OpenAI misses internal targets for ChatGPT audience growth and revenue

For the first time in a long while, OpenAI is facing not a question about model quality but a far less welcome problem for any fast-growing AI business: user and revenue growth has started to lag internal expectations. According to the Wall Street Journal, as cited by Bloomberg and Reuters, the company missed its own targets for new users and sales, and ChatGPT did not reach the mark of 1 billion weekly active users by the end of 2025. Inside OpenAI, this has prompted more than irritation over a missed KPI. It has raised a harder question: is commercial growth keeping pace with the scale of the infrastructure the company is already building and reserving years ahead?

The key detail is that this is not a drop in demand in absolute terms. By OpenAI's own figures as of March 31, 2026, ChatGPT already has more than 900 million weekly active users and over 50 million subscribers. In December, the company also said that more than 1 million business customers worldwide use its products for business directly.

The scale is still enormous, and it looks exceptional next to most technology companies. But when a company sets an internal goal of 1 billion weekly active users by a specific date, the gap between a "very large audience" and the "planned target audience" becomes financially significant, because hiring, sales, marketing and, above all, computing capacity are planned in advance around those expectations.

According to the WSJ, OpenAI also missed several monthly revenue targets in 2026. The publication links this, among other things, to Anthropic strengthening its position in programming and the enterprise segment, and to the growing popularity of Google Gemini making it harder to retain some subscribers. This is an important shift. Not long ago, OpenAI could grow almost automatically on the back of general interest in generative AI, but the market is now noticeably more competitive and segmented. A large user base no longer guarantees dominance in the most lucrative categories. Corporate customers compare not only model quality but also total cost of ownership, reliability, integrations, data control, agent features and how easily a tool fits into real workflows.

Against this backdrop, infrastructure costs look especially sensitive. According to Reuters, over the past year OpenAI has taken on commitments of roughly $600 billion tied to building data centers in the coming years. The Wall Street Journal also writes that chief financial officer Sarah Friar has voiced concern in conversations with management: if revenue does not grow fast enough, it may become harder for the company to pay for future compute contracts.

It is also reported that the board is scrutinizing recent data center deals more closely and asking questions about further expansion of computing resources. For a business that depends on ever more expensive clusters, chips, energy and long-term infrastructure agreements, even a moderate deviation from internal forecasts quickly turns into a strategic problem.

From the outside, though, the picture still looks very strong. In late March, OpenAI announced it had raised $122 billion in new capital, and in early April it said the enterprise segment already generates more than 40% of its revenue and could match the consumer business by the end of 2026. In other words, the company does not look weak in the classic sense.

It is more a clash of two speeds: on one side, almost unprecedented growth in product and customer base; on the other, an even more aggressive pace of infrastructure bets that demands flawless execution of the commercial plan. At this scale, even a small shortfall in revenue or audience stops being cosmetic and starts to affect the investment story, partner negotiations and preparation for a potential IPO.

The main takeaway is simple: leadership in AI is no longer measured only by user numbers, demo quality or the volume of new releases. OpenAI now has to prove that it can at once retain its audience, win enterprise budgets and turn colossal demand for AI into a cash flow that justifies the cost of its data centers. If it succeeds, the current miss will remain an episode against a backdrop of overall growth. If not, the market will view every new funding round, compute contract and model launch through the same question: is the business model keeping pace with the ambitions of the infrastructure?