2025 revealed how data center outages, from fires and mechanical failures to hyperscale cloud region events, can cascade quickly in an AI-driven world, highlighting the growing importance of physical resilience, control-plane reliability, and clean recovery.

As AI workloads scale, data center infrastructure faces mounting pressure from power demand, grid constraints and cooling limits.

Helios will go head-to-head with Nvidia’s own NVL systems, matching its latest NVL72’s 72 Rubin GPUs with 72 of AMD’s MI455X chips. It’s another sign that AMD is working to move further in on Nvidia’s turf in the AI data center market.

Comfort Systems USA, Inc. (NYSE:FIX)’s third quarter stood out in a way that is hard to miss. Same-store revenue rose 33%. Margins moved higher. Earnings per share more than doubled from a year ago. Free cash flow for the quarter came in above $500 million, a level that reshapes expectations for the business.

Forbes contributors publish independent expert analyses and insights. Anna Shedletsky writes about manufacturing technology, data, and AI. America’s AI race is accelerating at a blistering pace, and with it, the construction of the most expensive ...

FranklinWH has set a new global benchmark for home energy reliability by becoming the world's first residential battery system to earn TIA-942 Rated 1-4 certification, the same data-center standard trusted for always-on infrastructure.

DCaaS helps colleges and universities bridge the gap between on-premises infrastructure and cloud computing while reducing costs and operational burdens.

The advanced GPUs, cooling systems, and other hardware required to support AI demand a massive amount of electricity. Many AI data center campuses need over 1 GW of electricity, enough to power about 750,000 homes. According to an estimate by leading AI company Anthropic, the U.S. will need at least 50 GW of power for AI by 2028 alone.