Jensen Huang claims that Nvidia will “surprise the world” with a new chip at GTC 2026
Nvidia plans to “surprise the world” at GTC 2026
Next month, Nvidia will host its annual GTC (GPU Technology Conference). According to Korean media (as reported by WCCFTECH), Nvidia’s CEO, Jensen Huang, said the company plans to announce a chip that will “surprise the world” at the event.
Unfortunately, Huang did not provide any details about the new chip. I guess that would ruin the surprise. However, we do know that GTC 2026’s keynote will be on March 16th at 11 am Pacific Time (PT). We also know that Nvidia’s focus for this event is “AI, accelerated computing, and transformative tech”.
A chip that will surprise the world will be unveiled at GTC next month.
Nvidia’s new chips might not be a new graphics card or AI accelerator. After all, Nvidia offers CPUs, networking solutions, and other accelerator chips. Several reports indicate that Nvidia has partnered with SSD makers such as SK Hynix and Kioxia to develop “AI SSD” hardware. These solutions reportedly target 100 million IOPS of performance, accelerating AI workloads constrained by the speed of traditional SSDs and the capacity of HBM memory modules. Could Nvidia unveil its first “AI SSD” at GTC 2026?
We’ve prepared several new chips the world has never seen before.
Nothing is easy because all technologies are at their limits,
With a team like this (consisting of NVIDIA and SK Hynix memory engineers), nothing is impossible.
With hyperscalers building datacenter-scale AI setups, Nvidia has several avenues to boost AI performance. First, they can build more powerful AI accelerators, and second, they can boost the performance of the solutions used to interconnect AI accelerators and datacenter racks. Finally, Nvidia can utilise more performant memory and storage mediums to either feed more data to their AI hardware or do so at a faster rate. Major improvements in any of these areas is good news for Nvidia.
You can join the discussion on Nvidia’s GTC 2027 plans on the OC3D Forums.
