CoreWeave Deploys 16,000 GPUs at Delayed Texas Data Center for OpenAI
AI computing provider CoreWeave has begun deploying large volumes of GPUs at Core Scientific’s Texas data center, which had previously been delayed.
According to a report by The Information on Tuesday, citing internal company communications, CoreWeave has now brought more than 16,000 GPUs online at the Denton, Texas facility, a site being built and converted by Core Scientific to support workloads for OpenAI.
The deployment marks a significant step forward for a project that CoreWeave acknowledged last quarter had fallen behind schedule due to construction issues tied to a third-party developer.
As previously reported, the data center builds had been delayed due to weather-caused disruption at Core Scientific’s site. CoreWeave noted in a November earnings call that OpenAI had agreed to shift contract start dates and that no revenue would ultimately be lost.
The Information reviewed CoreWeave’s internal Slack messages, which described the chip deployment as occurring “under intense spotlight,” with executives noting that the process had been affected by setbacks from partners and other factors outside the company’s control.
According to the report, the site had only “a few racks delivered” as of November. By the end of December, more than 16,000 GPUs were installed, including a single day in which over 2,000 GPUs were brought online.
Core Scientific previously disclosed plans to invest roughly $6.1 billion to convert the Denton site from a bitcoin mining facility into an AI-focused data center campus. The Denton buildout is part of CoreWeave’s broader expansion tied to OpenAI, which has signed multi-billion-dollar capacity agreements with the company. In September 2025, OpenAI expanded its relationship with CoreWeave, bringing the total value of its contracts to about $22.4 billion.
