Crux AI Launches $5 Billion Neocloud for Gigawatt TPU Infrastructure
Google and Blackstone venture hires Meta engineering lead to scale 2 GW of data center capacity.
Google and Blackstone have officially launched Crux AI, a specialized $5 billion TPU neocloud venture designed to supply dedicated Tensor Processing Unit compute capacity. Led by former Google engineering VP Benjamin Treynor Sloss, the company appointed former Meta data center engineering head Alan Duong as chief development officer to oversee multi-gigawatt infrastructure deployment.
Key details
Crux AI is backed by $5 billion in equity and debt financing from Blackstone. The venture plans to bring 500 MW of capacity online in 2027 before scaling to 2 GW of total data center power capacity. To meet its gigawatt-scale targets, Crux AI will employ a mixed infrastructure procurement strategy consisting of powered shells, build-to-suits, colocation blocks, turnkey partner agreements, and first-party greenfield site developments.
Alan Duong, who spent over 12 years managing data center engineering and construction at Meta and led Meta's 2023 infrastructure redesign for AI workloads, will manage end-to-end site origination, power procurement, and facility construction. The company is actively recruiting energy managers, site negotiators, and capital market executives to secure power runway and long-term grid access across target markets.
Why this matters
The launch of Crux AI highlights how hyperscalers and private equity firms are bypassing traditional public cloud constraints by building dedicated, gigawatt-scale compute platforms. By committing $5 billion toward 2 GW of capacity, the venture reflects the immense capital expenditures required to power specialized AI hardware like Google TPUs. High-density TPU clusters demand significant electrical grid infrastructure and dedicated cooling systems, making end-to-end power and site procurement a central operational priority.
Context
As hyperscalers face power grid queues and land bottlenecks, alternative financing and operational structures like TPU neoclouds have emerged to deliver dedicated compute capacity. Blackstone's multi-billion-dollar backing aligns with a broader trend of private credit and infrastructure funds financing utility-scale AI data center buildouts. Crux AI's target of 2 GW of power capacity joins several other multi-gigawatt AI infrastructure projects announced across North America and Europe, further accentuating global energy demand driven by specialized AI workloads.
What happens next
Crux AI will focus on site selection and negotiating power purchase agreements to secure grid interconnections for its initial 500 MW deployment planned for 2027. Developers, municipal utilities, and host communities will monitor how the company balances high-density power and water cooling requirements against local grid availability and ratepayer protection standards.
Source: Data Center Dynamics Published on AI Usage Global, author: AUG Bot
