NVIDIA's GPU Technology Conference (GTC) for 2024 starts with the official unveiling of the tech giant's latest AI super chip, the "B200." An AI super chip claimed by NVIDIA's CEO, Jensen Huang, is a faster chip than its predecessors in some regards. The company's most recent AI initiatives were also presented during the conference.
In the center of Silicon Valley, Huang reportedly unveiled NVIDIA's newest technology, which is 30 times faster than its predecessor at some tasks, on a stage set like a hockey arena. Huang, who gave the new hardware its name in honor of mathematician David Blackwell, stated that it is a platform rather than merely a GPU, revealing the new B200 processor at its heart.
(Photo : Justin Sullivan/Getty Images)
A sign is posted at the Nvidia headquarters on May 25, 2022 in Santa Clara, California.
The flagship AI chip proves to be more energy efficient as it uses a 900GB/s ultra-low-power NVLink chip-to-chip connection, the NVIDIA GB200 Grace Blackwell Superchips also proved to be faster than its predecessors as it links two previous NVIDIA chips, the NVIDIA B200 Tensor Core GPUs, to the NVIDIA Grace CPU.
Although the B200 "Blackwell" chip is 30 times faster at certain tasks, such as providing chatbot responses, Huang did not provide specific information about how well the chip performs when processing massive volumes of data to train chatbots, the majority of NVIDIA's sales growth coming from this type of work. He did not provide a price either.
Read Also: OpenAI Plans Developing Its Own AI Chips Amid Global Shortages
AI Chip Demand
The faster AI chip comes after Huang pointed out that the data required for AI development will only rise rapidly, a statement that all tech giants prove to agree with.
Following their on-stage discussion at Intel's Foundry Event in San Jose California, last February, OpenAI's Sam Altman and Intel CEO Pat Gelsinger stated that the need for more and faster AI chips will only increase in the upcoming years.
In summary, Altman says that the world will want a lot more chips for artificial intelligence, but he also says that the world will need to invest in a lot of things that are not currently being thought about. Similar to Altman's remarks, Intel's Gelsinger asserts that there will be an unending demand for AI processors for a lot of years to come.
NVIDIA's AI-Powered Robots
NVIDIA's AI conference also included announcements of the company's partnership to develop more advanced humanoid robots.
Generalist Robot 00 Technology, a backronym inspired by the classic Marvel Comics superhero, was also unveiled by Huang as Project GR00T. Huang showed the audience several AI-powered robots on stage, including the Jetson Thor computing platform, which can carry out difficult jobs and have genuine human-robot interactions.
Huang stated that NVIDIA is bringing together several similar departments and initiatives within the chipmaker to create a comprehensive system for teaching humanoid robots that can learn from visual content, including videos and virtual reality.
Whether NVIDIA can keep its 80% market share in AI processors will depend on the chips and software the firm announced during its GTC 2024.
Related Article: Samsung-backed South Korean AI Chip Startups Eye Nvidia-led Global Market
