Jensen Huang Promises “Chips the World Has Never Seen” at NVIDIA GTC

NVIDIA is preparing major silicon reveals at GTC 2026, according to comments made by CEO Jensen Huang and reported by Wccftech. The company has reportedly lined up “new chips the world has never seen before,” setting high expectations for its upcoming keynote.

Huang did not confirm specific product names, but the statement alone has fueled speculation across the AI and semiconductor industry.

Rubin Success Sets the Stage for the Next Leap

At CES 2026, NVIDIA formally introduced its Vera Rubin AI lineup into full production. The platform includes six new chips, combining Vera CPUs with Rubin GPUs, aimed at accelerating AI workloads at scale.

Rubin is also expected to significantly drive HBM4 demand, reinforcing NVIDIA’s influence across the memory supply chain. With Hopper and Blackwell previously optimized for large-scale AI model pre-training, NVIDIA’s newer architectures are increasingly shifting toward inference, latency reduction, and higher memory bandwidth.

Grace Blackwell Ultra and Vera Rubin reflect that transition, targeting real-time AI applications and enterprise deployment rather than purely training-focused clusters.

What Could NVIDIA Reveal at GTC 2026?

The exact chips that will be unveiled remain unknown, but industry observers point to possible Rubin derivatives such as Rubin CPX, or even next-generation Feynman chips.

Feynman remains largely unconfirmed, yet rumors suggest a potentially disruptive design. Reports hint at SRAM-heavy architectures and advanced 3D stacking approaches, possibly integrating LPUs for improved efficiency and throughput.

As AI compute demand shifts from pre-training toward inference-heavy workloads, NVIDIA appears to be evolving its silicon strategy accordingly.

AI as a Full-Stack Industry

Huang described AI not simply as a chip race, but as a full-stack transformation spanning semiconductors, energy infrastructure, hyperscale data centers, cloud platforms, and end-user applications.

He emphasized partnerships, startup collaborations, and acquisitions as core pillars of NVIDIA’s continued leadership in the AI sector. This broader ecosystem approach positions NVIDIA as more than a GPU manufacturer, as it operates as a central infrastructure provider in the AI economy.

NVIDIA’s GTC 2026 keynote is scheduled for March 15 in San Jose, California. The event is widely expected to outline the next phase of AI infrastructure development, potentially introducing silicon that reshapes compute efficiency, memory integration, and inference scaling.

In related news, NVIDIA has also expanded DLSS 4 support across four additional games, while early comparisons show DLSS 4.5 outperforming competing upscaling technologies.