Nvidia's Vera CPU and $30 Billion Sovereign AI: The Two Stories Behind the Stock's Stalled Momentum

Nvidia reports $215.9B revenue (+65%) but stock lags; sovereign AI revenue triples to $30B, Vera CPU unveiled, while China export issues persist.

Nvidia is navigating a landscape of stark contrasts. Revenue hit $215.9 billion in fiscal 2026, up 65% year-on-year, and the latest quarter alone brought in $81.6 billion. Yet the stock trades around €178–€179, roughly 12% off its 52-week high of €202.50, with an RSI of 46.7 that signals a market deeply ambivalent. The tension between explosive operational growth and unresolved geopolitical risk has left investors waiting for the next catalyst.

That catalyst may come from two very different directions. On one hand, Nvidia is pushing deeper into enterprise infrastructure with the unveiling of the Vera CPU, its first processor purpose-built for agentic AI. On the other, the sovereign AI segment—governments building their own national AI infrastructure—has quietly tripled its contribution to the top line, generating $30 billion in fiscal 2026, up from roughly $10 billion in the prior year.

The Vera CPU represents a strategic long bet. Announced at HPE Discover in Las Vegas, it will handle tool-calling, orchestration and real-time data processing for AI agents that act autonomously. Slated for deployment in HPE ProLiant Compute DL394 Gen12 systems from 2027, the chip is embedded in HPE Private Cloud AI and comes bundled with the Nvidia Agent Toolkit and Confidential Computing across the entire HPE AI Factory. New configurations at the AI Factory at Scale level now scale up to 256 GPUs and cut inference costs to a tenth of the previous Blackwell generation.

The timing underscores that Nvidia is thinking beyond the current GPU-centric era. But the more immediate growth driver remains the sovereign AI boom. Governments from the Gulf to East Asia are treating GPU access as strategic infrastructure, not a commercial procurement. The model works for all parties: states get a credible AI strategy, data stays within national borders, and domestic champions gain real capability. Nvidia’s data centre revenue, which hit $75.2 billion in the most recent quarter—up 92%—drew roughly half from hyperscalers and half from a widening base including cloud providers, industrial firms, and these sovereign customers.

That sovereign surge partly offsets a painful loss. China once accounted for at least a fifth of Nvidia’s data centre sales. Eighteen months of US export controls have reshaped that market repeatedly. The Biden-era AI diffusion rule was revoked in May 2025, and a final rule in January 2026 shifted the H200 chip’s export licence presumption from denial to case-by-case review, tied to a 25% tariff, a 50% volume cap, third-party audits and stringent due diligence. From February 2026, small H200 licences were granted to select Chinese customers. Yet Nvidia has booked zero revenue under that programme. The door is technically open; commercially, it remains shut.

In late May 2026, the Commerce Department made clear that export licensing requirements for advanced AI semiconductors also apply to China-headquartered companies operating outside China, explicitly including Nvidia’s Blackwell and Rubin chip families. Nvidia’s response was that this merely confirmed its existing approach of seeking licences for deliveries to China-based firms. The net effect: a market that once delivered billions of dollars in quarterly revenue is now effectively blocked.

Against this backdrop, the Vultr deal announced mid-week offers a different growth path. Vultr, the largest privately held hyperscaler globally, has chosen Nvidia’s GB300 NVL72 platform built by HPE, coupled with Spectrum-X Ethernet networks at 400GbE and 800GbE. HPE will provide AI services, deployment expertise, lifecycle support and future liquid-cooling technology. The partnership shows Nvidia’s reach extending beyond the big public clouds into the expanding private cloud market.

The financials paint a broadly healthy picture. The stock sits around €179, roughly 10% above its 200-day moving average of €162.66, and the consensus analyst target of €258.02 implies upside of nearly 45%. But 30-day annualised volatility of around 40% means every regulatory headline can shift sentiment in an instant. The annual shareholder meeting on 24 June at 18:00 CET, held virtually, is likely to see management elaborate on the agentic AI strategy that took shape in Las Vegas—and perhaps offer clues on how Nvidia plans to navigate a regulatory environment that could tighten further.

The Trump administration is reportedly working on rules that would require US government approval for exports of Nvidia and AMD AI chips to every country, including close allies such as the UK, Germany and Japan. Those rules are not yet finalised. The direction is unmistakable. For a company valued at over €4 trillion, the gulf between regulatory tailwind and headwind is measured in hundreds of billions of dollars. Nvidia’s current consolidation near its moving averages is not a sign of weakness so much as a market counting the days until Washington makes its next move.