NVIDIA AI Updates: May 19, 2026

1. Jensen calls AI demand “utterly parabolic” as Dell’s Vera Rubin NVL72 PowerEdge ships for agentic inference

NVIDIA. Jensen Huang used his Dell Technologies World keynote on May 18 to call AI demand “parabolic, utterly parabolic,” and anchored the message around a new Dell PowerEdge generation built on the NVIDIA Vera Rubin NVL72 platform aimed at large-scale agentic inference at sharply lower cost-per-token. The flagship liquid-cooled PowerEdge XE9812 is the headline SKU on Vera Rubin NVL72, with XE9880L, XE9882L, and XE9885L filling out an HGX Rubin NVL8 tier for existing data center power envelopes, and the R9822 and M9822 debuting NVIDIA Vera CPU options for enterprise workloads. Huang named Lilly, Samsung, and Honeywell as enterprises already deploying Dell AI Factories at scale, framing this as the moment “AI factories” stop being a marketing pitch and start showing up as line items on enterprise capex plans. Source

2. Dell AI Factory partner lineup pitched as the standard build for agent-era enterprise inference

NVIDIA. Beyond the headline Vera Rubin announcement, the joint Dell-NVIDIA pitch at Dell Technologies World framed the Dell AI Factory as the default enterprise stack for the agent era — compute, networking, storage, software, and validated reference architectures shipped as a unit rather than assembled piecemeal. Over 4,000 customers are now cited as deployed on Dell AI Factory, with early adopters reportedly seeing roughly 2.6x ROI in the first year. The framing matters going into NVIDIA’s May 20 Q1 fiscal 2027 earnings: it reinforces the agentic-inference narrative the company has been pushing to offset concerns about hyperscaler ASIC encroachment and the steep drop in NVIDIA’s reported China market share. Source