The AI race heats up as bubble fears continue to worry investors

The past few months have felt like watching a new industrial revolution develop at fast-forward speed.

Each week, another breakthrough model arrives. Another tech giant hints at a new data centre the size of a town.

Another start-up floats a valuation that would have seemed absurd a year ago.

Investors have been trying to ride the wave, yet many global assets sold off recently, as if markets suddenly realised something wasn’t quite matching up.

The noise was loud, and the numbers were bigger. But the picture was getting harder to read.

Some worry about who’s leading the AI race, while others are still voicing their concerns about the bursting of the “AI bubble”.

How the race became a sprint with no finish line

When ChatGPT appeared in late 2022, Silicon Valley behaved like someone had discovered oil under every office park. The bet was simple.

Scale the models. Scale the compute. Scale the revenue. In that first phase, it didn’t matter that costs were rising. What mattered was speed.

By late 2025, the field has changed shape, and OpenAI no longer looks untouchable.

The company recently declared a company-wide “code red” to shore up ChatGPT after Google’s Gemini 3 outperformed rivals on several benchmarks.

Sam Altman paused new ventures to focus on speed, reliability and personalisation.

That kind of urgency from the market leader tells investors something important.

The lead time at the frontier is shrinking, and the moat around “best model wins” is thinner than assumed.

But Google and OpenAI are not fighting alone.

China’s DeepSeek says its latest V3.2 model matches GPT-5 in reasoning tests and reaches Olympiad-level performance in mathematics.