Anthropic and OpenAI Pursue Smaller Data Center Deals Across UK and Nordics
Key Points
- Anthropic has explored 20-30 MW capacity agreements in the UK and Nordics, while OpenAI pursued similar smaller deployments in the Nordics.
- Anthropic signed a $45 billion deal with Nscale in August for 460 MW of compute capacity at a West Virginia data center development.
- Inference workloads are projected to consume 37% of global data center capacity by 2030, compared to just 13% for training, according to JLL.
Shift From Training to Inference Infrastructure
The move toward smaller deployments comes as artificial intelligence infrastructure demands evolve. Anthropic inked a roughly $45 billion cloud deal with Nscale in August 2024, which will see the AI lab rent approximately 460 megawatts of compute capacity at a data center development in West Virginia. OpenAI has committed to developing 3 gigawatts of capacity in Georgia and 8 gigawatts in Ohio, beyond its original Stargate AI infrastructure project commitment of 10 gigawatts.
Read Next
OpenAI and Anthropic Generate 10x More Revenue Than All Chinese AI Models Combined
1 days ago
Intel, SK Hynix Discuss U.S. Memory Chip Manufacturing Deal
2 days ago
Jabez Tan, head of research at Structure Research, explained the strategic rationale for smaller deployments. "Securing a few megawatts at an existing powered site can be more practical than waiting for a much larger block in one location," Tan told CNBC. "For workloads that can operate across separate sites, a collection of smaller deployments can add up to substantial capacity."
The distinction between training and inference workloads drives much of this strategy. Training artificial intelligence models requires large amounts of computing power to process enormous quantities of data, while deploying those systems day-to-day—known as inference—can be performed with smaller clusters of chips. "Training a large model typically requires many chips working closely together," Tan said. "Many inference workloads can instead serve separate requests across multiple smaller clusters, opening up more locations."
Related coverage: OpenAI and Anthropic Generate 10x More Revenue Than All Chinese AI Models Combined
Data on projected workload distribution supports this shift. According to a report by real estate company JLL, inference made up 9 percent of global data center workloads in 2025 compared to 14 percent for training. By 2030, inference is projected to consume 37 percent of global data center capacity, compared to just 13 percent for training.
The industry is responding to these infrastructure trends. In February, Nvidia announced collaboration with several data center stakeholders to study smaller-scale data centers designed for distributed inference operations.
Why this matters: As the ratio of inference to training workloads shifts dramatically over the next five years, the ability to deploy smaller distributed computing clusters becomes critical competitive advantage for AI labs. Companies that secure flexible access to 20-30 megawatt facilities now may avoid lengthy construction delays while maintaining capacity scalability, potentially translating to faster model deployment and reduced operational costs during the transition from model training to production serving phases.
Related Guide: Read our complete guide →
What This Means
The shift toward distributed smaller-scale deployments suggests AI labs increasingly view compute capacity as a commodity requiring geographic diversification rather than centralized mega-facilities. By 2027, when inference workloads are projected to exceed training demands, companies with established multi-region 20-30 MW commitments will likely enjoy faster deployment cycles and lower latency for end-user applications. Expect competing data center operators and regional cloud providers to aggressively target these mid-scale contracts through 2026.
Sources: CNBC and other international news outlets.
Disclaimer: This article was produced with AI assistance based on publicly available news sources. While we strive for accuracy, NewsOracle makes no warranty as to the completeness or accuracy of the information. Errors and omissions may occur. Readers should independently verify all information before acting on it. NewsOracle does not intend to defame any individual or organisation and accepts no liability for any loss or damage arising from reliance on this content. Content is for informational purposes only and does not constitute legal, financial, medical, or professional advice. All rights reserved. Unauthorised reproduction prohibited.
NewsOracle Editorial
The NewsOracle Tech Desk covers breaking technology news including AI, Apple, Google, Tesla, Meta, OpenAI and product launches.
Latest coverage: OpenAI
