Google and private equity giant Blackstone late Monday announced an artificial intelligence (AI) cloud business aimed at whetting the tech industry’s insatiable thirst for AI computing power.
The U.S.-based joint venture will operate under a compute-as-a-service model, combining Blackstone’s vast infrastructure portfolio with Google Cloud’s proprietary AI chips, known as Tensor Processing Units (TPUs).
To kickstart the endeavor, Blackstone will make an initial $5 billion equity investment to bring 500 megawatts of data center capacity online by 2027. Financial details were not disclosed, but sources familiar with the matter indicate Blackstone will hold a majority stake. Bloomberg News reported that the total investment value could reach $25 billion when factoring in debt leverage.
Longtime Google executive Benjamin Sloss has been tapped to serve as the new company’s CEO.
“This venture helps address the soaring demand for our TPUs by offering organizations entirely new pathways to access high-performance computing capacity,” Google Cloud CEO Thomas Kurian said in a statement.
Ron Westfall, an analyst at HyperFRAME Research, expects the partnership to “disrupt the AI ecosystem by decoupling Google’s TPUs from its standard cloud suite and offering them independently.”
“Driven by Blackstone’s extensive data center network and financial backing, the standalone venture plans to deliver 500 megawatts of AI capacity by 2027,” Westfall said. “This initiative introduces a competitive alternative to NVIDIA-dominated GPU infrastructure, providing enterprises with greater supply chain resilience. From my viewpoint, by treating AI hardware as an independent economic asset class, the venture looks to undercut niche neocloud rivals and simplify access to top-tier training and inference. By structuring this as a majority-owned Blackstone vehicle that scales to a potential $25 billion total project value with leverage, the deal acts as Google’s off-balance-sheet answer to CoreWeave, offloading the heavy infrastructure-financing burden while expanding the commercial footprint of its custom silicon.”
The move highlights a broader shift as institutional finance increasingly bankrolls the physical architecture required to power AI.
Tech giants are projected to spend more than $700 billion on AI infrastructure and capital expenditure this year. However, private sector demand continues to outpace hyperscaler capacity. Asset management firm Ares recently estimated that the market opportunity for third-party data centers alone could hover near $900 billion.
“This isn’t just the biggest headline number we’ve seen, but it’s a high-quality bet on sustainable growth in AI infrastructure,” said Brittain Ladd, an AI and supply chain consultant at Chang Robotics.
For Blackstone, which manages the world’s largest pool of alternative assets, the deal cements its position as a dominant force in the AI land grab. CEO Stephen Schwarzman recently declared the firm to be “the largest investor in AI-related infrastructure in the world.” Blackstone already owns QTS, a major North American data center operator, alongside private equity stakes in premier AI firms like Anthropic and OpenAI. Secure energy pipelines have become prime real estate, forcing asset managers to aggressively snap up power generation and transmission assets to ensure long-term stability for energy-intensive data hubs.
The partnership also underscores an intensifying proxy war between NVIDIA Corp. and its largest cloud customers. By leasing out its custom TPUs via specialized joint ventures, Google is positioning itself as a direct alternative to NVIDIA’s dominant graphics processing units (GPUs).
Google is not alone in this strategy. E-commerce and cloud rival Amazon.com Inc. has experienced explosive growth with its proprietary Trainium chips. Amazon CEO Andy Jassy recently revealed that AWS’s custom chip business grew 40% quarter-over-quarter, achieving an annualized revenue run rate exceeding $20 billion.
As alternative chipmakers like Cerebras Inc. draw massive investor interest following recent public market debuts, the Google-Blackstone alliance signals that the race to control the physical bedrock of the AI boom is entering a highly capitalized, infrastructure-heavy new phase.
