces or their friends’. In the long run, if the app gains traction, OpenAI could open it to brands and tap into the vast advertising market currently dominated by TikTok and Instagram.