Three years ago, OpenAI could do no wrong. ChatGPT had just landed, users were sprinting past 100 million in under three months and the company enjoyed the kind of cultural dominance most tech firms can only dream of.
Investors piled in, inflating the valuation to $500 billion. It looked like the model that would redefine software, search and everything in its blast radius.
Today, the picture is less triumphant. According to Deutsche Bank Research, CEO Sam Altman has privately declared a “code red” as the company faces what analysts describe as three simultaneous threats.
First, subscription growth is slowing, even though user counts continue to rise. Second, substitutes are finally emerging, with Google’s Gemini 3 now credible enough to enter the conversation. Third, OpenAI’s investment needs are vast, with $1.4 trillion flagged as planned outlay for models, data and compute.
It is the first warning sign that the pioneer of the generative AI boom is encountering friction at scale. Deutsche Bank’s dbDataInsights shows monthly spending on OpenAI subscriptions flattening in major European markets through 2024 and 2025.
The chart included in the note shows spending growth slipping into negative territory in several months, recovering only patchily. November 2025 data, projected from partial figures, continues the trend.
Slowing spending is one thing. Competitive pressure is another. For more than a year after launch, no model approached ChatGPT in capability or momentum.
That moat has narrowed. Gemini 3 is highlighted as a material substitute, part of a growing set of alternatives that includes open-weight models and increasingly capable in-house systems built by enterprise customers. The challenge is not that OpenAI is losing users. It is that the willingness to pay is becoming more elastic as options multiply.
The third threat is the most structural. That projected $1.4 trillion investment burden is breathtaking for any company, let alone one whose revenue model still depends heavily on subscriptions and usage-based pricing.
Deutsche Bank’s analysts note that the weight of expectations for the AI boom now rests disproportionately on OpenAI’s shoulders. Scaling frontier models requires capital, energy, chips and data pipelines at levels closer to national infrastructure than startup-style iteration.
The note does not predict collapse. Rather, it sketches a turning point. ChatGPT remains a global product with unrivalled reach, and its user numbers continue to rise. But the economics of generative AI are shifting.
OpenAI must convert enthusiasm into durable revenue, defend its lead as imitation accelerates and secure financing for an investment bill usually associated with governments, not growth companies.
Three years after launch, the era of easy dominance is over. The next phase will test whether OpenAI can still run faster than its own shadow.