Skeleton Technologies, a global leader in high-power energy storage, launches GrapheneGPU, a high-performance system designed to solve the key challenge of AI growth: energy inefficiency and power infrastructure limitations. Skeleton’s proprietary Curved Graphene material reduces AI energy consumption by up to 45% and lowers power connection requirements by 44%, while boosting computing performance in FLOPS by 40%.
By cutting energy use and peak power demand nearly in half, these improvements proportionally reduce both capital investments and operational costs in AI infrastructure. GrapheneGPU has successfully completed validation under the most demanding power profiles set by leading hyperscalers. Initial shipments are scheduled from Skeleton’s German facility for June 2025. US manufacturing expansion is scheduled for the first quarter of 2026 to support growing international demand.
AI data centers are already consuming up to twice the energy they require, as inefficiencies stemming from limited grid capacity, infrastructure constraints, and intermittent outages continue to mount. According to the International Energy Agency, global electricity consumption by AI data centers is projected to reach 945 TWh by 2030. The US will account for around 240 TWh and Europe for over 150 TWh, with AI data centers becoming dominant drivers of electricity demand growth in both regions. Energy waste and constrained power availability are poised to become the main barriers to AI growth and AI data centers development worldwide.
A key contributor to these issues is the fluctuating power demand of GPUs, which can cycle between 0% and 100% within seconds—experiencing sharp spikes during intensive processing, followed by idle periods. To manage these fluctuations and avoid grid instability or equipment damage, many data centers currently rely on artificial loads — commonly referred to as dummy loads—during idle phases to maintain a constant power draw. While this helps stabilize operations, the approach leads to significant energy losses: up to 45% of energy can be wasted as heat, offering no computational benefit. At the same time, peak power availability is constrained by limited grid capacity, and prolonged timelines for building new substations and transmission lines can span up to 7 years.
GrapheneGPU stores energy during idle periods and releases it during peak demand, effectively smoothing out the fluctuations. This eliminates the need for dummy loads, reduces cooling requirements, and enables GPUs to operate more efficiently.
GrapheneGPU PCS50
As AI computing and energy demands double annually, AI data centers are facing a critical bottleneck. GrapheneGPU delivers up to 40% more computing with the same energy footprint, while cutting both capital and operating costs by reducing grid upgrade needs, energy waste, and cooling, said Taavi Madiberk, CEO of Skeleton Technologies.
“Powered by our patented Curved Graphene, this is a fundamental shift in how AI infrastructure can scale — sustainably and economically“.
GrapheneGPU, powered by Skeleton Technologies’ patented Curved Graphene, combines peak-shaving supercapacitors with AI-driven control software to smooth out GPU power spikes and significantly enhance energy efficiency. Fully compatible with existing infrastructure, it enables sustainable AI scaling while reducing reliance on additional power generation or costly grid upgrades. Validated under real-world GPU load profiles, GrapheneGPU is certified for data center peak shaving and ready for immediate deployment.
By cutting energy waste and thermal stress, it helps data centers meet performance targets without compromising environmental goals—setting a new standard for efficient, climate-resilient AI infrastructure.
