Without the active contribution from IT, data center infrastructure energy performance and sustainability will fall short of aspirations. Server power management features remain unexplored and underused by most efficiency initiatives.
Germany’s Energy Efficiency Act makes a PUE of 1.2 mandatory for all new data centers starting in 2026. This has reignited a debate: can a data center be both highly available and highly efficient?
Large colocation and public cloud companies have been growing strongly and very publicly, but enterprises continue to add data center capacity too, Uptime Intelligence data shows.
Data center and IT managers face growing demand to publish comprehensive carbon inventory reports. But estimates for the carbon content embedded in IT equipment have questionable accuracy and usefulness in informing decisions.
Pressure to improve data center efficiency and sustainability is driving interest in artificial intelligence (AI) technologies. Several startups aim to deliver new capabilities in IT power management and cooling optimization.
A new approach to data center management, proposed by data scientists and statisticians, looks to augment the functionality of tools like BMS and DCIM software by focusing on data, not equipment.
Despite high expectations, most operators will only see moderate impact from specialized AI hardware installations in the immediate future. The emergence of AI as a major force will sway the industry in a more profound, but less direct, fashion.
As operators deploy cold plate and immersion cooling, the cost or operational efficiency benefits are bound to disappoint. DLC alone will not bring the breakthrough in energy or sustainability performance the industry needs.
Thermal trends in server silicon will challenge assumptions that underpin efficiency and sustainability expectations around DLC. Limited visibility of future server cooling requirements means operators can only make an educated guess.
New survey data allows Uptime Institute to look at industry power usage effectiveness (PUE) in finer detail. Comparing PUE by compute capacity shows that, on average, larger sites have lower PUEs.
The thresholds set by Germany’s Energy Efficiency Act could make many legacy enterprise data centers obsolete — or else require operators to upgrade or move to compliant facilities
The final publication of the approved EED means that data center operators will need to undertake internal efforts to comply with energy management system and data reporting requirements.
The propensity to confidently give false information likely disqualifies generative AI from operational decision-making. However, this type of AI, with human supervision, could enhance other aspects of data center management.
Gigantic “gigawatt” data center campuses are being proposed or planned in many locations around the world. They could change the digital infrastructure landscape, but many questions — including how many will be built — remain.
Executives in many industries have been bold with their sustainability claims, setting ambitious net-zero goals and heralding minor successes. For digital infrastructure, as elsewhere, a painful and expensive correction is coming.
Daniel Bizo
Andy Lawrence
Douglas Donnellan
Jay Dietrich
Rose Weinschenk
John O'Brien
Max Smolaks
Jacqueline Davis