Performant cooling requires a full-system approach to eliminate thermal bottlenecks. Extreme silicon TDPs and highly efficient cooling do not have to be mutually exclusive if the data center and chip vendors work together.
China wants to expand its data center capacity, and to achieve net-zero emissions. To this end, the nation is regulating for efficiency and renewable energy, as well as harnessing centralized control, to balance growth with sustainability.
Uptime’s 2025 cooling survey found that fewer respondents cited sustainability as a primary driver for direct liquid cooling (DLC). Gradual advancement of DLC plans may be refining operator understanding of its incentives.
The Azure outage highlights a blind spot in resiliency planning. It is not only cloud compute that can fail — shared global network services such as DNS and CDNs can disrupt access to systems anywhere, including on-premises.
A fire in South Korea’s government data center shows how a misjudged safety fix and a lithium-ion battery can spark a national outage, proving that battery chemistry, placement, and procedures are central to resilience.
Financial institutions are embracing public cloud for some mission-critical workloads, and using it as a launchpad for AI development.
Consensus is growing that a major market “correction” is coming: while some infrastructure operators are highly exposed, others may stand to benefit.
As IT organizations embrace AI, data center facilities and colocation providers need to plan to deploy the supporting infrastructure — despite many uncertainties. Most, however, are still moving cautiously.
Superconducting busbars could offer a solution for delivering power to high-density racks, eliminating resistive heating and removing the need to shift to medium voltage distribution equipment.
A major outage at AWS’s Virginia region took many global organizations offline. What can enterprises do to reduce or negate the impact of such widespread outages?
Research into neuromorphic computing could lead to the creation of smaller, faster and more energy-efficient AI accelerators. This would have a transformative impact on digital infrastructure.
AI is changing how data centers operate, what began with algorithmic fine-tuning of chilled-water plants is now moving into the IT side of operations, closer to the load. But will operators ever trust AI enough to let it run the room?
Currently, the most straightforward way to support DLC loads in many data centers is to use existing air-cooling infrastructure combined with air-cooled CDUs.
This report analyzes facility-level PUE data from the Global Data Center Survey 2025, highlighting differences by region, facility size and age. It assesses the vulnerability of the industry to PUE minimum performance standards.
Large-scale AI training is an application of supercomputing. Supercomputing experts at the Yotta 2025 conference agree that operators need to optimize AI training efficiency and develop metrics to account for utilized power.
Daniel Bizo
Peter Judge
Jacqueline Davis
Dr. Owen Rogers
Dr. Rand Talib
Andy Lawrence
Max Smolaks
Jay Dietrich
Douglas Donnellan