Summer savings are heating up. From the Steam Summer Sale to GeForce NOW membership discounts, this week’s GFN Thursday delivers double the deals and more ways to get the most value from cloud gaming.

Plus, Dark Scrolls joins the growing Devolver lineup, alongside Square Enix’s The Adventures of Elliot: The Millennium Tales. They lead the charge for six new games joining the GeForce NOW library this week.

Steam Dreams Are Made of These

The Steam Summer Sale is here, bringing discounts across thousands of PC games as one of the year’s biggest opportunities to grow a gaming library.

Supported Steam games can be streamed across devices with GeForce NOW, making it easy to buy a game once, keep progress synced and pick up where the gameplay left off on PCs, Macs, handheld devices, phones, TVs and more.

In other words, the Steam Summer Sale brings the deals; GeForce NOW adds the flexibility. 

As new titles expand collections, storage demands and hardware requirements expand with them. GeForce NOW helps remove those barriers by streaming supported games from powerful GeForce RTX servers in the cloud, allowing members to enjoy today’s biggest games on devices they already own. Since downloads and installs are handled in the cloud, games can be added to the cart without being added to storage.

Check out the “Sales & Special Offers” row in the GeForce NOW app to discover the discounts today.

The Ultimate Upgrade to Level Up for Less

The deals don’t stop there. Pair GeForce NOW’s summer sale with the Steam Summer Sale to spend less time waiting on downloads, managing storage or needing pricey hardware upgrades — and more time gaming.

Get $70 off a 12-month Ultimate membership or $35 off a 12-month Performance membership and experience GeForce RTX-powered gaming in the cloud across devices.

The Ultimate membership unlocks GeForce RTX 4080- and 5080-class performance in the cloud with up to 4K resolution, up to 120 frames per second (fps) and advanced technologies like NVIDIA DLSS, ray tracing and NVIDIA Reflex.

Dig Into Devolver

Dark Scrolls, the kinetic action roguelite from Devolver Digital, arrives on GeForce NOW with its blend of fast combat, evolving builds and unapologetic chaos. Set in a warped fantasy world that doesn’t take itself too seriously, players battle through shifting arenas packed with enemies, hazards and power-ups that can turn a run from fragile to unstoppable in seconds.

Stack abilities, experiment with wild combinations and adapt on the fly as the game constantly raises the stakes — rewarding bold play as much as careful movement.

On GeForce NOW, Dark Scrolls is ready the moment players are, streaming across devices with no downloads or setup required. It joins a growing lineup of Devolver Digital titles on the service — Cult of the Lamb, Hotline Miami, Hotline Miami 2: Wrong Number, Inscryption, Enter the Gungeon and Ball x Pit — each delivering that distinct mix of style, surprise and controlled chaos, and just a click away with GeForce NOW.

A Storybook Across Centuries

The Adventures of Elliot: The Millennium Tales arrives on GeForce NOW, delivering a charming, narrative-driven adventure filled with mystery and discovery. With a hand-crafted world and a focus on exploration, it blends classic adventure gameplay with modern, character-driven storytelling.

Follow Elliot — a curious traveler bound to a mysterious Millennium Core — as he’s pulled across eras in a journey that spans neon skylines, forgotten ruins and quiet villages on the edge of legend. Each era has its own rules and rhythms, with Elliot’s reactions and scribbled journal notes giving the story a warm, personal touch.

Stream the cinematic-quality visuals and responsive gameplay with GeForce RTX power in the cloud for Ultimate members. Experience Elliot’s time-twisting journey in sharp detail across supported devices — no high-end rig, patches or paradox prep required. Just jump in and pick up from wherever the last chapter left off.

In addition, members can look for the following:

• Dark Scrolls (New release on Steam, available June 22)
• SAND: Raiders of Sophie (New release on Steam, available June 22)
• Deer & Boy (New release on Steam, available June 23)
• EMPULSE (New release on Steam, available June 24)
• The Adventures of Elliot: The Millennium Tales (Steam)
• FATAL FURY: City of the Wolves (Steam)

Leaving the last word to the Community Corner, one GeForce NOW member recently shared being “so impressed” by GeForce NOW. They gave the service a spin because of affordable pricing and took a lower-end PC from 20-30 fps on low settings to 60+ fps with settings maxed out — really putting the WoW in their World of Warcraft. 

What are you planning to play this weekend? Maybe even more importantly, what device are you planning to play on? Let us know on X or in the comments below.

Building AI systems at scale is demanding, requiring low-latency inference, fast vector search, strong GPU price-performance and infrastructure that can grow without multiplying operational complexity. 

NVIDIA’s latest work with Amazon Web Services (AWS) addresses each of those constraints. Across Amazon OpenSearch and Amazon EC2, NVIDIA AI infrastructure is giving enterprises more practical paths to deploy AI at production scale. 

EC2 G7 instances powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs expand the compute layer for AI, graphics, video and data analytics workloads, while the NVIDIA cuVS library accelerates the retrieval layer by making GPU-powered vector indexing the default in OpenSearch Serverless. And with AWS achieving NVIDIA Exemplar Cloud status for NVIDIA GB300, customers can trust they’re receiving peak optimized performance for their training workloads.

NVIDIA RTX PRO 4500 Blackwell Server Edition Multi-Workload GPUs Power New Amazon EC2 G7 Instances

Amazon EC2 G7 instances bring NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs to AWS for AI inference, graphics, spatial computing and GPU-accelerated data analytics — delivering a new instance type engineered for production workloads that need performance without the operational overhead of a customer-managed GPU platform.

Compared with G6 instances, G7 delivers up to 4.6x AI inference performance, up to 2.1x graphics performance and significantly faster GPU-accelerated data analytics on Amazon EMR using the NVIDIA cuDF library for Apache Spark workloads. 

With support for up to eight GPUs, 256GB of total GPU memory, 700 Gbps of EFA-enabled networking and up to 7.6TB of local NVMe SSD storage — across one-, two-, four- and eight- GPU configurations plus bare metal, coming soon — G7 instances let customers right-size infrastructure for their workloads instead of over-provisioning for them.

The platform’s versatility means AI teams get lower-latency inference. Media and entertainment teams get high-resolution video workflows and rendering. Simulation, computer-aided design, virtual desktop infrastructure, gaming and spatial computing teams get the same instance type for graphics-intensive applications. And data teams can apply the GPU memory, local storage and networking improvements to analytics pipelines and vector database workloads. 

G7 instances are accessible through AWS Deep Learning Amazon Machine Images (AMIs), Amazon Deep Learning Containers, Amazon EMR, Amazon EKS, Amazon ECS and graphics AMIs — and coming soon to Amazon SageMaker AI.

NVIDIA cuVS Makes GPU-Accelerated Vector Search the Default in Amazon OpenSearch

The next generation of Amazon OpenSearch Serverless powers agentic AI and dynamic workloads with no infrastructure management required. It uses GPU-accelerated vector indexing, powered by NVIDIA cuVS, as the default compute choice for all vector collections.

For teams building retrieval-augmented generation, semantic search, recommendation systems and agentic AI applications, that shift matters. It turns GPU-powered vector search from a specialized optimization project into a standard AWS capability.

The customer impact is direct: vector indexing up to 10x faster at a quarter of the cost, compared with CPU-only builds — making billion-scale vector databases practical to build in under an hour. 

By making NVIDIA cuVS the default in OpenSearch Serverless, AWS customers get a much faster path from raw data to production-ready AI retrieval infrastructure — with serverless scaling that reduces operational overhead when workloads are idle.

AWS Achieves NVIDIA Exemplar Cloud Status for GB300 Training Performance

AWS has achieved NVIDIA Exemplar Cloud status on NVIDIA GB300 for training workloads. This means AWS meets the rigorous performance thresholds that NVIDIA uses to benchmark AI workloads against its reference architecture. 

This achievement is the result of deep co-engineering efforts between AWS and NVIDIA teams. Through the NVIDIA Exemplar Clouds initiative, developers and AI leaders can be confident they’re using consistent, high-performance cloud infrastructure for large-scale training, helping teams evaluate cloud providers with greater confidence, improve total cost of ownership and move AI projects from planning to production more efficiently.

Together, these advancements reinforce every layer of the AI infrastructure stack on AWS. The throughline is the same: production-grade AI infrastructure that performs at scale, without adding operational burden to the teams running it.

Learn more in this AWS blog.

Editor’s note: This post is part of the Nemotron Labs blog series, which explores how the latest open models, datasets and training techniques help businesses build specialized AI systems and applications on NVIDIA platforms. Each post highlights practical ways to use an open stack to deliver real value in production — from transparent research copilots to scalable AI agents. 

Companies are asking how to build specialized AI that fits with the way their workflows actually run. 

The first wave of enterprise AI was about access. Companies experimented with new frontier and open models, ran pilots and explored how AI can help. 

Now, specialized agents — systems of models that can reason, use tools and take action even for the most complex workflows — put more useful AI within reach of the people who already know the work best.

Agents are already helping life sciences researchers accelerate medicine discovery, security teams investigate vulnerabilities with more context and operations teams seamlessly coordinate supply chains. 

To tap into these specialized agents, businesses are using a foundation they can adapt and own: one built on models they can customize, tools that connect to systems they already use and infrastructure that lets agents operate safely at scale.

NVIDIA Agent Toolkit — comprising models, tools, skills and a secure runtime — provides an open, modular foundation for building safer, faster, lower-cost digital AI coworkers that enterprises and developers can customize, specialize, control and trust.

The Building Blocks for Specialized AI Coworkers

Enterprises and developers building secure, specialized AI agents require:

• Models, which provide the reasoning foundation. 
• Tools and skills, which connect agents to the actions and domain expertise needed to get work done. 
• Runtime support, which helps agents execute workflows. 

NVIDIA Agent Toolkit includes all three:

• NVIDIA Nemotron open models give teams flexibility to customize, evaluate and deploy agents for their own needs. 
• NVIDIA NemoClaw blueprints provide patterns for safer agent behavior, delivering accurate results at lower costs, with tools and skills connecting agents to concrete actions.
• The NVIDIA OpenShell runtime helps agents operate safely inside the systems where work gets done. 

NVIDIA technologies accelerate all the pieces needed to turn a powerful frontier model into a fully functional digital coworker. The toolkit’s users can work with third-party agent harnesses — or agent orchestration frameworks — of their choice, including Hermes Agents and OpenClaw.

This unlocks enterprise AI momentum with control. And that matters because the most valuable agents across industries will be specialized. 

Agents Take Shape Across Industries

The specialized AI foundation is already at work.

In life sciences, agents can help researchers call domain models for protein design, virtual screening, genomics analysis and biomarker discovery. The new NVIDIA BioNeMo Toolkit enables work that previously took months to be completed in days. 

In healthcare, agents support clinical documentation, clinical decision support and care coordination. Plus, physical agents in robotics systems trained in digital twins of hospitals can scale surgical assistance and hospital automation to meet care demands.

In software, cybersecurity, industrial operations and customer workflows, agents can connect to the tools and data teams already use, helping people move faster through complex workflows.

For example, Cadence and Synopsys are building autonomous agents for chip design and engineering workflows. CrowdStrike is running specialized security agents that triage alerts with 98.5% accuracy. Palantir, SAP, ServiceNow, Siemens and Dassault Systèmes are embedding agent capabilities into the enterprise platforms where critical decisions get made. 

It all points to the same larger shift: Agents become more useful when they can combine models, tools, skills, runtime and infrastructure in ways companies can adapt to their own workflows. NVIDIA Agent Toolkit provides an open, modular foundation that enables this combination.

Learn more about NVIDIA Agent Toolkit and NVIDIA BioNeMo Agent Toolkit.

News Highlights:

• NVIDIA technology runs 81% of the TOP500 and 90% of the systems new to the list.
• 26 systems on the TOP500 adopted the NVIDIA Grace CPU, up eight from the previous list.
• The top eight systems on the Green500 run on NVIDIA GPUs and nine of the top 10 use NVIDIA technologies.
• No. 1 on the Green500, KAIROS, uses a single NVIDIA Grace Hopper Superchip.
• 376 of the TOP500 systems are interconnected using NVIDIA networking.

NVIDIA technologies power more than 400 of the world’s 500 fastest supercomputers — 81% of the TOP500 — according to the latest rankings released this week at the ISC High Performance conference in Hamburg, Germany.

That’s a gain of 17 systems from the previous list, with the momentum in new deployments: nearly nine of every 10 systems new to the ranking are built on NVIDIA technologies.

That percentage reflects a deliberate preference for machines built for AI, simulation and science together. And it’s compounding: NVIDIA systems across the TOP500 now deliver more than 2x the AI training and nearly 3x the AI inference throughput of every other platform combined.

GPU and networking adoption each hit new highs, with NVIDIA GPUs accelerating a record 238 systems and NVIDIA networking connecting a record 376 — the vast majority on NVIDIA Quantum InfiniBand, the backbone of large-scale AI and high-performance computing, and the rest on Ethernet. 

The trend behind the numbers is bigger than any one list: Accelerated computing is becoming the foundation for the systems taking on the world’s most demanding work, across AI and science.

Updated twice a year, the TOP500 ranks the world’s fastest supercomputers, while the Green500 list measures how much computing each delivers per watt.

A Full-Stack Footprint

NVIDIA’s reach now spans the full system — GPU, networking and, increasingly, the CPU — with NVIDIA Grace CPU adoption reaching 26 systems, up eight from the previous list, with nearly 2.5 million Grace CPUs shipped.

NVIDIA Grace-based machines sit atop both rankings: JUPITER at No. 5 and Alps at No. 10 on the TOP500, and KAIROS at No. 1 on the Green500.

Each pairs an NVIDIA GPU with the Grace CPU in a single NVIDIA Grace Hopper Superchip, letting the two share memory with minimal overhead — a design built for the memory-intensive demands of modern AI.

The NVIDIA Vera CPU, announced earlier this year, builds on the success of Grace, taking CPU performance and energy efficiency to new levels for the most demanding AI workloads in modern data centers — where agents move from answering basic questions to taking actions, running code, using tools and evaluating results.

Topping the Efficiency List

NVIDIA swept the Green500 ranking of the most energy-efficient supercomputers: The top eight all run on NVIDIA GPUs and nine of the top 10 use NVIDIA technologies. 

Leading the list is KAIROS, an NVIDIA Grace Hopper system at France’s University of Toulouse, at 73.3 gigaflops per watt — with Grace Hopper systems taking the top four spots, across France, Germany and the U.K.

From Exascale Science to the Next Wave

A record 35 NVIDIA AI HPC supercomputers are in development across Europe — equipping more than 3 million researchers with next-generation infrastructure for continental AI, accelerated science and industrial innovation.

Among these systems is JUPITER, Europe’s fastest supercomputer and its first to reach exascale, at the Jülich Supercomputing Centre in Germany.

JUPITER is mapping the human brain at cellular scale, simulating Earth’s climate and advancing the AI behind next-generation 6G networks.

The newest arrivals to the list run on the NVIDIA Blackwell architecture, with B200 and GB200 systems entering the rankings across Asia, Europe and the U.S. — and the first GB200 systems debuting in Japan.

The buildout is global, from a new AI factory in South Africa to national AI systems in Saudi Arabia, Singapore and Vietnam.

It’s the same story up and down the list: the world’s AI buildout is running on NVIDIA.

Telecom operators have seen remarkable returns from using generative AI to automate network management, customer care and back-office operations. Most of that impact has been task‑based: automation that speeds up predetermined steps while people manually correlate insights and direct next steps.

Automation is no longer the finish line — it’s the launchpad to autonomy. 

The industry is now pushing toward truly autonomous networks and operations, where AI agents proactively watch for problems and coordinate changes across network, IT and business systems.

Together, synthetic data, telecom-domain models, secure agent runtimes and simulations form critical pieces of a secure, telecom autonomy platform, where agents understand operator intent, act safely across business and network domains and keep humans in control of policy.

NVIDIA and its partners are demonstrating these building blocks at TM Forum’s DTW Ignite 2026 — running this week in Copenhagen — giving operators a practical path to running more autonomous, resilient networks and powering richer AI‑driven services for consumers and businesses.

Unlock Privacy‑Safe Telecom Data for AI Models

Reasoning models that understand the telecom domain are the foundation of autonomous networks. These specialized models require fine‑tuning on high‑quality datasets, yet 54% of operators cite data‑related issues as their biggest barrier, with the most valuable network and customer data too sensitive to use directly.

Synthetic data is enabling operators to safely increase the volume and diversity of training data, protect sensitive information and democratize access to production‑like telecom datasets across internal teams and external developers, without exposing raw customer records.

SoftBank Corp. is using technologies such as NVIDIA NeMo Safe Synthesizer and NVIDIA NeMo Anonymizer to generate privacy‑preserving synthetic datasets that reflect the structure and distribution of real network performance and configuration datasets. These datasets are being used to fine-tune its large telecom model and build specialized network agents.

Securely Deploy Autonomous Telecom Agents 

As telecom operators look to achieve autonomy across end-to-end workflows, they need AI agents that can stick with a complex job from start to finish, not just execute a pointed task. Long‑running autonomous agents that operate under strict service-level agreements, change‑management policies and regulatory constraints are key to this shift.

NVIDIA NemoClaw blueprints and the NVIDIA OpenShell secure runtime give these agents policy‑based guardrails and sandboxed access to telecom systems, so operators can more safely expand the role of agents in operations while keeping behavior predictable, auditable and governed.

AdaptKey is collaborating with operators to pilot security‑hardened, long-running agents for self‑healing 5G network operations. NemoClaw and OpenShell power agents that detect security and connectivity issues and submit scoped remediation requests into AdaptKey’s KeySmith platform for execution, which orchestrates diagnosis and runs agents that apply auditable fixes across core, radio access network (RAN) and billing systems.

Amdocs is showcasing the potential of NemoClaw and OpenShell for proactive customer-care agents, including roaming assistance scenarios where autonomous agents can identify customers whose roaming package is nearing depletion, engage them with approved options and execute actions within defined business policies and operational controls.

Amdocs is applying this runtime to autonomous data‑science agents that analyze customer accounts and assess migration eligibility, producing ranked, decision‑ready views that help operators intelligently sequence migrations to modern billing and business platforms at the right time and in the right order.

NTT DATA is using NVIDIA Nemotron open models with NemoClaw to build long‑running agents for proactive detection of network degradation. These anomaly agents track long‑term performance trends and escalate relevant cases to research agents for fine‑grained telemetry analysis and clear remediation proposals.

ServiceNow is bringing Project Arc to telecom, enabling autonomous network operations center agents that run incident response. Arc pulls context from emails, logs and diagnostics across disconnected systems and orchestrates the full lifecycle from initial alerts to assigned work orders. Secured by NVIDIA OpenShell and governed by ServiceNow AI Control Tower, every Arc action stays contained, auditable and within policy.

Tata Consultancy Services (TCS) is building a multi‑fidelity “AI sensor” architecture that helps operators spot and resolve network issues faster. NemoClaw orchestrates long-running agents powered by Nemotron and NVIDIA NV‑Tesseract that scan broadly for issues and selectively trigger deeper diagnosis, giving operators a faster, more efficient path from anomaly to action.

Bring Trust to Autonomy With Accelerated Simulation

As AI agents take on more responsibility in telecom operations, simulation is becoming an integral part of decision support. By accelerating simulation workloads on GPUs, operators can give agents a safe, near-real-time environment to validate their recommendations before acting on live network and business systems.

Forsk has integrated an AI‑based radio propagation model into its Naos RAN planning platform, achieving ray‑tracing‑level accuracy up to 200x faster than CPU‑only baselines on NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. The resulting RAN digital twin lets operators safely optimize the network in near real time, enabling use cases such as network self‑healing and automated antenna tilt.

VIAVI Solutions is accelerating its TeraVM AI RAN Scenario Generator by moving large‑scale RAN simulations from CPUs to NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. Early results show order‑of‑magnitude improvements in simulation throughput, letting operators run high‑fidelity scenarios at a real deployment scale so autonomous agents can de‑risk proposed network changes. 

In addition, VIAVI has released an IP Network Configuration Blueprint that extends validation into the IP and transport network domains, enabling operators to safely validate routing, traffic‑engineering and resilience changes, before they touch the live network.

KDDI and KDDI Research are bringing accelerated simulation into the 6G era through a collaboration with NVIDIA, Keysight and Samsung Research America to build a high‑fidelity RAN digital twin using NVIDIA Aerial Omniverse Digital Twin and Keysight’s digital‑twin‑ready emulation tools running on KDDI’s AI data centers. In this environment, multiple autonomous agents will be able to safely simulate and validate RAN “what‑if” scenarios, ranging from area‑optimization strategies to future radio conditions, traffic shifts and new AI air‑interface functions.

Dive deeper into the telecom autonomous networks stack by reading this NVIDIA technical blog.

JUPITER, Europe’s first exascale supercomputer at Germany’s Forschungszentrum Jülich, runs on NVIDIA Grace Hopper Superchips and NVIDIA Quantum-X800 InfiniBand networking — and it’s had a busy year.

As the international supercomputing community gathers at ISC in Hamburg this week, four projects running on JUPITER point to what exascale computing can actually do: map the human brain at cellular scale, simulate the entire Earth’s climate at 1-kilometer resolution, build AI systems for the next generation of wireless networks and simulate a universal 50-qubit quantum computer.

“With JUPITER, Europe doesn’t just join the exascale era — it leads it, across the widest range of science and AI of any system worldwide,” said Thomas Lippert, director of the Jülich Supercomputing Centre and professor at Goethe University Frankfurt. 

Four projects, detailed below, share a throughline: scientific problems that were out of reach on previous hardware are now tractable at exascale.

A Foundation Model for Mapping the Brain

The Jülich Brain Atlas project — anchored at Jülich’s Institute of Neuroscience and Medicine with partner Helmholtz AI, partner hospital and other Helmholtz institutions — has produced CytoNet, a foundation model for brain microarchitecture analysis.

The complexity of the human brain is astonishing. With 86 billion neurons and about 100 trillion connections between them, understanding brain function at single neuron resolution has been out of reach, until now.

The research is led by neuroscientist Katrin Amunts and computer scientist Christian Schiffer at INM-1, Jülich’s Institute of Neuroscience and Medicine. The model learns from brain imaging data at cellular scale, building a map that links individual cell structures to broader patterns of brain organization and function.

Training ran on JUPITER in under five days, using 6.5 petabytes of data from 21 post-mortem brains on 4,096 NVIDIA Grace Hopper Superchips. A paper describing the work is available on arXiv.

“For the first time, we’re not just using AI to analyze the brain — we’re building an agent that can think through the experiment itself,” said Katrin Amunts, director of INM-1 at Forschungszentrum Jülich and professor of brain research at Heinrich Heine University Düsseldorf. “That changes what neuroscience will be, and JUPITER is what makes that sentence possible to say today.”

That agent is the team’s next step: building an AI agent for brain researchers — integrating multimodal reasoning, language interfaces and Q&A capabilities using open models, including NVIDIA Nemotron 3 120B, working toward AI assistants that can help scientists interrogate brain data directly.

Climate at Kilometer Resolution

A novel ICON configuration — developed by researchers at the ETH Zurich, German Climate Computing Centre (DKRZ), Jülich Supercomputing Centre (JSC), Max Planck Institute for Meteorology, NVIDIA, Swiss National Supercomputing Centre (CSCS) and the University of Hamburg — won the Gordon Bell Prize for Climate Modelling at SC25 last November.

The breakthrough isn’t resolution alone. ICON is the first model to simulate a coupled Earth system at 1-kilometer resolution, with ocean, atmosphere and land, biogeochemistry and the full carbon cycle, with carbon exchanged, between all components. It can simulate and visualize complete ecosystems, such as phytoplankton blooms and zooplankton grazing. Previous systems could model pieces of this; ICON runs it all. This allows a much more precise and complete simulation of the Earth — observable at that level of detail for the first time.

Running on 20,480 NVIDIA Grace Hopper Superchips on JUPITER, the model simulated roughly 146 days of real climate into 24 hours of compute, setting a world record in global climate simulation. NVIDIA’s involvement in the ICON community spans more than a decade.

“Our simulations resolve the fine-scale winds, ocean eddies and upper-ocean mixing that shape marine ecosystems and regulate the ocean’s uptake of carbon,” said Daniel Klocke, computational infrastructure and model development group leader at the Max Planck Institute for Meteorology. “At a global resolution of just 1 kilometer, many of these interactions emerge directly from the laws of physics rather than being approximated. This gives us an unprecedented view of how the atmosphere, ocean and biosphere work together, helping us understand the processes driving our changing climate.”

6G Gets an Exascale Partner

In March, Ericsson and Forschungszentrum Jülich announced a collaboration to develop AI for the continued evolution of 5G and for 6G networks — with JUPITER as the compute engine for large-scale AI model training and testing.

The collaboration targets brain-inspired architectures designed to handle complex network operations at far lower energy costs. 

Research priorities include AI models for Ericsson’s radio and core networks, energy-efficient AI inference at the radio edge using neuromorphic approaches, and modular supercomputing architecture concepts drawn from JSC’s exascale work. 

Breaking Quantum Records

Researchers at the Jülich Supercomputing Centre (JSC), working with the jointly run NVIDIA Application Lab, also set a world first by fully simulating a universal 50-qubit quantum computer, surpassing the previous 48-qubit record. 

The simulation was made possible by drawing on the coherent, tightly coupled CPU-GPU memory architecture of JUPITER’s NVIDIA GH200 Grace Hopper Superchips, which lets data exceeding GPU limits spill seamlessly into CPU memory with minimal performance loss — allowing JUPITER to hold a far greater quantum state than GPU memory alone, which is what pushed the simulation past the previous 48-qubit record.

For now, that kind of simulation is the most powerful tool quantum research has: today’s quantum hardware can’t yet outperform classical computers on useful problems, so simulating quantum machines at the largest possible scale is how researchers design and stress-test the algorithms that future hardware will run.

This powerful quantum simulator, JUQCS-50, will be accessible to explore the performance of quantum algorithm designs within JUNIQ, the quantum computer user facility at JSC, led by Kristel Michielsen, director of JSC and professor at the University of Cologne. JUQCS-50 turns Europe’s first exascale system into a powerful testbed for tomorrow’s quantum-GPU supercomputers.

Exascale’s Impact

The range of science running on JUPITER — from neurons to atmosphere to wireless infrastructure to quantum — makes a case that exascale computing has moved from a research category into production. 

The results are a proof point for the Grace Hopper platform at the frontier of science.

Learn more about NVIDIA AI for science.

For the past two years, the U.S. National Science Foundation’s National Artificial Intelligence Research Resource (NAIRR) pilot program has driven innovative research across the U.S. for over 700 projects — spanning protein prediction and infectious disease outbreak management. 

NVIDIA contributed to the NAIRR pilot through a cloud-based resource that gives researchers dedicated access to a minimum of four NVIDIA DGX nodes for at least a month. NVIDIA also provided technical support to onboard and assist the researchers throughout their projects. 

With NVIDIA’s AI infrastructure support and DGX reference architecture providing dedicated resources, researchers have collapsed workflow timelines and uncovered groundbreaking technologies that will reshape and advance industries such as healthcare, agriculture and energy. 

The potential for scientific exploration and discovery across the nation through NAIRR is boundless. Learn more about a few NAIRR projects below. 

Physical Simulations With Polymathic AI’s Well Dataset

Simulation-to-real pipelines are becoming increasingly common across industries as a safer, more cost-efficient deployment method. 

Polymathic AI — a coalition of international scientists from Flatiron Institute, Cambridge University and Lawrence Berkeley National Lab — with the help of NVIDIA GPUs and NVIDIA NVLink interconnect technology, is strengthening physical, fluidlike simulations with its large-scale dataset called the “Well.” The dataset will be used to train the largest and most broadly applicable foundation model for fluidlike behavior to date. 

This foundation model, named Walrus, has been made publicly available along with its data, code and pertained weights. 

Polymathic AI’s approach builds on previous work in physics pretraining environments — addressing current limitations in scale and pretraining diversity. The research group also plans to explore scaling laws to help accelerate the development of more powerful foundation models for scientific applications.

University of Michigan’s Fusion Model for Energy Storage

Energy, a foundation of society, requires designing novel and efficient materials for energy storage and conversion.

Researchers at the University of Michigan, led by Professor Venkat Viswanathan in the Department of Aerospace engineering, are developing a model-fusion framework that brings together domain-specific molecular AI and general-purpose large language models. The goal is to help computational scientists more easily explore chemical space, ask chemistry-specific questions in natural language and identify promising materials for next-generation energy technologies. 

The family of molecular foundation models, MIST (the Molecular Insight SMILES Transformers), is designed for discovery and exploration across chemical space. 

MIST models were pretrained on large unlabeled molecular datasets and use a novel tokenizer, Smirk, to better capture nuclear, electronic, geometric, isotopic and stereochemical information from molecular representations. MIST models have been fine-tuned on more than 400 structure-property relationships and can match or exceed state-of-the-art performance across benchmarks spanning electrochemistry, quantum chemistry, physiology and other domains. 

MIST was developed on a 40-GPU NVIDIA DGX cluster the researchers gained as part of a NAIRR allocation and an additional 200,000 NVIDIA GPU hours on ALCF’s Polaris cluster. The team used NVIDIA’s NGC PyTorch container to support reproducible GPU-accelerated development across the different clusters.

Fusing MIST with general-purpose LLMs makes accurate quantum-chemical calculations more broadly accessible and accelerates the design of energy storage and conversion systems needed to enable widespread electrification of transportation, such as in the heavy-duty and aviation sectors.

Boston University’s BEACON AI Pipeline for Infectious Disease Detection 

Infectious diseases can spread rapidly in communities, causing surges in outbreaks. 

Boston University’s Hariri Institute for Computing and the Center on Emerging Infectious Diseases is working to train and evaluate a LLM using NVIDIA accelerated compute, through an AI pipeline to support an outbreak monitoring program called BEACON — Biothreats Emergence, Analysis and Communications Network.  

This LLM is being trained using a large corpus of documents on infectious diseases and epidemic-prone priority pathogens to support the work of field experts and outbreak analysts working on BEACON.

The model will be capable of analyzing online posts of emerging disease outbreaks on a global scale to extract features for downstream categorization and prioritization. BEACON will process signals from a variety of sources — including global disease-tracking platform HealthMap, news and social media feeds, subject-matter experts and individual communications via community boards or social media — to generate concise outbreak reports.  

These comprehensive outbreak analyses can inform clinical practice guidelines for emerging infectious diseases and identify gaps where further data is needed. 

Internationally deployed doctors, government organizations and academic researchers are already using the BEACON model to quickly identify and treat infectious diseases. 

“When you talk to infectious disease experts about what they used to do before we developed this pipeline, it used to take several hours for them to compose a report,” said Ioannis Paschalidis, director of Boston University’s Hariri Institute. “Now, producing a report gets done in roughly two minutes.” 

NAIRR and NVIDIA Across the Nation 

The latest scientific research doesn’t end there. Many other universities — including Harvard, Stanford, Colorado State University and more — are pioneering scientific breakthroughs with the help of NAIRR and NVIDIA. 

With scientists gaining broader access to AI and accelerated computing, innovation for a safer and healthier nation are more tangible than ever. 

Learn more about the NAIRR pilot program and explore how NVIDIA is driving academic research.

At the ISC conference running in Hamburg this week, NVIDIA is introducing new software that speeds AI for science, from chemistry and materials discovery to the search for dark matter. 

The NVIDIA DAQIRI library and new NVIDIA ALCHEMI NIM microservices — as well as the NVIDIA cuPhoton reference code, coming soon — turn work that once took hours or days on CPUs into real-time, GPU-accelerated pipelines. 

They’re a part of NVIDIA CUDA-X, a collection of tools and libraries that deliver dramatically higher performance across application domains, including AI and high-performance computing.

These performance gains are large and have real impact. Across disciplines, scientists are using AI and accelerated computing to generate data and insights with instruments and surveys faster than ever.  

For example, running on NVIDIA GB200 NVL72 systems, cuPhoton speeds loading, reading, processing and analysis of FITS data — the standard astronomical file format — from observatories and telescopes. In early access, cuPhoton accelerated loading and reading of FITS images collected by the Rubin Observatory’s Legacy Survey of Space and Time (LSST) by 14,900x. It also enabled up to 8,400x faster signal processing and analysis using 32 NVIDIA Grace Blackwell superchips. 

Ultimately, this means faster insights from the LSST camera — the largest digital camera ever built — which captures images of billions of far-away galaxies, as well as closer, faint objects that don’t reflect much light.

New Software, From the Lab Bench to the Telescope

The new software accelerates research on dark matter, materials simulation and more.

NVIDIA cuPhoton is a reference code for scientists looking to extract insights from multidimensional data collected from telescopes, X-rays and laser experiments. It’s built to load, process, analyze and visualize petabytes of data, and can be used alongside other NVIDIA CUDA-X technologies to build an end-to-end accelerated pipeline for work in fields including astrophysics and astronomy. 

Researchers at Princeton University collaborated with NVIDIA to develop cuPhoton and will use it — along with Harvard University — for processing and analysis of massive data collected from observatories and  dark energy surveys. 

NVIDIA DAQIRI — short for Data Acquisition for Integrated Real-time Instruments — is a high-performance networking library that streams data from fast detectors and sensors into NVIDIA software. Older systems are tied to fixed hardware and can drop data when instruments produce it faster than they can save it. DAQIRI keeps up by handling the stream as it arrives. 

A research project called A-GHOST was developed by scientists from CERN, the University of Chicago and University College London, in the framework of CERN openlab. It uses DAQIRI to run AI in real time on collision data recorded by the ATLAS Experiment at CERN. A-GHOST analyses data that  would normally be rejected by ATLAS  — over 99% of it, due to storage constraints — allowing it to catch potentially interesting signals that would otherwise be lost.

NVIDIA ALCHEMI comprises a collection of domain-specific microservices and a toolkit for accelerating chemical and materials discovery, with applications across battery materials, catalysts, OLED displays, beauty products and more. 

NVIDIA released in March two ALCHEMI NIM microservices for batched geometry relaxation (BGR) and batched molecular dynamics (BMD). These AI-accelerated tools let researchers simulate millions of molecules and materials at once: BGR to find their most stable structures, BMD to simulate how they move over time.

In addition, ALCHEMI is expected to soon include a microservice for the widely used Vienna Ab initio Simulation Package (VASP), enabling researchers to run materials simulations with higher GPU throughput. By running multiple VASP calculations on a single GPU with the NVIDIA Multi-Process Service, the microservice achieves a 3x speedup for geometry optimization — the process of finding the most stable arrangement of atoms in a material.

Plus, developers and researchers can use the ALCHEMI Toolkit to accelerate training of AI surrogate models called machine learning interatomic potentials and easily build custom, high-performance atomistic simulation workflows.

How Lila Sciences Runs the Scientific Method Nonstop With NVIDIA ALCHEMI 

Lila Sciences — which is building a scientific superintelligence platform and autonomous lab for life sciences, chemistry and materials science — collaborated with NVIDIA on a high-fidelity magnet simulation using ALCHEMI, demoed at NVIDIA GTC San Jose in March. 

Lila Sciences accelerated high-throughput materials screening by 50x using the ALCHEMI NIM microservice for BGR, identifying stable candidates that have higher chances of being synthesized. It then accelerated the calculation of magnetic properties by 30% for shortlisted candidates using the ALCHEMI VASP microservice in early access.

The speedups compound. ALCHEMI’s specialized kernels for TensorNet gave Lila a 6x speedup in training and inference and reduced memory usage by 3x, enabling simulations that previously took weeks in just days. 

Instead of running one experiment at a time, this approach evaluates multiple materials simultaneously in GPU memory and can be generalized for use cases spanning: 

• Materials discovery — screening novel, stable compositions at scale 
• Energy — discovering active, earth-abundant catalysts for producing chemicals and fuels
• Electromagnetics — understanding and predicting complex magnetic behaviors

ALCHEMI sits at the simulation layer, generating the physical-science data that feeds the rest of the loop.

In addition, Lila Sciences accelerates scientific discovery with the full NVIDIA stack, using NVIDIA Megatron-LM and NVIDIA Nemotron for training — including the Nemotron 3 Nano and Nemotron 3 Super open models, as well as the NeMo RL and NeMo Gym libraries. The company also taps into NVIDIA BioNeMo for molecular generation, NVIDIA Triton and NIM microservices for inference serving, and NVIDIA Omniverse libraries for digital twins. 

“The work showcases using a powerful computing stack assembled to accelerate discovery at a scale no individual scientist could achieve alone,” said Andy Beam, cofounder and chief technology officer of Lila Sciences.

Availability

The NVIDIA ALCHEMI Toolkit and Toolkit-Ops are available for download from Github and PyPI. ALCHEMI NIM microservices are available for download from the NVIDIA NGC catalog. The ALCHEMI NIM microservice for VASP is expected to be available later this summer. 

DAQIRI is now available on GitHub. CuPhoton is expected to be available this summer.

Learn more about NVIDIA AI for science.

See notice regarding software product information. 

Mission, Vision and Veritas — new Los Alamos National Laboratory (LANL) supercomputers to be built with HPE and NVIDIA — are tapping NVIDIA Vera CPUs to accelerate scientific discovery, unlocking agentic AI for science.

The supercomputers will use the HPE Cray Supercomputing GX5000 architecture with the NVIDIA Vera Rubin platform, combining NVIDIA Vera CPUs, NVIDIA Rubin GPUs and NVIDIA Quantum-X800 InfiniBand networking.

Under the planned configuration, Mission will include NVIDIA Vera Rubin GPU nodes and 2,300 standalone NVIDIA Vera CPUs using the HPE Cray Supercomputing GX240 blade. Veritas will feature approximately 1,150 standalone NVIDIA Vera CPUs to complement NVIDIA Vera Rubin nodes. 

Veritas will arrive alongside Mission and Vision and serve the Laboratory Directed Research and Development program, helping accelerate agentic AI for science. The system will test these technologies for use in larger systems being built out at LANL. 

Researchers are adding a new tool for science with AI agents that can form hypotheses, choose tools, launch simulations, analyze outputs and refine the next step. LANL’s public work on URSA, the Universal Research and Scientific Agent — running on Venado and soon Mission and Vision — points in this direction: a modular, feedback-driven AI framework designed to help scientists brainstorm hypotheses, plan experiments, run simulations and analyze results. 

LANL demonstrated that the Vera CPU delivered 7x higher performance on URSA workloads than the CPUs in the Crossroads x86 supercomputer.

Vera CPU for Agents and Simulation

In LANL’s early testing of NVIDIA Vera CPUs on Branson — an open source Monte Carlo heat transfer simulation tool — Vera outperforms the CPUs used in the Crossroads x86 supercomputer by over 3x. 

These results were made possible by Vera, including its custom Olympus core, LPDDR5 memory and fast on-chip fabric. 

A single Vera CPU outperforms a single socket x86-based CPU by more than 3x while providing more than 4x the memory per core and 6x the memory per node. Ultimately, this means faster  scientific results for LANL.

All of the lab’s supercomputers were codesigned by hardware architects, system software developers, domain scientists, computer scientists and applied mathematicians — helping ensure systems are shaped by real scientific workloads, not abstract benchmarks alone. 

Building on Generations of LANL Systems

Mission, expected to be operational in 2027, will be the fifth Advanced Technology System in the National Nuclear Security Administration’s Advanced Simulation and Computing program and will replace Crossroads for classified national security workloads. 

Vision, also expected to be operational in 2027, will serve as a resource for fundamental science, including materials and nuclear science, energy modeling, biomedical research and AI — letting more scientists test methods, train models and explore ideas before moving into higher-consequence work.

The work extends more than a decade of LANL and NVIDIA’s deep collaboration on CPUs, from Grace to Vera, using extreme codesign for LANL simulation workloads.

The three new supercomputers build on Venado, the HPE Cray EX supercomputer installed at Los Alamos in 2024 with NVIDIA GH200 Grace Hopper Superchips and NVIDIA Grace CPU Superchips. 

Learn more about the NVIDIA Vera CPU.

The next era of AI will not be defined by compute alone. Its growth will be determined by energy.

 As accelerated computing scales across AI factories, agentic AI, industrial AI, edge computing and physical AI — including robotics and autonomous systems — global electricity demand is rising at unprecedented speed. 

In many regions, expanding grid infrastructure to meet that need requires years of permitting, transmission upgrades, land acquisition and capital investment.

This challenge is reshaping how the world thinks about energy infrastructure for AI.

Eco Wave Power, a member of the NVIDIA Inception startup program’s Sustainable Futures initiative, is developing technology — powered by NVIDIA AI infrastructure and digital twins — that converts energy from ocean waves into clean electricity using existing marine infrastructure. By using already-built coastal structures, wave energy generation can be deployed closer to areas with growing power demand — including ports, industrial zones and future AI infrastructure hubs.

“Wave energy is one of the largest renewable energy sources that exists,” said Inna Braverman, cofounder and CEO of Eco Wave Power. “Everybody wants it, but nobody can do it, so I looked at the current problems with harnessing wave power and I asked: How do we simplify it?”

Turning the Sea Into a Power Source 

Harnessing Earth’s natural cycles for power generation isn’t a new concept. Wind and solar energy have been well established industries for decades. 

Waves are on the way to completing this trifecta of power-producing elements. 

In the U.S. alone, wave energy could produce over 60% of annual energy consumption, according to the Energy Information Administration. 

It all starts with floaters — noninvasive floating infrastructure attached to breakwaters or sea walls to capture the power generated by waves breaking against the shoreline. 

The density of seawater is roughly 800x the density of air, allowing larger amounts of energy to be generated using much smaller devices than wind turbines. 

The next step is managing and distributing that power. While previous companies faced a bottleneck at this stage — due to having their computer hardware in the floater, leading to potential damages during rough currents — Eco Wave Power puts its computers, sensors, hydraulic conversion and electric parts on land at centers, keeping expensive hardware dry and safe from storms. 

“Wave energy is the least intermittent source of renewable energy,” Braverman said. “Solar energy — for example — is great, but you have night, winter, cloud coverage and pollution that all impact production. With wave energy, you can generate around the clock.” 

AI Wave Energy Layer Using NVIDIA Omniverse Libraries and Accelerated Compute

As AI infrastructure expands, energy systems themselves are becoming increasingly intelligent.

Digital twins of wave patterns and floating infrastructure — built with NVIDIA Omniverse libraries — can simulate wave conditions, structural behavior, deployment configurations and operational scenarios before physical installation begins. These virtual environments can help optimize engineering decisions, reduce deployment risk and accelerate infrastructure planning.

At the operational layer, NVIDIA accelerated computing and AI technologies enable real-time optimization of wave energy systems through predictive analytics, anomaly detection, environmental forecasting and predictive maintenance. AI models can continuously analyze ocean conditions, equipment performance and energy generation patterns to improve efficiency and operational resilience.

AI can also orchestrate energy-aware computing infrastructure by aligning energy-intensive workloads with periods of stronger renewable generation and dynamically optimizing power utilization across distributed systems. 

Ocean Powered Data Centers on the Horizon 

Eco Wave Power operates projects in Jaffa Port, Israel, created in collaboration with EDF Power Solutions and the Israeli Energy Ministry, and in the Port of Los Angeles, developed in collaboration with AltaSea and Shell. Eco Wave Power is also developing new projects in Portugal at the Port of Leixões, Suao Port in Taiwan, and Mumbai, India, with Bharat Petroleum. 

Wave power has already demonstrated its ability to handle consumer energy needs — and is now showing potential to support data centers. 

“We have a possibility to link AI factories directly to wave energy, because a lot of data centers are moving toward the coast,” Braverman said. “They need cooling and water, so they’re now located in ports.” 

Pilots are already underway at the port of Los Angeles to showcase how wave energy can be the sole power source for a data center without tapping into the existing grid energy.

AI software serves as the control layer for this data center pilot, planning compute tasks based on the available power supply. For example, the software can monitor and predict when waves will be stronger throughout the week based on weather patterns — and accordingly allocate more intensive compute tasks for these periods. 

“We exist, we work, we’re grid connected and we have so much of this resource,” Braverman  said. “The energy is needed now, so I think we’re in the right place at the right time and we’re innovative, but we’re not futuristic, and that’s what sets us apart.” 

Explore how NVIDIA is driving the future of energy.

Hot tubs sit at about 38 to 40 degrees Celsius, warm enough that most people can only soak for about 15 minutes. NVIDIA’s newest AI servers can run their cooling liquid even hotter — up to 45 degrees Celsius, or 113 degrees Fahrenheit. That higher temperature limit is precisely what makes them more energy efficient.

The Rubin generation of NVIDIA AI infrastructure is the world’s first to achieve 100% liquid cooling — every chip, every networking component, cooled entirely by liquid in a closed loop with no fans anywhere in the system. This liquid cooling methodology is outlined in the NVIDIA DSX AI factory reference design, a guide that outlines best practices to design, build and operate the entire AI factory infrastructure stack.

Although each generation offers significantly more computing power for each watt, full liquid-cooled AI compute infrastructure enables data centers to dramatically reduce cooling energy consumption — making a meaningful difference to overall data center energy use at hyperscale.

“The NVIDIA DSX reference design for AI factories has zero water consumption — we have eliminated massive amounts of power usage and pretty much all water usage,” said Ali Heydari, director of data center cooling and infrastructure at NVIDIA. “With dry-cooler-based designs, it’s a closed-loop system with no evaporative water cooling — outside of maybe 1% of the year when we might need chillers in some climates.”

Historically, cooling alone has accounted for up to 40% of a data center’s electricity consumption, making it one of the most significant areas where efficiency improvements can drive down both operational expenses and energy demands.

Industry estimates suggest that raising chiller plant temperatures by just one degree can cut cooling energy costs by about 4%. At scale, those savings add up quickly. A 50-megawatt hyperscale facility can save over $4 million annually in cooling-related energy and water costs by moving to liquid-cooled infrastructure. 

In favorable climates, NVIDIA’s 45-degree liquid-cooling architecture can enable chiller-less operation with dry coolers, reducing facility cooling water consumption from roughly 2.6 million gallons per megawatt per year for conventional cooling-tower-based systems to near zero — up to a 100% reduction in water use. 

The reason: traditional air-cooled data centers depend on large volumes of cooled air to remove heat from IT equipment, often requiring energy-intensive cooling infrastructure during hot weather. With NVIDIA’s 45-degree liquid cooling, heat is captured directly at the chip and transported through liquid loops operating at much higher temperatures, allowing outdoor dry coolers to reject heat efficiently for much of the year while significantly reducing mechanical cooling requirements and facility water consumption. 

The data center ambient temperature is flexible — warm summer air is fine — because nothing in the server depends on cool air. The liquid does all the work — and the same liquid can be recirculated in a closed loop so no new water is consumed to cool the chips.

A New Standard for the Industry

Because the NVIDIA Rubin platform integrates 100% liquid-cooled infrastructure, every cloud provider and data center operator building for it is making the transition. 

The ecosystem is keeping pace. Motivair, the advanced cooling division of Schneider Electric, has worked alongside NVIDIA’s product roadmap for nearly a decade — and Richard Whitmore, its president and CEO, says the relationship only intensified as power densities crossed the threshold where air cooling was no longer a viable option.

“Once the watts per chip crossed a certain level, liquid cooling became mandatory,” said Whitmore.

Too Hot to Cool AI Infrastructure Is Hotter Than You’d Think

There’s a long-standing misconception in the industry that a cold data center is an efficient one. Decades ago, if a data center didn’t feel like a walk-in freezer, people would assume something was wrong. 

In reality, chips can sustain far warmer environments than that instinct suggests. Silicon processors generate enormous internal heat — the coolant entering a fully liquid-cooled chip at 45 degrees Celsius exits at roughly 55 degrees, having absorbed that heat load across the chip surface. Yet performance doesn’t degrade. 

The processors continue to operate at full performance because liquid-cooled cold plates keep device temperatures within validated operating limits, even with coolant entering the rack at 45 degrees Celsius. 

No Fans, No Cold Aisles — A Fundamentally Different Machine

Walk into a traditional data center and notice two things: the noise — cooling fans contribute to total noise levels at or above 85 decibels, loud enough to require ear protection — and the physical choreography of hot aisles and cold aisles, carefully managed to push cooled air across components. 

The Rubin architecture changes the picture.

Coolant — 75% water and 25% propylene glycol — flows through cold plates that sit directly on processors, pulling heat out at the source. Running that coolant at up to 45 degrees Celsius means that in many climates, the facility loop can reject heat without turning on mechanical chillers and noisy fans. 

That unlocks something beyond energy savings: the possibility of eliminating water consumption entirely. 

In the right geography — somewhere with reliably cool outdoor air — a liquid-cooled data center can reject its heat through coolant distribution units that capture heat directly at the source and transport it to outdoor dry coolers, essentially large radiator coils positioned outside the building. 

The loop is filled once and runs closed for the life of the facility. And it takes dramatically less space in the AI factory compared to traditional air-cooling infrastructure.

“In the right geographic location, with the right system design, you don’t need any refrigeration equipment,” Whitmore said. “You can just put big radiator coils outside and use the air temperature for all your cooling. It’s incredibly efficient.”

The geography caveat matters. A data center in the Scottish Highlands and one in Phoenix, Arizona, face very different realities. But even in warmer climates, the shift toward 45-degrees-Celsius coolant moves operators significantly closer to that chiller-less ideal — where chillers may turn on just a few days a year when the outside air temperature demands it.

Another key benefit of this new model for AI factories is the potential for waste heat recovery, where residual heat from AI factory operations can be repurposed to heat commercial or residential buildings nearby. 

The Engineering Problem Nobody Had Solved

Previous liquid-cooled servers were hybrid: GPUs and CPUs got cold plates, but the rest of the system stayed air-cooled, with finned heat sinks designed to shed heat into moving air. In a fully liquid-cooled server, the cooling for these components needed to be completely redesigned to use liquid.

NVIDIA’s thermal engineering team reworked how those components handle heat, designing cooling loops that simplify how liquid is routed to multiple high-power chips on the board using a single inlet and outlet, resulting in a cleaner tray-level cooling architecture.

One visible outcome: Rubin servers have clean, sealed front panels where air-cooled servers have perforated bezels. Another: fully liquid cooled servers enable higher rack density than air-cooled servers, so a system that previously occupied six rack units now fits in two — more compute, less space, less noise.

AI workloads are not getting lighter. The compute demand driving data center construction is growing faster than almost any other category of infrastructure investment. 

Without efficiency improvements in how that compute is cooled, the energy cost of running AI at scale would grow in lockstep with the hardware. Liquid cooling at up to 45 degrees Celsius — hotter than a hot tub, cooler for the planet — is one of the most important tools the industry has to close that gap.

Learn more about liquid cooling, the NVIDIA DSX platform for AI factories and NVIDIA’s approach to energy-efficient AI infrastructure.

In a consequential grid infrastructure decision, the Federal Energy Regulatory Commission (FERC) today issued a major milestone on large-load interconnection impacting how those building AI factories, semiconductor fabrication support systems and advanced manufacturing facilities can connect to the grid. 

In the era of AI, which NVIDIA founder and CEO Jensen Huang has described as a five-layer cake, energy is the critical foundation of technological innovation. 

FERC’s actions do more than modernize the grid interconnection queue — the approval process power developers must complete to safely connect new energy generation to the electrical grid. Following U.S. Secretary of Energy Chris Wright’s order directing FERC to address large-load interconnection, the actions establish national policy for how America can simultaneously lower energy costs, grow its industrial base, scale AI and strengthen the electrical grid.

For policymakers, utilities and technology partners, the message is clear: This is a pro-growth, pro-affordability and pro-reliability policy.

Faster Connections, Stronger Grid

At its core, the new framework cuts through burdensome bureaucratic red tape and aligns industry incentives.

Large customers are no longer passive entrants into an overburdened interconnection queue. They’re active participants in building the infrastructure they require. That means:

• Funding their own network upgrades, reducing cost pressure on existing ratepayers.
• Bringing new energy generation online, increasing supply alongside demand.
• Offering flexible load, allowing grid operators to manage peaks more efficiently.

Customers that can demonstrate flexibility — shifting or curtailing load in response to grid conditions — can move through the process on accelerated timelines, with study periods potentially as short as 60 days, per Secretary Wright’s directive.

This is not just faster interconnection. It’s smarter interconnection.

The Math Adds Up

Electric grids are capital-intensive systems with high fixed costs. When more demand is added efficiently, those costs are spread across a broader base — lowering prices per unit.

The data backs this up.

Lawrence Berkeley National Laboratory found that every 10% increase in state electricity consumption correlates with an approximately 6-cents-per-kilowatt-hour reduction in retail electricity prices. In other words, grid growth — when done right — lowers costs.

This dynamic is already playing out at the state level:

• North Dakota, after adding 23 data centers, saw the nation’s largest decrease in electricity prices.
• Mississippi, Louisiana and Virginia moved early to attract large loads and are now seeing tangible ratepayer, grid modernization and investment benefits.
• PG&E has forecast that, under the right conditions, each new 1 gigawatt of data center load could reduce electric rates by 1-2% by spreading fixed grid costs over more usage.

Inversely, states that fail to attract new load risk concentrating system costs on a shrinking customer base — putting upward pressure on rates for households and small businesses.

FERC’s actions create a national pathway to avoid that outcome. They build on the successes of communities across North Dakota, Mississippi, Louisiana and Virginia to create a national on-ramp, enabling every region to compete for and benefit from the next wave of industrial and technological investment.

Infrastructure That Powers the Modern Economy

This is not abstract infrastructure. It underpins the technologies shaping the next generation of American competitiveness.

The facilities enabled by this framework will power:

• AI-driven drug discovery that accelerates breakthroughs in medicine.
• Semiconductor design and advanced manufacturing that secure domestic supply chains.
• Weather modeling and climate analytics that improve resilience.
• Next-generation energy systems that are more adaptive and reliable.

The benefits extend beyond any single facility or industry. They can reach every American who visits a doctor, buys a product or pays an electricity bill.

The Moment to Engage in a Decade-Defining Opportunity

The framework is in place — but how it’s implemented, refined and scaled will depend on the stakeholders who engage now. Across government and industry, those who engage today will define what this system looks like for the next decade — how fast it grows, how resilient it becomes and how broadly its benefits are shared. 

NVIDIA is not waiting.

In parallel with FERC’s action, NVIDIA and Emerald AI are already working with partners across the ecosystem to build a new class of AI factories — designed from the ground up as flexible grid assets.

These facilities will:

• Bring their own generation to the grid
• Respond to grid conditions in real time
• Act as stabilizing forces for surrounding communities

Commercial deployment begins later this year.

This is what the future of large-load interconnection looks like: not a burden on the grid, but a backbone of reliability and efficiency.

FERC has taken an important step forward, and NVIDIA welcomes this leadership.

The digital era gave the advertising and marketing industry speed; the AI era is giving it autonomous operations. 

For companies building next-generation technologies for advertising and marketing, the question is no longer whether to adopt AI but whether their infrastructure can support it at the speed and scale the industry demands. 

At Cannes Lions, running June 22-26 in France, industry leaders including Alembic, Amazon Web Services (AWS), Criteo, Higgsfield, KERV.ai and Taboola are showcasing how NVIDIA technologies help unlock greater creativity and enable faster, autonomous operations at enterprise scale.

Decision Intelligence at Enterprise Scale

Causal AI platform Alembic helps solve one of enterprises’ biggest challenges: proving what marketing initiatives actually drive growth, not just reporting on what happened. Modeling true causation simultaneously across every channel, market and audience requires AI infrastructure that can process enormous, fast-changing datasets without reducing them to correlation-based assumptions.

NVIDIA DGX Vera Rubin NVL72 systems enable Alembic to scale its Causal AI models to analyze more variables, run larger simulations and quantify the true drivers of growth across marketing investments. Alembic will be the first Causal AI company to use NVIDIA DGX Vera Rubin SuperPODs for enterprise-scale causal modeling, giving executives a single source of unbiased truth on what drove business outcomes and where capital is being wasted, so they can act with confidence on future decisions.

Alembic’s inference runs on private supercomputing infrastructure inside Equinix data centers where the enterprise data already lives, keeping AI workloads local. World Wide Technology extends this to secure and regulated environments. Together, the companies offer a complete enterprise AI stack purpose-built for executives and data leaders accountable for capital decisions. 

Smarter Bidding at Auction Speed

For advertisers, serving ads and relevant recommendations across billions of daily transactions requires AI that’s accurate, fast and affordable enough to run at scale.

Amazon Web Services (AWS) is bringing cloud infrastructure, foundation models and NVIDIA GPU-accelerated computing together into a cohesive stack for the adtech industry that can scale for the era of AI agents. AWS is giving advertisers and demand-side platforms, supply-side platforms and independent software vendors a production-ready reference implementation to run AI-powered bidding directly inside auctions — powered by NVIDIA Triton Inference Server, which delivers deep learning inference fast enough to fit within real-time auction windows. 

That means adtech companies can move from rules-based decisioning to AI-powered models for bid price optimization, audience activation and deal scoring directly within the live auction pipeline.

Advertising company Criteo helps retailers show the right product to the right shopper at the right moment, across one of the largest recommendation networks in digital advertising. Keeping those recommendations relevant means continuously retraining its AI on billions of shopper timelines, a process where speed directly translates to quality. 

Collaborating with NVIDIA, Criteo achieved a roughly 2x speedup in model training on NVIDIA Blackwell GPUs, driven by the NVIDIA cuEmbed open library. That efficiency already frees roughly 17,000 GPU hours a year, and the companies are now scaling the work further.

Taboola is applying the same infrastructure logic to conversational AI, using NVIDIA GPUs to power DeeperDive, its AI answer engine, and extending that infrastructure to AI platforms and chatbots so they can generate revenue from advertising.

Agentic AI Across the Marketing Workflow

In marketing and other industries, AI agents are increasingly acting as digital coworkers, taking on long-running tasks across planning, execution and optimization. But these agents are only deployable for enterprises when they come with proper controls, including safety guardrails, auditability and role-based permissioning. 

The NVIDIA Agent Toolkit, which includes NVIDIA NemoClaw blueprints and the NVIDIA OpenShell secure runtime, provides these controls.

For example, Higgsfield AI, an AI video and image generator production platform, offers Higgsfield Supercomputer agents that manage the full marketing automation lifecycle: from campaign ideation, planning, creative production to posting and autonomous campaign optimization — in a single interface. It orchestrates leading large language models alongside 35+ image, audio and video models, including Higgsfield’s proprietary Soul and Soul 2.0 models built on NVIDIA Blackwell architecture. 

As part of the collaboration, NVIDIA Agent Toolkit software, including NVIDIA Nemotron open models, powers specialized subagents within the Higgsfield Supercomputer, running continuously inside every campaign. NemoClaw and OpenShell are being integrated to provide the enterprise trust layer.  

The result: the full marketing lifecycle, from ideation and creative production through posting, performance analysis and optimization, is available in a single interface. Marketing campaigns for nearly 400 of the Fortune 500 companies are created on the platform. 

Video courtesy of Higgsfield.

Contextual and Content Intelligence at Scale

AI understanding content at the level of meaning requires advanced infrastructure. NVIDIA’s multimodal stack provides the vector search, data processing and video understanding capabilities that make this kind of intelligence viable at production scale.

AI-powered media leader KERV’s Moment Match Engine evaluates a multitude of signals across every video frame and media asset to understand individual scenes, objects and products, providing content recommendations based on ad creative — the visual and textual elements of an advertisement — to drive improved engagement.

Video courtesy of KERV.ai.

KERV.ai recently optimized its processing pipeline, achieving over 10x improvements in speed and efficiency when using the NVIDIA Nemotron 3 Nano Omni open model in the platform. KERV’s solution analyzes what each ad or media brief contains, who it resonates with and which exact moment within content environments to target. 

On MediaPerf, an open benchmark for AI video understanding, Nemotron 3 Nano Omni — adopted by ecosystem partners including PYLER, which uses NVIDIA DGX B200 systems — delivered the highest throughput and lowest inference cost of any model evaluated, open or closed source.

Learn more about how NVIDIA powers advertising and marketing technologies.

Featured video courtesy of Higgsfield.

Play favorite titles from popular game libraries, keep progress synced and jump back into gaming sessions on virtually any device.

That’s the power of GeForce NOW cloud gaming.

From providing access to members’ favorite game libraries to offering some of the season’s best membership pricing, GeForce NOW is making it easier than ever to get more from the cloud this summer.

Plus, check out the seven new games arriving in the GeForce NOW library this week.

Hit Play Across Top Game Stores

Whether it’s blockbuster franchises, competitive shooters, sprawling role-playing games (RPGs) or indie gems, GeForce NOW supports thousands of games from leading PC game stores, making it easy to bring existing game libraries to the cloud.

Connect supported game store accounts and stream titles with GeForce RTX power. Games that include cloud-save functionality help keep progress intact across devices. Start a game on one screen, pick up where playtime left off on another and spend less time managing installs and storage space.

Jump into fan favorites like Cyberpunk 2077 and The Witcher 3: Wild Hunt streaming through GOG. These sci-fi and fantasy favorites support cloud saves, making it easy to continue where play left off. GOG single sign-on and game library syncing — coming this summer — will make that process even smoother.

Explore ancient mysteries in the Assassin’s Creed franchise or team up for tactical action in Tom Clancy’s Rainbow Six Siege through Ubisoft+, delivering iconic worlds and competitive experiences to the cloud. 

Whether joining massive, all-in combat in Battlefield 6 or shaping the fate of Thedas with compelling companions in Dragon Age: The Veilguard, the EA app offers action-packed experiences and beloved RPGs.

Find them in the GeForce NOW app by simply searching and filtering by store.

Choose the Screen, Keep the Adventure

Gaming doesn’t always happen at a desk — community members are sharing how GeForce NOW lets them play across their favorite devices.

GeForce NOW Ambassador and cloud gaming creator Anytime Anywhere Gaming recently showcased Dead as Disco — a PC title without a native mobile version — running on an iPhone through the cloud.

As he put it: “One of the coolest benefits of GeForce NOW is I can take the same copy of the game I own on Steam and carry it to my iPhone … it’s super awesome being able to play a game on a multitude of devices without having to buy it again.”

That flexibility extends beyond mobile. One GeForce NOW member on Reddit shared being “blown away” by the experience of playing on Mac through the cloud, highlighting how GeForce RTX-powered gaming can be enjoyed across devices, no upgrades required.

Whether via phone on the go, on a MacBook between work meetings or on a TV in the living room, GeForce NOW helps make PC gaming accessible across nearly every screen.

Upgrade and Unlock More This Summer

For a limited time, annual memberships are available at the best prices of the year:

• $35 off a 12-month Performance membership
• $70 off a 12-month Ultimate membership

Performance memberships deliver high-quality cloud gaming across devices, while Ultimate unlocks GeForce RTX 5080-class performance in the cloud with up to 5K resolution, up to 120 frames per second and advanced technologies like NVIDIA DLSS, ray tracing and NVIDIA Reflex.

Pair these special prices with thousands of supported games across top PC game stores. Plus, with GeForce NOW, spend less time downloading, upgrading hardware and managing storage — and more time gaming.

New on the Cloud 

In addition, members can look for the following:

• Embers of the Uncrowned Demo (New release on Steam, available June 13)
• Pro Cycling Manager 26 (New release on Steam, available June 15)
• Aphelion (Steam)
• Citizen Sleeper (Epic Game Store, free from June 18-25)
• Megastore Simulator (Steam)
• OPERATOR (Steam)
• Super Meat Boy 3D (Xbox, available on Game Pass)

What are you planning to play this weekend? Let us know on X or in the comments below.

A year ago at NVIDIA GTC Paris at VivaTech, France laid out plans to advance local AI — from new AI factories and national compute capacity to open frontier models and industrial platforms.

Now, that AI infrastructure is coming online. AI agents are running in production, startups are deploying applications and the French AI ecosystem is developing models, datasets and platforms designed around local languages, cultural context and European requirements.

French AI Infrastructure Takes Shape

France’s AI ambitions are gaining momentum. Billions in investment commitments through France 2030, the 2025 AI Action Summit and this year’s Choose France Summit are reinforcing the country’s position as one of Europe’s leading destinations for AI infrastructure.

As part of these efforts, Mistral is building a new 44-megawatt data center in Bruyères-le-Châtel, a commune in northern France. Announced at GTC Paris last year, Mistral’s first deployment is already operational with 18,000 NVIDIA GB200 systems — laying the foundation for the company’s roadmap of 200 megawatts of compute capacity across Europe by 2027.

The NVIDIA Blackwell platform is designed to help AI factories maximize throughput within fixed power budgets, combining higher performance‑per‑watt silicon with software features that boost data center throughput in power‑constrained environments.

Mistral is also working with French public investment bank Bpifrance, AI and advanced tech investment company MGX and NVIDIA to expand Campus AI, a network of AI factories anchored by a planned 1.4-gigawatt facility, making it one of Europe’s largest AI campuses. 

This momentum reflects a broader wave of AI infrastructure investment in France.

Scaleway, a European public cloud provider, now offers NVIDIA Blackwell B300‑SXM instances, giving developers and enterprises access to accelerated computing on demand. 

Bull and Foxconn have announced the production of NVIDIA Vera Rubin NVL72 in Europe. Systems will be manufactured and initially tested at Foxconn’s facilities in the Czech Republic before being assembled, integrated and fully validated at Bull’s factory in Angers, France. And a consortium of eight leading French companies has submitted a bid to host a European AI gigafactory in France to strengthen European AI infrastructure and accelerate AI adoption.

Meanwhile, Schneider Electric has teamed with NVIDIA to develop blueprints for gigawatt-scale AI factories, helping organizations accelerate AI infrastructure deployment.

Open Models Underpin AI Development

France’s AI ecosystem is producing models, datasets and platforms tailored to local languages, cultural context, and European business and regulatory requirements. As AI agents become more capable, organizations are increasingly adopting systems of models, using the right model for the right task to improve accuracy, reduce costs and accelerate outcomes.

On stage at this year’s VivaTech event, leaders from Gradium, H Company, LINAGORA, Pleias and NVIDIA explored the role of open models in enabling more transparent, customizable and locally relevant AI for governments, enterprises and developers.

“What we see now is a shift from building one isolated model to running continuous model infrastructure, where models train the next models, curate data, generate synthetic environments and verify reinforcement learning,” said Pierre-Carl Langlais, chief technology officer of Pleias. “Open model infrastructure is simply the way to ensure that many people can build AI and frontier-level practice can disseminate throughout the entire economy.”

The discussion underscored a key theme: combining open models with energy‑efficient infrastructure gives organizations the control they need to inspect, adapt, deploy and audit AI that meets Europe’s compliance and trust requirements.

NVIDIA Nemotron is advancing this with open models, datasets and playbooks that help model builders accelerate workflows from training to deployment.

• Mistral, a founding member of the NVIDIA Nemotron Coalition — a network of AI builders collaborating on open frontier models — is contributing model-development expertise and multimodal capabilities to help advance open frontier models through open collaboration.
• LINAGORA is building multilingual large language models with strong focus on the French language with its Luciole model family, developed using NVIDIA Nemotron and NeMo libraries and designed for local language and cultural context. Luciole 1B, 8B and 23B were pretrained on Jean-Zay — one of Europe’s most powerful and eco-efficient AI supercomputers — in collaboration with CNRS/IDRIS in the frame of the OpenLLM-France project. These open source models are distributed on Hugging Face (OpenLLM-France) along with their pretraining datasets. 
• H Company, also part of the Nemotron Coalition, is developing Holotron, a family of AI agents built on open NVIDIA Nemotron models. These computer-use agents can interact with any software interface in the same way a human would, without needing application programming interfaces or custom integrations, and automate complex enterprise workflows from end to end.  
• Pleias, in collaboration with NVIDIA, developed Nemotron-Personas-France and Nemotron-Personas-Belgium, privacy-preserving synthetic persona datasets grounded in French and Belgian demographics and cultural context. The startup is also using Jean Zay train compact language models entirely on open, well‑documented datasets, making it easier for customers to address EU AI Act requirements around data provenance and transparency. The team is now building specialized versions for search, retrieval-augmented generation and public sector document workflows.

AI Production Pays Off

The shift from pilot to production is the defining story of the past year, as organizations across every major industry in France use AI to boost efficiency, quality and speed.

Initiatives like the collaboration — announced at Adopt AI — between AI Factory France (AI2F), led by GENCI, NVIDIA Inception and NVIDIA Connect programs are helping startups gain access to national supercomputing resources, including Jean Zay. Early participants, including Pleias, Nebula and Ryax Technologies, are already turning that access into deployable applications.

Similar efforts are expanding across Europe, including a collaboration between NVIDIA and the Barcelona Supercomputing Center, creating a network that connects local infrastructure with startups and public sector institutions.

In healthcare, Sanofi is deploying AI agents across the value chain, from research, manufacturing and commercial to daily operations like procurement and IT, helping teams automate complex workflows at global scale. The company is also working with startups Owkin and Biolevate to develop autonomous agents for drug discovery and development.

Orange Business, the B2B subsidiary of telecom company Orange, adopted a lead-with-internal-use approach by first testing and scaling its Live Intelligence GenAI platform internally, with more than 100,000 active users across the company. At the same time, Orange Business made the platform available as a trusted agentic AI solution, enabling businesses and public sector organizations across Europe to adopt AI securely while keeping data hosted within the region.

Stellantis announced a strategic initiative to advance AI-enabled digital twins across its global manufacturing footprint, powered by real-time data, simulation and AI, to improve efficiency, quality and operational decision-making. Dassault Systèmes is combining virtual twins with AI infrastructure and open models on its agentic 3DEXPERIENCE platform, powered by science-validated industry world models that enable the design, simulation and operation of complex systems with confidence. This establishes a secure and trustworthy foundation for industrial AI, helping scale innovation across the generative economy.

TotalEnergies is building Pangea 5, a next-generation supercomputer developed with Dell Technologies and NVIDIA that will increase the company’s computing power to support seismic imaging, advanced simulation and AI-driven research in the energy sector.

L’Oréal is using its CreAltech platform to combine generative AI and 3D digital twins, helping creative teams scale content production while maintaining brand consistency, quality and responsible AI practices across global markets.

France’s trajectory has moved from announcing its AI ambitions to deploying the infrastructure, models and applications needed to realize them. As new AI factories come online and adoption accelerates across industries, the country is emerging as one of Europe’s most active environments for AI development.

The foundations are in place. What gets built on top of them is just getting started.

Join NVIDIA at VivaTech 2026 in Paris, running June 17-20.

NVIDIA XR AI is now available in public beta, giving developers a framework for building multimodal AI agents for AR glasses and XR devices.

AI runs at the speed of light. More and more, that light is made in Texas.

Coherent broke ground today on an expanded manufacturing building in Sherman, Texas. 

The company makes the lasers, optical components and compound semiconductors that wire AI systems together — and runs what it calls the world’s first 6-inch indium phosphide fab. 

NVIDIA founder and CEO Jensen Huang and Coherent CEO Jim Anderson were on hand for the ceremony, joined by Sherman Mayor Shawn Temann and Adriana Cruz, executive director of Texas Economic Development and Tourism, who delivered remarks.  

The expanded building will scale production of the same InP wafers that carry data between chips, servers and data centers at the speed of light — the optical backbone of modern AI infrastructure.

It’s the kind of milestone that turns a commitment into construction: a concrete step in expanding advanced semiconductor manufacturing in the United States.

“AI is the ultimate general-purpose technology,” Huang said during a conversation with Anderson at the groundbreaking. “Because intelligence is fundamental — the ability to process information, to reason and solve problems — it affects every single industry.”

Public programs like the CHIPS Act, funded at roughly $50 billion, were designed to bring chip manufacturing back to the U.S. 

As part of today’s event, Coherent is announcing a $50 million CHIPS Act grant to help finance the expanded Sherman facility — building on roughly $17 million in earlier support from the Texas CHIPS program and the Sherman Economic Development Corporation.

NVIDIA’s own commitment to produce up to $500 billion of AI infrastructure in the U.S. through industry partnerships with new sites in Arizona and Texas adds private-sector momentum.

“Coherent is a world-class company, and the work you do is vital to our future, vital to the future of artificial intelligence and vital to reindustrializing the United States,” Huang said.

Compound semiconductors like indium phosphide and gallium arsenide — the materials behind the high-speed networking and optical interconnects that modern AI runs on — don’t get the headlines that logic chips do. But their domestic supply chains have been thin for years. Today’s event was an argument that the gap is closing.

When 576 GPUs span eight racks and operate as a single system — as they will in NVIDIA Vera Rubin Ultra NVL576, which links eight NVLink racks of 72 NVIDIA Rubin Ultra GPUs into one 576-GPU domain — copper can’t carry the signal across that distance. 

To connect hundreds of thousands of processors separated by hundreds or thousands of feet across a data center, the only way to solve that problem is silicon photonics, Huang explained. 

As signaling rates climb, the reach of a metal trace shrinks, and spanning eight racks in copper would burn power on retimers and signal conditioning that a data center would rather spend on compute. 

Optics pays a one-time penalty to move from electrical to light, but once paid, distance is nearly free. At NVL576 scale, light is the most power-efficient option. 

NVIDIA and Coherent aren’t new to each other — they’ve worked together for roughly two decades. 

In March, they deepened the relationship into a multiyear strategic partnership: NVIDIA is investing $2 billion in Coherent to support R&D, future capacity and U.S.-based manufacturing, alongside a multibillion-dollar purchase commitment for advanced laser and optical networking products.

Sherman, a city of roughly 45,000 people an hour north of Dallas, has become the latest dateline for the AI era — emblematic of a boom built as much on picks, shovels and manufacturing muscle as on software.

“When we get to full capacity, this site will support more than 550 direct jobs — and thousands of jobs, direct and indirect,” Anderson said.

What the factory ships isn’t a single product dropped into a single slot. It’s the lasers, transceivers and pluggable optical modules that move data across NVIDIA networking — each enabling a different part of the system.

“As AI systems grow larger and more powerful, connectivity is just as important as compute,” Anderson said. “AI runs on compute, but it scales on connectivity — and Sherman is where that connective tissue gets built.”

Today’s event made that visible.

Before the groundbreaking, guests toured the existing fab and previewed the equipment that will populate the expanded building once it’s running. An NVIDIA rack stood on the factory floor, one of the six stops on the tour. 

The tour was followed by a fireside chat with Huang and Anderson, where the two CEOs discussed the partnership and what scaling domestic optical manufacturing means for the AI buildout ahead.

“Today marks an important milestone — not just for Coherent, but for American manufacturing and for the future of AI infrastructure,” Anderson said. 

The semiconductor laser was born in U.S. labs — Bell Labs demonstrated a room-temperature version in 1970 — before the technology and its manufacturing largely migrated overseas.

“We were founded as a manufacturing company in 1971. We’ve always been a U.S. manufacturing company — and after 50 years, the most advanced 6-inch indium phosphide line in the world is right here in Sherman,” Anderson said. 

That manufacturing gap shows up in the wafers themselves: while silicon fabs run on 12-inch wafers, most of the world’s InP production is still stuck on 3- and 4-inch wafers — lower yields and far fewer components per run. 

Moving to 6-inch wafers roughly quadruples the usable area of a 3-inch wafer (area scales with the square of the diameter), driving down cost and unlocking the volume the AI buildout demands.

It took 50 years to build the first line, Huang said — and in one year, they’ve quadrupled it, a measure of the demand for accelerated computing.

Inside, the core processes are familiar: lithography, photoresist, depositing and etching materials, layer by layer. The difference is the material. On an InP substrate, engineers grow exotic compound-semiconductor layers and tune them for precise optical properties — the physics that lets a chip emit and modulate light.

Today, that InP travels inside Coherent’s pluggable optics — transceivers about the size of a USB stick that plug into the front of NVIDIA networking switches and move data between racks across the data center floor, where copper can’t reach. Each module carries an indium phosphide laser. 

Those same modules now help enable NVIDIA Spectrum-X Photonics and Quantum-X Photonics switches with co-packaged optics: Coherent supplies the external laser module that plugs into the switch’s front plate. 

And as NVIDIA works to keep optics from becoming the next bottleneck, demand for those lasers only climbs.

“Ten years from now, I think we’ll look back and realize AI is what made it possible to invest in sustainable energy, upgrade our energy grid and reconstitute a workforce,” Huang said. “You can’t have only information workers in an economy — you also have to have builders. We have an opportunity over the next 10 years to reshape our communities and be much more balanced.”

Enterprises are moving agentic AI from proof of concept to production — and the next generation of AI factories are built for the era of agents.

At HPE Discover Las Vegas, running through Thursday, June 18, NVIDIA and HPE are expanding the HPE AI Factory with NVIDIA, including NVIDIA Vera CPU and NVIDIA Agent Toolkit for HPE Private Cloud AI. 

Plus, NVIDIA Confidential Computing extends across HPE AI Factory and enhanced full-stack NVIDIA integration — with NVIDIA accelerated computing, NVIDIA AI software and NVIDIA networking — is available throughout the entire portfolio.

NVIDIA Vera CPU Available With HPE Private Cloud AI

The HPE ProLiant Compute DL394 Gen12 with the NVIDIA Vera CPU will be available in 2027 with HPE Private Cloud AI, a turnkey AI factory co-engineered with NVIDIA. Vera is the first CPU built for agents — designed for the tool calls, orchestration and real-time data processing required across the agent loop — bringing deterministic, low-latency performance into HPE Private Cloud AI.

The New York Stock Exchange, in collaboration with Redpanda and HPE, is an early enterprise customer exploring Vera CPU with the HPE ProLiant Compute DL394 Gen12 server.

The Vera CPU is part of the NVIDIA Vera Rubin platform, which is ramping into full production with the NVIDIA Vera Rubin NVL72 rack-scale system available from HPE. Vera Rubin was built for frontier-scale models larger than 1 trillion parameters and will ship with full-stack NVIDIA Confidential Computing across every chip. 

HPE is also bringing the HPE Compute XD700 — built on NVIDIA HGX Rubin NVL8 — to the HPE AI Factory, supporting up to 128 Rubin GPUs per rack.

NVIDIA Agent Toolkit Now Available With HPE Private Cloud AI

NVIDIA Agent Toolkit — including NVIDIA Nemotron open models, the NVIDIA OpenShell secure runtime and NVIDIA NemoClaw blueprints — will be available with HPE Private Cloud AI. Together, they give enterprises an agentic AI operating system for monitoring agent behavior, enforcing governance policies, and safely building and running autonomous, long-running multi-agent systems.

HPE Private Cloud AI adds secure local agent registration, letting customers approve AI models, skills and tools against centralized governance and security policies before they run. New HPE Zerto Software capabilities detect rogue agent actions and use continuous data protection to rewind to a clean state.

On the data side, HPE Alletra Storage MP X10000 — which achieved the foundation level of NVIDIA-Certified Storage — automatically applies metadata and governance policies to prepare unstructured data for AI pipelines, improving token throughput.

NVIDIA Confidential Computing Across All HPE AI Factory Solutions

NVIDIA Confidential Computing is now available across the HPE AI Factory through HPE Services — including HPE AI Factory at Scale, HPE Sovereign AI Factory and HPE Private Cloud AI. 

AI applications access and use private and sensitive data that needs to be protected and secured. In addition, models trained with proprietary data or techniques need to be safeguarded from exfiltration. Confidential computing is essential for these modern AI workloads, as it protects models and private data during execution for on-premises and sovereign deployments, establishing a chain of trust through cryptographic attestation and encryption at every stage.  

In addition, HPE ProLiant Compute DL380a achieved certification as part of the NVIDIA-Certified Systems for NVIDIA Confidential Computing program, which validates robust application performance with confidential computing. These systems provide hardware-based protection for AI workloads and sensitive data assets while maintaining optimal NVIDIA acceleration.

Across the HPE AI Factory solutions, NVIDIA BlueField DPUs and NVIDIA DOCA provide in-silicon zero-trust policy enforcement, runtime threat detection and network encryption — protecting AI workloads, agents and data without performance tradeoffs.

Enhanced Full-Stack NVIDIA Integration Across the Portfolio

HPE AI Factory at Scale, HPE Sovereign AI Factory and HPE Private Cloud AI are now available with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, NVIDIA Spectrum-X Ethernet networking, NVIDIA BlueField-3 DPUs and NVIDIA ConnectX-8 SuperNICs. 

For next-generation AI factories, every Vera Rubin NVL72 system will ship with NVIDIA networking built in — NVIDIA Vera BlueField-4 DPUs, NVIDIA ConnectX-9 SuperNICs and NVIDIA Spectrum-X Ethernet — with NVIDIA Spectrum-6 switching delivering 1.6x higher networking performance for AI communication versus off-the-shelf Ethernet. 

Spectrum-X Ethernet networking is the standard for HPE AI Factory with NVIDIA — including at-scale, sovereign and turnkey AI factory solutions available now. For large-scale and sovereign workloads, HPE announced at NVIDIA GTC in March that it’s also adding NVIDIA InfiniBand networking options — including NVIDIA Quantum-X800 InfiniBand with the HPE Cray Supercomputing GX5000.

These configurations are based on NVIDIA reference architectures and support use cases from AI development through production-scale deployment, with NVIDIA AI Enterprise software and the HPE Unleash AI ecosystem.

At HPE Discover this week, the Unleash AI partner program is expanding with nearly a dozen new AI software partners — including Aizen, BridgeTEK, deepset, Deliverance, Faclon Labs, Gallop, Rocket, Supervity, Thales, Trustwise and Vortiqx.

Attendees can explore these solutions all week at the show and learn more about the HPE AI Factory with NVIDIA, part of the NVIDIA AI Computing by HPE portfolio.