The Entire AI Industry Wants to End CUDA, According to Intel's CEO
Pat Gelsinger, Intel's CEO, has recently stated that the entire AI industry wants to eliminate CUDA (Compute Unified Device Architecture) from the market. This statement was made during the "AI Everywhere" event, where Gelsinger defended the official stance of his company regarding the AI sector.
CUDA is a software development platform that includes a compiler and tools used by programmers to develop software for NVIDIA's GPUs. Replacing it with another option in ongoing projects is a challenge. This is precisely the strongest suit that this technology has in its favor. While there are alternatives to this work environment, such as CANN (Compute Architecture Neural Networks), which is Huawei's proposal, they all cost significantly less to abandon NVIDIA's technology.
Gelsinger assured that Intel is not alone in the group of companies that want to eradicate CUDA from the AI industry. However, he may have exaggerated when he said "the entire industry." During his speech, he mentioned that Google and OpenAI are some of the companies with a significant weight in the AI industry that want to leave CUDA behind. Furthermore, he stated that Intel is determined to pursue all commercial opportunities that arise as they compete with NVIDIA, AMD, or any other player.
JENSEN HUANG RESPONDS WITH A FIRM CONVICTION
As could be expected, NVIDIA's CEO, Jensen Huang, has responded to Gelsinger's statements. During an event organized by CASPA (Chinese American Semiconductor Professional Association), Huang discussed a very interesting topic that is currently in the limelight: how GPUs have evolved from being specialized graphics chips to taking center stage in the AI field. He also explored where the future of this hardware lies, which is undoubtedly promising.
If the industry as a whole wants to leave CUDA behind, NVIDIA will accept it. However, one of the most interesting points that Huang addressed was CUDA's role in NVIDIA's future. Huang stated that their goal is not to manufacture GPUs with CUDA. From this statement, it is clear that NVIDIA's business is entirely focused on hardware. Therefore, if the industry as a whole wants to leave CUDA behind, NVIDIA will undoubtedly accept it. However, Jensen Huang did not miss the opportunity to remind his interlocutors that his company currently leads the AI industry with undisputed prominence.
One of the founders of CASPA, who is also an AI specialized engineer, took advantage of the occasion to ask Huang if the architecture of GPUs will continue to dominate the hardware industry for AI in the future and if it will have the opportunity to dethrone NVIDIA. Huang responded by ensuring that GPUs have the capacity to adapt to different use scenarios as human needs change. He concluded by stating that NVIDIA will maintain its leadership position in the future, and the opportunities for its competitors, although they exist, will be minimal. And we will see.
More information: CASPA
China is unbeatable in artificial intelligence. Not everyone says it, but the director general of Google does.