Katarzyna Petru
Nvidia is once again heating up the atmosphere around its hardware plans. Jensen Huang announced the launch of a "chip that will surprise the world," and the industry is already trying to decipher what lies behind this slogan. The declaration was made during an informal meeting with engineers in Santa Clara, as reported by The Korea Economic Daily. It was not an official presentation but a spontaneous Q&A following dinner. Such moments often reveal more than carefully prepared conferences. The key question, however, is different. Will the announced chip have real significance for ordinary users, or is it just another step in the race for dominance in data centres?
Rubin, Feynman or something completely new?
Nvidia has not provided any specifics. We do not know the name, segment, or parameters. This opens the floor to speculation. One of the candidates is the development of the Vera Rubin platform with HBM4 memory. It has already been officially announced, so it is hard to consider it a surprise element, but a more advanced version could possibly be showcased. The second hypothesis concerns the Feynman architecture, which has been the subject of industry rumours for months. It is said to involve: greater integration of SRAM memory directly within the chip structure, aggressive use of 3D stacking technology, and further optimisation for inference and working with large AI models. Such a chip would be a natural step following the Hopper and Blackwell generations, which solidified Nvidia's position in the artificial intelligence sector.
Or perhaps finally something for PC?
One cannot rule out a less obvious scenario. The industry has long been talking about the N1X processor, which is ARM64 for personal computers. Such a chip could compete with the upcoming constructions from Intel and open doors for Nvidia to the classic PC market. The problem is that Huang has been focusing almost exclusively on AI and server infrastructure in his recent appearances. It is this segment that drives the company's valuation and investor interest. Therefore, the most likely scenario is another accelerator for data centres, rather than a graphics card for gamers.
Everything will be clarified during the Nvidia GTC 2026 conference, which will take place on 16 March in San Jose. This event has been the stage for Nvidia's biggest premieres for years and a signal of the direction in which the semiconductor industry is heading. If the announced chipset indeed turns out to be groundbreaking, it may further strengthen the company's dominance in the AI sector. However, for gamers and PC users, the most important question will be whether the new technology translates in any way into consumer products.
Nvidia is once again building anticipation ahead of GTC. The slogan about the "chip that will surprise the world" sounds impressive, but for now we don't know if it pertains to a breakthrough for the mass market or another step in the development of AI infrastructure for the largest tech companies. We will only find out the answer in March.
Source: The Korea Economic Daily
