THE APEX TIMES
NVIDIA highlights a surge of AI-factory builds that tie power grids to AI compute
In a new NVIDIA post, utilities and industrial partners describe how they are using the NVIDIA DSX AI Factory Platform, reference architectures, and digital twins to plan and optimize power and cooling as one system from the substation to the rack.
NVIDIA says a growing share of AI-factory projects are being reshaped by a single constraint: power. In a blog post published Tuesday, the company described how customers are using its NVIDIA DSX AI Factory Platform to plan, simulate, and run the electrical grid and the AI factory together, aiming to reduce delays and stranded capacity that can occur when grid, facility power, and compute are engineered independently.
NVIDIA framed the challenge in stark terms. It cited more than $300 billion in equipment that it says is backlogged, and said over 200 gigawatts of projects are waiting in U.S. interconnection queues. In NVIDIA’s account, the bottleneck is not just supply, but coordination, because long-running engineering practices keep the grid, the facility, and the compute system in separate silos.
The DSX platform is described as a way to connect those layers. NVIDIA said DSX brings together grid and power expertise, data-center engineering tools, and simulation workflows so organizations can design, simulate, operate, and optimize data centers with an eye toward energy efficiency, grid reliability, and performance. The company’s pitch is that treating power and cooling as first-class system components can accelerate the time it takes to produce usable AI output.
A number of energy and industrial partners were mentioned as examples of how DSX is being integrated into broader projects. NVIDIA said GE Vernova is deploying grid-to-facility and rack power expertise to complement DSX systems, positioning the effort as a way to transform power constraints into “competitive advantages” while maximizing compute and token output, as well as sustainability and grid reliability.
NVIDIA also pointed to Hitachi Energy, saying its battery energy storage system has qualified as “NVIDIA DSX Ready,” tied to identifying power infrastructure that meets NVIDIA DSX reference architecture design requirements for resilient, reliable, and scalable AI factories. The post also named Schneider Electric, which NVIDIA said, together with AVEVA and ETAP, is advancing DSX with validated power and cooling reference architectures and an NVIDIA Omniverse-powered AI factory digital twin.
NVIDIA’s post further described how software and simulation layers are being connected. It said AVEVA engineering software expertise, NVIDIA SimReady assets, OpenUSD, and Omniverse can link simulations, real-time operational data, and AI-driven optimization across AVEVA, Schneider Electric, and third-party solutions. The goal, in NVIDIA’s framing, is to improve energy efficiency and workload performance while accelerating time to token across the data-center lifecycle.
Siemens Energy was presented as another participant expanding a DSX-aligned reference design. NVIDIA said Siemens Energy is developing reference designs for resilient, scalable, grid-aware power infrastructure spanning power generation, transmission, and distribution, aligned with NVIDIA DSX’s “Vera Rubin” AI factory design requirements. NVIDIA added that Siemens Energy is also exploring NVIDIA technologies including RAPIDS, Metropolis, and Isaac Sim through its Noedra digital twin platform to help monitor grid infrastructure risks and support unplanned outage prevention.
Beyond infrastructure vendors, the post also highlighted efforts from software and agent-focused companies. NVIDIA said Phaidra integrated DSX Flex, DSX Max-Q, and DSX Exchange into self-learning agents to optimize cooling and power allocation, targeting more compute within the same power envelope, and delivering about 10% more compute. NVIDIA also said Emerald AI integrated its Conductor platform into DSX Flex to help utilities tap underused grid capacity, modeling approaches designed to unlock up to 100 gigawatts on the existing U.S. grid.
NVIDIA did not provide detailed customer contracts, deployment timelines, or third-party verification in the blog post, and the figures it cited appear to be broad industry characterizations rather than company-specific backlog reconciliations. It also did not disclose whether these DSX integrations are universally deployed across projects or limited to early reference builds, nor did it break out performance results by geography, data-center design, or grid constraints.
Still, the direction NVIDIA is pointing is clear: the company is trying to make power planning and simulation part of the same workflow as AI-system delivery. With DSX-linked reference architectures, digital twins, and optimization approaches being promoted by multiple grid and data-center stakeholders, the next question for the market is how quickly these methods move from demonstrations to repeatable deployments that can reduce interconnection timelines and lower the risk of overbuilding capacity that cannot be fully used.
Why It Matters
- If power and cooling planning move closer to compute deployment, AI data-center operators may be able to reduce stranded capacity risk and shorten time to usable AI output.
- Digital twins and connected simulation workflows could make grid-aware operations more feasible, especially as projects face tight interconnection windows.
- Reference architectures promoted across energy and engineering software ecosystems may push standardization, which can lower integration friction for new sites.
- Optimization at the level of cooling and power allocation could become a differentiator as utilities seek ways to use existing grid capacity more efficiently.
Sources
Key Facts
- NVIDIA says power has become the biggest bottleneck in infrastructure, citing more than $300 billion in equipment backlogs and over 200 gigawatts waiting in U.S. interconnection queues.
- NVIDIA attributes a root cause of delays to siloed engineering, where grid, facility power, and compute have historically been designed independently.
- NVIDIA DSX is positioned as a platform to plan, simulate, and run the grid and AI factory as one system from the substation to the rack.
- NVIDIA said GE Vernova is using grid-to-facility and rack power expertise to complement DSX systems.
- NVIDIA said Hitachi Energy’s battery energy storage system is “NVIDIA DSX Ready,” aligned with DSX reference architecture design requirements for resilient, reliable, and scalable AI factories.
- NVIDIA said Schneider Electric, AVEVA, and ETAP are advancing DSX with validated Vera Rubin power and cooling reference architectures and an Omniverse-powered AI factory digital twin.
- NVIDIA cited reported outcomes including Phaidra’s self-learning agents targeting about 10% more compute within the same power envelope, and Emerald AI modeling up to 100 gigawatts of unlocked grid capacity.
Technology Related
Supermicro unveils NVIDIA Vera CPU rack aimed at scaling agentic AI workloads
Super Micro Computer said it is shipping an NVIDIA Vera CPU rack configured for dense, liquid-cooled rack-scale deployments, positioned as a complement to NVIDIA’s Vera Rubin platform.
Meta turns free cash flow negative and creates a new non-voting Class C share
Meta Platforms said its free cash flow turned negative and it is launching a new non-voting share class, a structure that can expand future financing options without immediately bringing in cash.
Zuckerberg points to Muse as a potential profit engine, but analyst flags heavy losses
Meta’s CEO says the company’s Muse AI assistant could earn more per user than Facebook or Instagram, even as an analyst argues Muse is already consuming more money than any other Meta launch.
Netflix partners with hoichoi to expand South Asian storytelling across Hindi, Tamil and Telugu
The streaming giant says the tie-up will bring hoichoi titles to Netflix in multiple Indian languages, alongside new premieres and additional adaptations tied to select Netflix originals.
NVIDIA shows how European retailers are shifting from AI “discovery” to physical operations with open models and digital twins
NVIDIA says European retail and logistics partners are using accelerated open and frontier AI models alongside NVIDIA digital twin technology to cut inference costs, improve catalog and search quality, and simulate warehouse workflows before deploying robots and systems.
NVIDIA spotlights open AI models powering European government projects, from agentic labor offices to sovereign classified-cloud plans
NVIDIA says European public-sector agencies and partners are using open models to keep sensitive data under government control, while speeding delivery of AI agents for tasks ranging from document search to medical imaging. The push includes platform upgrades for the European Commission’s shared AI infrastructure and a planned “sovereign” cloud stack designed for classified workloads in Germany.
NVIDIA promotes open Earth-2 weather models across Europe with new StormScope EU pipeline
The company says its Earth-2 open model family is being adapted for regional forecasting, climate research and environmental intelligence, with partners demonstrating near-real-time outputs every 10 minutes using satellite data.
NVIDIA highlights European finance’s shift toward open AI models built on company-owned data
At GTC Berlin, banks, payments firms and data providers showcased production deployments using open weight models and NVIDIA-accelerated computing, aiming to control cost, risk and data access while scaling AI beyond pilots.
NVIDIA expands Cosmos with new training tools, simulation checkpoints and data factory blueprints for world models
The company said its Cosmos platform now includes additional training and optimization capabilities, cloud-native data and labeling workflows, and updated microservices aimed at accelerating how teams build and deploy “world models” for robotics and other physical AI systems.
Microsoft shares weigh near historic levels as analysts debate the value of the latest AI hardware push
A Yahoo Finance analysis on Microsoft’s stock argues the market may be discounting future cash flow after the unveiling of its Maia 300 initiative, putting the stock’s valuation in a narrower band than recent trading suggests.
