Nvidia GTC Preview: Sovereign AI and Blackwell Take the Stage

Nvidia is set to host its upcoming GTC conference in Washington, D.C., for two days. Unlike the developer-oriented GTC event held in March, this gathering primarily targets government contractors.

CEO Jensen Huang will attend and deliver his trademark keynote speech.

Why did Jensen Huang suddenly move Nvidia’s GTC to Washington, D.C.?

The answer might go far beyond a change in venue. With global governments rushing to build “sovereign AI” systems, this year’s GTC could reveal how Nvidia plans to dominate the next phase of the AI arms race — one that’s no longer just about chips, but national power.

The rise of sovereign AI has already become a global reality. During the previous earnings call, Nvidia executives revealed that the European Union plans to invest €20 billion to build 20 AI factories across France, Germany, Italy, and Spain, including five “super factories.” These facilities will expand the EU’s AI computing infrastructure tenfold.

In the United Kingdom, the Isambard-AI supercomputer, powered by Nvidia’s technology, is now the country’s most powerful AI system.

Nvidia’s revenue from sovereign AI projects is expected to exceed $20 billion in 2025, doubling from the prior year.

Once a company is listed on national sovereign AI procurement rosters, it can secure large, long-term contracts — and Nvidia has no intention of missing out.

By hosting the GTC conference in Washington, D.C., Nvidia is signaling strong confidence in joining the U.S. sovereign AI procurement programs.

In truth, the U.S. government has few alternatives — Nvidia’s technology remains unmatched globally. Still, Jensen Huang aims to show deference to President Trump by holding the event in the nation’s capital, a strategic gesture in both optics and diplomacy.

What Might Jensen Huang Announce?

Beyond sovereign AI, the GTC conference will also focus on Nvidia’s latest AI product lineup.

The highlight is expected to be the “Blackwell” series GPUs, particularly the high-end “Blackwell Ultra.”

These chips feature about 50% more memory capacity than current models and offer a significant performance boost.

Investors are also eagerly anticipating details on Nvidia’s next-generation “Vera Rubin” architecture, planned for launch in 2026, which could be unveiled at this event.

Nvidia is also ramping up deployment of its Spectrum-X Ethernet infrastructure, optimized for AI workloads and capable of significantly reducing model training costs.

Many major cloud providers are already adopting this technology. At GTC, Nvidia is expected to announce the new Spectrum 5 switch and the next-generation ConnectX-8 SmartNIC.

Robotics, Quantum Computing, and AI Healthcare

Nvidia is also expected to share updates on projects it has invested in — spanning robotics, quantum computing, and AI-driven healthcare. These initiatives reflect the company’s ambition to expand beyond chips into broader AI ecosystems.

Importantly, this GTC will not be solely Nvidia’s stage.

The exhibition hall will feature a dedicated startup pavilion, where numerous AI startups will compete to attract investors and potential partners.

The collaboration and competition at this venue will highlight the growing AI industrial ecosystem that Nvidia has helped cultivate worldwide.