YouTube video by Simon Clark

AI is consuming staggering amounts of energy—already over 10% of U.S. electricity—and the demand is only accelerating. Now, researchers have unveiled a radically more efficient approach that could sla...

Cut AI's power bill by 1,000x with new tech, but can it live up to the hype?

The Electric Reliability Council of Texas (ERCOT) is preparing to survey 461 data centers to evaluate their water and energy usage as part of an audit initiated by Governor Greg Abbott. This audit has temporarily paused power grid connections for specific large data center facilities until the assessment is complete. Data centers have until mid-October

How much of the energy consumed in an AI chip is spent doing something useful? This question affects everything from software to system architecture to chip design.

America is reshaping its entire electric grid to power AI — but no one actually knows how much power AI will need. In Ohio alone, data center requests dropped by more than half once developers had to put money on the line. Find out who is making billion-dollar decisions on numbers nobody trusts, and who gets stuck with the bill if they are wrong. https://qz.com/power-play-1

The rapid expansion of artificial intelligence is driving a significant surge in energy consumption and reshaping the technology landscape. A primary concern highlighted across multiple sources is the escalating demand for power required to fuel AI development. xAI's decision to purchase and relocate an overseas power plant to the United States to support its vast GPU infrastructure exemplifies this intense power need. Pennsylvania's Governor has reportedly expressed worries about the state’s electrical grid’s capacity to handle the escalating demands of AI infrastructure, prompting discussions of potential grid strain and even possible exits. This rapid growth is exposing a critical imbalance: power generation isn't keeping pace with the burgeoning needs of the AI sector. Nvidia remains the dominant force in AI hardware, deploying its most advanced chips, including the Grace Blackwell Ultra, to power significant AI infrastructure. While Nvidia leads, AMD and Intel are actively developing next-generation CPUs, with AIDA64 preparing support for AMD’s Zen 6 and Intel’s Wildcat Lake. Intel is currently facing challenges, including layoffs and potential closure of its Fab 28 facility in Israel, signaling a period of restructuring within the company. Asus is celebrating 30 years in the graphics card business, marking a significant milestone in the industry’s evolution. Beyond CPUs and GPUs, innovation continues across the chip development sphere. IBM is focusing its new Power11 chips on AI and ransomware defense. GlobalFoundries has acquired MIPS, a move likely geared towards incorporating RISC-V and AI intellectual property into its offerings. The emergence of tools like the Snapmaker U1’s four-color tool changer demonstrates advancements in 3D printing technology. The overall situation reveals a complex interplay of factors. The relentless pursuit of AI innovation is pushing the limits of existing infrastructure, particularly in power generation and grid capacity. This is impacting major players, leading to strategic shifts like acquisitions and layoffs, while simultaneously accelerating the development of new hardware and software solutions. Resolving the power supply bottleneck will be crucial to sustaining the continued advancement of artificial intelligence.

Lawmakers and witnesses debated S.867, a bill to add state-level permitting and oversight for data center projects. Conservation groups pressed for guardrails to prevent cost shifting and require water and power transparency; industry warned prescriptive rules could deter investment.

State Utility Commission to Address Data Center Impact on Power Grid The state Public Utility Commission is initiating new efforts to address data center effects on energy expenses and power consumption following an executive directive from Governor Josh Shapiro. The commission’s initiatives are designed to ensure that data centers do not shift electricity costs onto