CERAWeek – the so-called Davos of energy – is where policy makers, manufacturers, technologists and investors gather to discuss what the world’s governments will do next.
NVIDIA and Emerald AI released at the conference last week a new way forward – treating AI firms not as static energy burdens but as assets in a dynamic, intelligent network. This collaboration combines high-speed computing, AI factory reference architecture and real-time electricity, enabling large-scale AI devices to connect to the grid faster, work more efficiently and boost system reliability.
Built upon NVIDIA Vera Rubin DSX AI Factory Reference Design and Emerald AI’s Conductor platform, the system integrates computing, power networking and control into a single architecture. The result is an AI firm that can produce high-quality AI tokens while dynamically responding to network conditions – changing as needed, supporting reliability and reducing the need to build multiple infrastructures for high demand.
AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power and Vistra are working to build the generation capacity needed to meet rapidly growing electricity demand. The companies plan to collaborate on innovative strategies to support AI firms built on NVIDIA and Emerald AI architectures, including hybrid projects that use combined power to accelerate power generation while delivering value to the wider grid. By combining AI-powered workloads with flexible operations, next-generation hardware and smart controls, this approach boosts grid reliability.
It is an important step in the assurance of a grid, supported by an ecosystem for advanced AI firms. This new paradigm of computer hardware – described by the founder and CEO of NVIDIA Jensen Huang as A five-stage AI cake – has power as its basis.
Driving updates in Tokens Per Second Per Watt
Energy disruptions are changing AI data centers, with energy efficiency or performance per wattspecifically tokens per second per watt, which is the defining measure of our modern computing power. By prioritizing computing efficiency, organizations can lower operating costs, increase revenue and create sustainable digital infrastructure for businesses and consumers across America and around the world.
“Energy is a concern, but it’s not the only issue,” Huang said said on Lex Fridman’s latest podcast. “That’s the reason why we strive for extreme codedesign, so that we can upgrade tokens per second with watts of size every year.”
NVIDIA has a long history of driving performance and energy efficiency. From the NVIDIA Kepler GPU in 2012 to the NVIDIA Vera Rubin platform this year, the number of tokens generated within the same power budget has increased by more than 1 million times.
It takes industry collaboration across the five-tier AI cake – from capabilities to switches, architectures, models and processes – to make this happen.
Robots, Digital Twins and AI Upskilling Drive Energy Advances
NVIDIA’s ecosystem partners demonstrated at the event how AI, simulation and workforce innovation are accelerating the power tools needed to support the intelligent age. Announcements from Maximo, TerraPower and Adaptive Construction Solutions exemplify how AI is pressing the times across construction, energy production and talent development.
Maximosolar robotics company AES, announced the completion of a 100-megawatt robotics plant at AES’ Bellefield facility. Using AI-powered robots developed with NVIDIA’s fastest computing, NVIDIA Omniverse libraries and NVIDIA’s Isaac Sim framework, Maximo has demonstrated that autonomous systems can now operate reliably in application environments. This method improves installation speed, safety and stability, helping to bridge the gap between the rising demand for electricity and building capacity.
TerraPowerworking with SoftServe, has envisioned the NVIDIA Omniverse digital twin platform designed to significantly shorten nuclear plant site development and design times. By applying AI and simulation to first-class engineering, the platform cuts the design cycle from years to months, accelerating the deployment of TerraPower’s Natrium power plants while improving design and grid connectivity.
Adaptive Construction Solutions has announced a national apprenticeship program, in partnership with NVIDIA, to help build the skilled workforce needed for AI firms and energy infrastructure. The program aims to increase job training, expand access to in-demand jobs while supporting the rapid construction of AI-driven power systems.
The initiative outlined how AI, digital twins and workforce innovation are coming together to deliver faster, more sustainable power systems.
Joining Together to Assess AI Factories for Grid Reliability
GE Vernova, Schneider Electric and Vertiv highlighted how digital twins, proven reference architectures and integrated architectures are becoming essential to develop AI firms as reliable grid participants. Publishers are tackling the “power-to-rack” challenge – designing AI devices as a unified power and computing system from day one.
GE Vernova explained how high fidelity digital twins are connected to NVIDIA Omniverse DSX Blueprint enable utilities and manufacturers to simulate the behavior of the grid, substations and AI factory loads together before deployment. Such system-level models help validate connection strategies, reduce risk and accelerate power uptime in critical grid areas.
Schneider Electric announced new NVIDIA Vera Rubin-certified architectures and digital twin architectures optimized for AVEVA. By simulating power, cooling and control in Omniverse, Schneider helps users optimize performance per watt, validate designs before building and operate AI factories more efficiently and predictably than standard.
Vertiv integrated, simulation-ready infrastructure built on renewable energy and cooling building blocks. Combined with the Vera Rubin DSX reference framework, Vertiv’s approach reduces design and deployment complexity while supporting the rapid and reliable scaling of AI firms.
https://www.youtube.com/watch?v=MyNer7tf-WQ
Together, these industry initiatives provide a digital path forward, including the proven architecture and physical infrastructure needed to transform AI firms into flexible, grid-aware assets to power the world efficiently.
Learn more about how NVIDIA and its partners are moving forward solutions powered by AI and high-performance computing.
#Work #Scale #NVIDIA #Power #Leaders #Accelerating #Factory #Shifts #Power #Grid