NVIDIA CEO Jensen Huang said AI infrastructure demand has gone parabolic as Dell Technologies unveiled a refresh of its AI Factory product line at Dell Technologies World on May 18. The centerpiece is the Dell PowerEdge XE9812 built on NVIDIA Vera Rubin NVL72, which delivers up to 10x lower cost-per-token than Blackwell for large-scale agentic AI inference. Dell cited 5,000 enterprises including Lilly, Samsung, and Honeywell now running AI workloads on Dell AI Factories with NVIDIA, with worldwide AI infrastructure spending projected to reach $3-4 trillion by 2030.
“We've now arrived at the era of useful AI, which is the reason why demand is going parabolic, utterly parabolic.”— Jensen Huang, NVIDIA CEO
Dell chairman and CEO Michael Dell framed the stakes in terms of token consumption, which the company projects will grow 3,400% by 2030. The new systems target that growth with liquid-cooled, high-density GPU configurations and NVIDIA's Vera CPU, purpose-built for agentic workloads that cycle between data queries, sandboxed code execution, and model inference.
The Dell PowerEdge XE9880L, XE9885L, and XE9882L servers are the first Dell systems built on NVIDIA HGX Rubin NVL8, supporting up to 144 GPUs per rack with 100% direct liquid-cooled compute nodes and up to 5.5x the performance of HGX B200. Dell also introduced PowerRack, a fully integrated system combining compute, networking, and storage with thermal design and power management engineered together.
Key facts
- 01Dell and NVIDIA launched Vera Rubin NVL72 systems with 10x lower cost-per-token than Blackwell for agentic AI inference.
- 025,000 enterprises including Lilly, Samsung, and Honeywell are running AI workloads on Dell AI Factories with NVIDIA.
- 03Worldwide AI infrastructure spending could reach $3-4 trillion by 2030, with token consumption projected to grow 3,400%.
- 04NVIDIA Vera CPU delivers 50% faster agent sandboxes and 3x faster enterprise data queries versus traditional CPUs.
- 05Dell PowerEdge XE9880L and related servers support up to 144 GPUs per rack with 5.5x the performance of HGX B200.
NVIDIA Vera CPU, introduced across Dell PowerEdge M9822 and R9822 servers, brings 1.2 TB/s memory bandwidth to the enterprise AI factory. The chip completes agentic workloads 50% faster than x86 processors, with Starburst delivering 3x faster query throughput on Vera for large-scale SQL analytics. Dell's AI Data Platform now includes accelerated data engines built on NVIDIA CUDA-X libraries, including cuDF for structured data and cuVS for unstructured data.
On-premises deployment is the dominant mode. Dell's survey found that 67% of AI workloads now run outside the cloud, with 88% of respondents running at least one AI workload on-premises. The company is addressing that shift with NVIDIA Confidential Computing as the foundation for securely deploying frontier models inside the enterprise without exposing model IP or data.
Google Distributed Cloud with Gemini 3.0 is now available in preview on Dell PowerEdge XE9780 servers accelerated by NVIDIA Blackwell and secured by NVIDIA Confidential Computing. SpaceXAI will bring its latest models on-premises to the Dell AI Factory with NVIDIA Confidential Computing keeping model weights and enterprise data protected end-to-end. NVIDIA Nemotron models and Reflection's open-source frontier AI models are also coming on premises.
Lilly executive vice president Diogo Rau joined the keynote to discuss the pharmaceutical company's AI-driven advancements in life sciences, powered by AI infrastructure deployed at scale with Dell and NVIDIA. Honeywell CTO Suresh Venkatarayalu walked through the company's move from public cloud to on-premises AI using the Dell AI Factory for industrial AI use cases, digital twins, and automation from data center to the edge. Hudson River Trading is expanding its Dell deployment to power AI-driven research with Dell PowerEdge XE9685L servers.
Dell announced new software partnerships spanning agentic AI and code assistants. Palantir's sovereign AI OS reference architecture with NVIDIA, announced in March, now runs on Dell infrastructure for on-premises deployment of Palantir Ontology and AIP. ServiceNow customers will be able to leverage the Dell AI Factory to bring together infrastructure and enterprise workflow automation. Dell also announced new solutions with Fogsphere, Ipsotek, Mistral AI, Poolside, and Uneeq, as well as security partnerships with CrowdStrike and Fortanix.
OpenAI Codex will connect with the Dell AI Data Platform to help customers bring Codex closer to internal context including codebases, documentation, business systems, and operational knowledge. Dell and OpenAI will also explore how Codex can connect with the Dell AI Factory.
The product line extends to deskside systems. Dell Pro Max with GB10 and GB300 powered by the NVIDIA Grace Blackwell architecture, as well as Dell Pro Precision systems powered by NVIDIA RTX PRO Blackwell workstation platforms, will run Dell Deskside Agentic AI with the NVIDIA NemoClaw stack, NVIDIA OpenShell runtime, and NVIDIA Nemotron open models.
The 3,400% token consumption growth projection assumes the shift from pilot to production continues at pace. If agentic AI proves less reliable in multi-step tasks than current benchmarks suggest, enterprises may throttle deployments, which would slow infrastructure demand. The 10x cost-per-token advantage of Vera Rubin NVL72 over Blackwell also assumes workloads are inference-heavy and long-running, a profile that fits some but not all enterprise AI use cases.
Dell and NVIDIA are building for a world where the enterprise AI factory runs continuously, not episodically. The question is whether the market validates the premise that most AI value accrues on-premises rather than in the cloud. If 67% of workloads are already running outside the cloud, the tipping point has passed. If that number represents early-stage experimentation rather than sustained production workloads, the $3-4 trillion infrastructure spend may land later than 2030.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
