Google will supply the new U.S.-based company with its tensor processing units<\/a> \u2014 chips purpose-built for processing artificial intelligence computations \u2014 bringing the first 500 megawatts of compute capacity online by 2027, with “plans to scale significantly over time,” Blackstone said in a statement.<\/p>\n “This new company has enormous potential as it helps to\u00a0meet the unprecedented demand for compute,” Jon Gray, President and COO of Blackstone, said in the statement. <\/p>\n The unnamed company will be helmed by Benjamin Treynor Sloss<\/a>, who most recently served as Google’s chief programs officer. A Google spokesperson declined to comment on whether Google would retain a direct leadership role in it.<\/p>\n The Wall Street Journal, which first reported<\/a> on the joint venture before Blackstone’s official statement, said the private equity giant would hold a majority stake, citing sources familiar with the matter. <\/p>\n Blackstone did not disclose the venture’s ownership structure in its statement, and did not respond to CNBC’s request for comment by publication time.<\/p>\n The Journal also reported that the joint venture has already identified likely data center locations, some of which are under construction.<\/p>\n Blackstone, which manages\u00a0more than $1.3 trillion in assets,\u00a0has invested aggressively across the AI ecosystem and, earlier this month, established a similar venture with Anthropic<\/a>.<\/p>\n Shares of Alphabet and Blackstone rose by about 1% in pre-market trading on Tuesday.<\/p>\n The deal underscores the growing rivalry between Google and Nvidia in AI hardware. Google’s TPUs have long been seen as the company’s answer<\/a> to industry leader Nvidia<\/a>‘s graphics processing units.<\/p>\n While Google continues to use Nvidia’s GPUs across its cloud architecture<\/a>, the company has sought to reduce its reliance on Jensen Huang’s Nvidia by developing its proprietary TPUs.<\/p>\n Other tech giants such as Amazon<\/a> Web Services have similarly sought to develop their own semiconductor chips<\/a>. Google was an early proponent<\/a> of in-house hardware production, manufacturing its first TPU in 2015.<\/p>\n Unlike GPUs, which are broadly suited for general usage<\/a>, Google touts its TPUs as purpose-built for more efficient processing of a narrower set of applications<\/a>, such as agentic AI applications. Google runs its Gemini AI model on its TPUs<\/a>, with Anthropic<\/a> and Citadel Securities<\/a> also among its clientele of TPU users.<\/p>\n Nvidia’s GPUs<\/a>, first developed in 1999 for rendering graphics on computers and gaming consoles, function by breaking complex computing problems into smaller tasks and resolving them simultaneously. <\/p>\n Demand for GPUs surged after the launch of OpenAI’s ChatGPT in 2022, propelling Nvidia to the mantle of world’s most valuable company<\/a> in 2024.<\/p>\n Earlier this month, Google briefly overtook Nvidia<\/a> by market value. Analysts told CNBC that\u00a0Google parent Alphabet remains well positioned in AI thanks to its in-house AI development, extensive distribution network, and profitable cloud division.<\/p>\n![\"Breaking](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/1779182891_340_108229615-1763686216714-251117-cg-32-ai-chips-gfx1.png\")
<\/p>\n