Data Center Knowledge

Asia-Pacific's hectic data center rollout has hit a series of speed bumps. The most dramatic by far was the collapse Friday of Australian-based Firmus’ planned AU$7.1 billion (US$4.97 billion) IPO, which had been expected to be the biggest ASX debut since Telstra’s 1997 IPO.

The Texas A&M University System (TAMUS) has explored building a data center at its RELLIS Campus in Bryan, Texas, since 2013. A private developer began building one in 2024 but filed for Chapter 11 bankruptcy protection in November 2025, leaving the project unfinished. ThisWay Global (TWG), an Austin-based AI software company, says it acquired the project out of bankruptcy and completed the transaction on September 30, 2026.

Finland told Google to halt major site work at two planned data centers after regulators found that substantial land clearing and other site preparation had occurred before the required environmental impact assessments were completed. For data center operators, Google’s experience shows that securing power and grid capacity does not guarantee that construction can proceed.

How can over 50,000 labor hours be wasted a year on walking? Like this: A worker strolls across a 500,000 sq. ft. data center site to retrieve a single connector. No big deal ... until you multiply that trip across dozens of technicians, shifts, and buildings. That's how 50,000 hours (or about $1,000,000 in production) can be spent walking instead of focusing on mission-critical tasks. This is the kind of hidden issue experts like me are trained to find and fix. What is Total Cost of Ownership (TCO)?

HPE today (Oct. 7) announced the first batch of next-generation, AMD-powered servers designed to enable enterprises to run AI inference and agentic AI, as well as traditional enterprise workloads such as analytics and virtualization. The company unveiled its first four ProLiant Gen13 servers, all powered by AMD’s sixth-generation Epyc 9006 SP7-series CPUs – code-named “Venice” – featuring new automated security and expanded management capabilities.

Data center electricity demand could grow at a compound annual rate of 24% through 2030, but the availability of power infrastructure will determine how much computing capacity developers can bring online, according to McKinsey’s Global Energy Perspective 2026. The projection places data centers among the most consequential sources of new electricity demand while highlighting uncertainty over how that demand will evolve beyond the decade.

Power constraints are pushing data center development beyond Europe’s established hubs, but emerging markets now face the challenge of turning ambitious project announcements into operating facilities. Grid connections, construction capacity, and community support will determine how much of the next wave gets built.

AI data center developers and operators are being forced to confront not only how much power their facilities draw, but how quickly that demand can surge and collapse, creating frequency excursions that can ripple across the grid. Panelists at the recent Data Center World Power conference in Dallas (Sept. 21-23, 2026) warned that extreme events, such as a sudden drop of hundreds of megawatts in AI power draw, could trigger wide-area disturbances if not mitigated.

The uninterruptible power supply (UPS) has long been a critical element in the data center power train, providing both power conditioning and resiliency. That central role is increasingly under question as infrastructure evolves at both ends of the power chain. On the load side, AI-driven densification is reshaping facilities from the rack outward. On the grid side, greater integration of renewables, microgrids, and on-site generation is changing the profile of power delivered to data centers.

For the past three years, site selection for hyperscale and enterprise data centers in North America followed a simple, singular rule: find power. As artificial intelligence workloads escalated electricity demand, developers rushed to secure interconnection capacity wherever megawatts were available. But that era of single-variable site selection has officially come to a close.