Google DeepMind chief says AI development could soon reach a choke point, here is why

Artificial Intelligence is developing rapidly, with companies such as Google and Anthropic periodically bringing more advanced models. However, Google DeepMind CEO Demis Hassabis has warned that one factor could put all this growth at a choke point.

It is 2026 and artificial intelligence (AI) models are better than ever. Google recently released Gemini 3.1 Pro, its most-advanced model yet, which even outperforms Anthropic’s Claude Opus 4.6 in certain benchmarks. However, Google DeepMind CEO Demis Hassabis believes that this rapid growth could come to a standstill due to one factor – memory.

In recent weeks, we have heard chatter surrounding a memory shortage. AI data centres need thousands of GPUs and computing power to run AI models. This has created a shortage in supply which is skyrocketing prices of various electronics, including smartphones.

Hassabis raised concerns that this shortage in supply of memory chips could become a major bottleneck for AI progress. He told CNBC, “You need a lot of chips to be able to experiment on new ideas at a big enough scale that you can actually see if they're going to work.” The Google DeepMind chief described this to be a potential “choke point.”

Previously Meta CEO Mark Zuckerberg has stated that AI researchers want "the most chips possible."

At a time when AI companies are pursuing ever-larger models and greater computational power, the constraints on memory chip supply are creating significant headwinds for the industry as a whole.

Google manufactures its own Tensor Processing Units (TPUs) and has the advantage of proprietary chip design, which helps reduce its reliance on third-party suppliers such as Nvidia. However, Hassabis emphasized that the company also faced an issue. He said, “It still, in the end, actually comes down to a few suppliers of a few key components.”