Jensen Huang’s Big Week: China Trip, Vera Rubin, and a UK AI

NVIDIA CEO Jensen Huang had one of his most consequential weeks in recent memory — joining President Trump’s China summit at Trump’s personal request, announcing a partnership with British AI startup Ineffable Intelligence, and delivering a keynote at Dell Technologies World where he declared AI demand is “going parabolic, utterly parabolic.” The flurry of activity, spanning May 12–14, 2026, reflects NVIDIA’s expanding role not just as a chipmaker but as a central actor in global AI diplomacy and infrastructure.

Huang Joins Trump’s China Trip After Presidential Call

NVIDIA confirmed to CNBC on Tuesday, May 12, that Huang would travel to China as part of President Trump’s delegation for a summit with Chinese President Xi Jinping scheduled for Thursday and Friday. The confirmation came after an unusual sequence: initial reports suggested Huang would not be part of the delegation, but after media coverage of his absence circulated, Trump personally called Huang to invite him, according to a source cited by CNBC.

Speaking to reporters on Thursday, May 14, Huang confirmed the account directly. “President Trump asked me to come,” he said, according to CNBC. He described the summit as “one of the most important summits in human history.”

The optics are significant. NVIDIA has faced sustained regulatory pressure over chip export controls to China, and Huang’s inclusion in a high-level diplomatic mission signals that AI hardware supply chains are now a live geopolitical variable — not merely a trade policy footnote. An NVIDIA earnings call is expected to follow the summit, and analysts are already watching for whether Huang addresses China sales directly, per CNBC reporting.

Vera Rubin NVL72 and the Dell AI Factory Push

At Dell Technologies World on Monday, May 12, Huang took the keynote stage alongside Dell founder and CEO Michael Dell to outline NVIDIA’s latest hardware milestones and the scale of enterprise AI adoption.

Dell opened with a market sizing claim: worldwide AI infrastructure spending could reach $3–4 trillion by 2030, with token consumption projected to grow 3,400% over the same period, according to the NVIDIA AI Blog. “There is a massive AI investment boom that’s already underway, and a productivity boom is beginning,” Dell said.

Huang followed with specific performance figures for NVIDIA’s next-generation hardware:

• NVIDIA Vera Rubin NVL72 delivers agentic AI inference at one-tenth the cost per token compared to prior configurations
• NVIDIA Vera CPU runs agent sandboxes 50% faster than traditional CPUs
• Enterprise data queries run up to 3x faster on the Vera CPU
• 5,000 enterprises — including Eli Lilly, Samsung, and Honeywell — are running AI workloads on Dell AI Factories with NVIDIA

The Vera Rubin NVL72 is NVIDIA’s rack-scale system combining the Vera CPU with Rubin-generation GPUs, positioned as the successor architecture to the current Blackwell generation. Its appearance in a production-deployment context at Dell Technologies World suggests commercial availability timelines are firming up, though NVIDIA has not announced a specific ship date in these sources.

NVIDIA Bets on UK Startup Ineffable Intelligence

On Wednesday, May 13, NVIDIA announced a partnership with London-based AI startup Ineffable Intelligence to develop what Huang described as the “next frontier” of AI systems, according to CNBC.

Ineffable Intelligence’s approach diverges from the dominant paradigm in foundation model development. Rather than training primarily on human-generated data, the startup focuses on reinforcement learning — training AI systems to learn continuously from experience rather than from static datasets.

“The next frontier of AI is superlearners — systems that learn continuously from experience,” Huang said in the announcement.

The partnership was announced just months after NVIDIA disclosed a separate investment in another AI startup, signaling an accelerating pattern of strategic bets on post-pretraining architectures. Ineffable Intelligence is a relatively early-stage company; CNBC described it as a startup without providing funding figures or headcount in the available reporting.

The reinforcement learning angle is notable given broader industry interest in moving beyond static model training. Systems that can update from real-world interaction — rather than requiring expensive retraining cycles — would reduce inference costs and improve adaptability, two constraints that currently limit enterprise AI deployment at scale.

Demand Signal: “Utterly Parabolic”

Huang’s language at Dell Technologies World was unusually direct, even by his standards. “We’ve now arrived at the era of useful AI, which is the reason why demand is going parabolic, utterly parabolic,” he said, according to the NVIDIA AI Blog. “What took months now takes weeks. What took weeks now…”

The phrasing tracks with NVIDIA’s financial trajectory. The company has posted consecutive quarters of record revenue driven by data center GPU sales, and the Blackwell architecture — which underpins the current H100/H200 successor lineup — has been supply-constrained since its introduction.

The Dell partnership provides a concrete distribution channel for that demand. Dell AI Factories, which bundle NVIDIA GPUs with Dell server infrastructure, now serve 5,000 enterprise customers. That figure represents a substantial installed base for NVIDIA’s next hardware generation to upgrade into — a dynamic that supports sustained revenue visibility even before new chip architectures ship at volume.

What This Means

Huang’s week illustrates how NVIDIA’s strategic surface area has expanded well beyond chip design. Three separate events — a geopolitical summit, a hardware keynote, and a startup partnership — each carry distinct implications for the company’s medium-term position.

The China trip is the most consequential near-term variable. Export controls have blocked NVIDIA from selling its most capable chips to Chinese customers, and the Trump-Xi summit could either tighten or relax those restrictions. Huang’s presence in the delegation suggests the administration views NVIDIA’s market access as a negotiating chip — or at minimum, that Huang’s direct relationships in China are considered useful context for the talks.

The Vera Rubin NVL72 performance figures, meanwhile, set a new cost-efficiency benchmark for agentic AI workloads. A 10x reduction in cost per token for inference — if it holds in production — would materially change the economics of deploying AI agents at enterprise scale, accelerating adoption timelines for the 5,000 Dell AI Factory customers already in the pipeline.

The Ineffable Intelligence bet is the longest-duration play of the three. Reinforcement learning-based “superlearner” systems are not yet production-ready at commercial scale, but NVIDIA’s early positioning here mirrors its historical pattern of backing infrastructure for the next compute wave before it arrives.

Collectively, the week’s activity reinforces that NVIDIA is not merely reacting to AI demand — it is actively shaping where that demand flows next.

FAQ

Why did Jensen Huang join Trump’s China trip?

According to CNBC, Huang said President Trump personally called him and asked him to join the delegation. Initial plans did not include Huang, but Trump reached out after media coverage noted his absence.

What is the NVIDIA Vera Rubin NVL72?

The Vera Rubin NVL72 is NVIDIA’s rack-scale AI system combining the Vera CPU with next-generation Rubin GPUs. According to the NVIDIA AI Blog, it delivers agentic AI inference at one-tenth the cost per token compared to prior configurations and is positioned as the successor to the current Blackwell architecture.

What does Ineffable Intelligence do?

Ineffable Intelligence is a British AI startup focused on reinforcement learning — training AI systems to learn from experience rather than static human-generated data. CNBC reported that NVIDIA partnered with the company on May 13, 2026, with Huang describing its goal as building “superlearners” that improve continuously through interaction.