It's not an AI bubble. It's an OpenAI bubble.

Every few months, the same question resurfaces: Are we in an AI bubble?

Every few months, the same question resurfaces: Are we in an AI bubble?

Barclays’ latest research suggests we might be, or at least that the growth curve is flattening. AI-related spending added roughly one percentage point to US GDP in the first half of 2025 but that boost is expected to fade.

Capital expenditure among the major hyperscalers is projected to slow from a 71% surge this year to around 30% through 2027. Adjust for inflation and the slowdown is even sharper.

So yes, the AI boom may be cooling, but that doesn’t mean AI itself is in a bubble. It means OpenAI is.

OpenAI has become the gravitational centre of the AI universe. Its valuation, its models and its partnerships move markets. Every time a new version of ChatGPT launches, it ripples through Microsoft, Nvidia and half the Nasdaq. The brand has become shorthand for the entire field and that’s precisely the problem.

Most companies building “AI solutions” aren’t really building AI. They’re renting it via an API, on someone else’s servers, running on someone else’s compute budget. Their growth is therefore dependent on another company’s roadmap, pricing and infrastructure decisions. The result is a narrow, circular form of value creation.

Meanwhile, the broader AI economy looks far less speculative.

Utilities are investing tens of billions to expand the power grid for data centres. Industrials are using machine learning to optimise logistics and reduce energy waste. CIOs are budgeting double-digit growth through 2030 for AI and cybersecurity.

These aren’t hype cycles, they’re infrastructure shifts happening quietly, in the operational trenches, not in model launches or valuation updates.

The danger isn’t that AI has been overvalued, it’s that our attention has been overconcentrated on one company, one narrative, one style of intelligence. Quieter innovators are building domain-specific models, self-learning architectures, capital-efficient alternatives to the compute-heavy status quo.

When the OpenAI bubble deflates, as all bubbles do, AI itself won’t collapse. It will mature.

The next phase won’t be about who can train the biggest model but who can make generative artificial intelligence useful, sustainable and scalable.

AI isn’t in a bubble but OpenAI, the brand, the symbol just might be. And when that imbalance corrects, the story of artificial intelligence will finally move from spectacle to substance.

Read more
All rights reserved  Boltzbit 2025