CoreWeave has completed a US$9 billion all-stock acquisition of Core Scientific, a data center infrastructure provider.

The company announced the deal on July 7, 2025.

With this acquisition, CoreWeave will gain access to over one gigawatt of data center capacity.

This capacity will be used to support AI training and inference workloads.

Core Scientific, which previously focused on bitcoin mining, will see its GPUs repurposed for training and running generative AI models.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ Power capacity has emerged as the critical bottleneck in AI infrastructure race

The CoreWeave acquisition highlights how access to power has become the defining constraint in AI infrastructure expansion, superseding traditional metrics like square footage or server counts.

CoreWeave’s primary motivation appears to be securing 1.3 gigawatts of power capacity across Core Scientific’s data centers, with an additional 1 gigawatt available for future expansion 1.

This power-centric focus mirrors other major industry deals, such as OpenAI’s reported arrangement to rent 4.5 gigawatts of data center capacity from Oracle mentioned in the original article.

The importance of power is further emphasized by data center vacancy rates hitting an all-time low of 2.6% in North America despite record construction levels, indicating that physical space is no longer the primary constraint 2.

The deal’s emphasis on eliminating $10 billion in future lease obligations demonstrates how controlling power infrastructure directly, rather than leasing it, has become a strategic imperative for companies competing in the AI infrastructure space 3.

2️⃣ Data center consolidation accelerating with increasingly larger transactions

The $9 billion CoreWeave-Core Scientific deal represents a continuation of the consolidation trend in data center infrastructure, but at a significantly larger scale than previous transactions.

Historical industry acquisitions show a pattern of growing deal sizes: Digital Realty acquired Interxion for $8.4 billion in 2019, DuPont Fabros for $7.8 billion in 2017, and Telx for $1.89 billion in 2015 456.