Datacenter Slowdown?

Next in Tech

A set of questions are being asked about whether the velocity of datacenter build outs is wavering. Analyst Dan Thompson returns to the podcast to explore the realities of the market with host Eric Hanselman. There are some facts that raised the questions, including Microsoft demurring on some datacenter projects, and they have fueled concerns that the heady pace might slacken. The DeepSeek model’s debut with claims of lower training requirements spooked the tech markets in January. There have been layoffs. Are these harbingers of AI doom?

In the same way that DeepSeek signaled the next phase of optimization in AI models, datacenter markets can be expected to adjust over time. The difference is that the time scales for the two areas are vastly different. With AI ecosystems seemingly cycling daily, the multiyear scheduling of datacenters might appear glacial by comparison. But it’s a market influenced by many factors, including a complex supply chain. Capital spending across hyperscale cloud providers remains healthy and maintains an eye on a future with much greater density of AI functionality. We’re still a long way from a correction.

More S&P Global Content:

• Powering AI – Opportunities, tensions in datacenter and energy markets

For S&P Global subscribers:

• Truths about how the power sector can (and cannot) respond to datacenter needs
• Potential impacts of DeepSeek on datacenters and energy demand
• 2025 US datacenters and nuclear energy report

Credits:

• Host/Author: Eric Hanselman
• Guests: Dan Thompson
• Producer/Editor: Adam Kovalsky
• Published With Assistance From: Sophie Carr, Feranmi Adeoshun, Kyra Smith