Google and Blackstone take aim at Nvidia in $5bn AI bet
Alphabet’s Google and Blackstone are launching a $5bn (£3.7bn) AI infrastructure firm in one of the clearest signs yet that Big Tech is trying to loosen Nvidia’s grip on the AI boom.
The deal, as first reported by The Wall Street Journal, will see the pair create a new US-based AI cloud and data centre group powered by Google’s Tensor Processing Units (TPUs), the company’s chips designed to train and run the likes of ChatGPT.
Blackstone will be the majority owner of the venture, with the alternative asset manager committing $5bn in equity capital upfront.
The total investment could reportedly rise to as much as $25bn once debt financing is included.
The companies said the new group would bring 500 megawatts of data centre capacity online by 2027, with plans to scale significantly over time as demand for AI computing power continues to surge.
The move marks Google’s most aggressive attempt yet to commercialise its proprietary chips outside its own cloud division and compete more directly with Nvidia, whose GPUs became the backbone of the AI industry following the launch of ChatGPT.
“We see a generational opportunity to invest capital at scale building AI infrastructure,” said Jon Gray, chief operating officer of Blackstone.
“This new company has enormous potential as it helps to meet the unprecedented demand for compute.”
Benjamin Treynor Sloss, a Google veteran of more than two decades who most recently served as chief programmes officer, will lead the newly formed business.
The partnership underlines how the AI arms race is shifting from building ever-larger models to controlling the infrastructure that powers them – and, increasingly, the chips themselves.
The AI boom is becoming a fight over infrastructure
Google has spent the best part of a decade developing its TPUs internally, but pressure is building across the industry to find alternatives to Nvidia’s increasingly expensive and power-hungry systems.
Nvidia has retained its hold over semiconductor production, but rivals like Google, Amazon and Microsoft are now pushing their own custom-built chips as they attempt to cut costs and reduce reliance on external suppliers.
Thomas Kurian, chief executive of Google Cloud, said the venture would “help meet growing demand for TPUs” while giving customers “more options” for accessing computing power.
Google already uses the chips to train its Gemini models, while companies like Anthropic and Citadel Securities also deploy them.
Blackstone, meanwhile, has become one of the most aggressive investors in AI infrastructure globally.
The firm, which manages more than $1.3tn in assets, owns data centre operator QTS and has financed deals linked to Corweave, Anthropic and OpenAI.
The scramble to build alternatives to Nvidia is now spreading well beyond Silicon Valley, with Softbank recently injecting $457m into British AI chip company Graphcore as part of its wider push into AI hardware and infrastructure.
Meanwhile, UK startup Fractile is also attempting to redesign AI chips entirely by combining memory and compute together to improve speed and reduce energy costs.