OpenAI's revenue and user growth slow as AI infrastructure costs climb

OpenAI is entering the hardest phase since ChatGPT's explosive success. The market has already shown it is interested in generative AI, but the company now has to prove it can turn that interest into durable revenue faster than its compute costs grow. Inside the startup, anxiety is rising because actual revenue growth and the inflow of new users are falling short of targets, and some users are leaving for competitors.

The problem stands out against OpenAI's capital-intensive model. The company is developing new models, widening its product lineup and securing huge amounts of computing power to train and serve its services. That takes more than research funding: it requires constant spending on data centers, chips, networking infrastructure and rented capacity from existing cloud providers. The more popular AI products become, the higher the load and the greater the dependence on outside financing.

ChatGPT gave OpenAI a powerful start and made the brand nearly synonymous with consumer AI. But after the first wave of hype, the market became far more crowded. Users are no longer tied to a single service: they can choose among large platforms, niche AI tools, assistants built into office products, and open models that can run locally or through cheaper cloud interfaces. In this environment, growth can no longer rest on novelty alone. The company has to retain its audience and prove practical value in everyday use.

Slower growth in new users is particularly painful for OpenAI because its costs scale poorly downward. If demand grows more slowly than expected, that does not automatically cut costs: infrastructure, cloud supplier contracts and development teams still have to be paid for. As a result, the gap between the ambitions built into growth plans and actual monetization becomes a strategic problem, not just a temporary swing in metrics.

The loss of some users to competitors adds to the picture. In generative AI, switching between services is easier than in many mature digital markets. If similar answer quality, a more convenient interface, a lower price or fewer restrictions are available in another product, a significant share of users is willing to experiment. For OpenAI, that means the fight is no longer only about model quality as such, but also about product speed, stability, transparent subscriptions, integrations with work tools and clear payback scenarios for businesses.

This is also an important signal for investors and partners. Until now, willingness to back OpenAI rested largely on the belief that research leadership would automatically convert into market dominance. As the industry matures, more down-to-earth questions come to the fore: how much it costs to acquire a user, how long that user stays active, how predictably revenue grows, and whether infrastructure can expand without an endless acceleration in spending. The AI market increasingly judges operational discipline, not just technological breakthroughs.

The enterprise segment is a separate challenge. Businesses care about more than impressive demos: they want security, cost control, legal predictability, quality integrations and service stability under load. If OpenAI wants to offset the slowdown in consumer use, it needs to strengthen the offerings that turn a model into a working tool inside companies rather than a one-off demonstration. That usually brings steadier revenue, but it requires a long sales cycle and serious support.

The situation around OpenAI shows that the generative AI market is entering a tougher phase. The winner will not be whoever surprised the world first, but whoever can retain users, sell clear value and control infrastructure costs at the same time. For OpenAI, this is a maturity test: the brand and technological leadership still give it a strong position, but it now has to back that position with the economics of the business, not just the volume of new model launches.