.svg){kind=spacer}
Details
- NVIDIA highlighted at Climate Week NYC that large language model inference energy efficiency has improved by 100,000 times over the last decade, promoting accelerated computing as a pillar of sustainable technology.
- The company participated in Climate Week panels with partners like Crusoe Energy Systems and Emerald AI, addressing how AI can optimize electric grids and support sustainable energy transitions through September 26, 2025.
- NVIDIA published new product carbon footprint reports, revealing a 24% reduction in embodied carbon emissions intensity comparing its HGX H100 and HGX B200 baseboards, and introduced an Omniverse Blueprint for building energy-efficient AI infrastructure.
- NVIDIA forecasts that AI applications could cut 4.5% from projected global energy demand by 2035, with sector-specific savings between 2% and 8% across industry, transportation, and buildings.
- The company featured its Earth-2 platform during Climate Week, showcasing its ability to deliver global-scale climate and weather simulations to help utilities integrate renewable energy and bolster grid reliability.
Impact
NVIDIA's announcements solidify its leadership in sustainable AI as the sector faces scrutiny over energy consumption and emissions. By combining dramatic efficiency gains, hands-on industry partnerships, and transparent carbon reporting, the company sets a high bar for rivals and appeals to enterprises balancing innovation with environmental mandates. As more tech giants face pressure to prove their climate credentials, NVIDIA's integrated efforts could accelerate sustainable AI adoption in critical infrastructure sectors.
