Intel CEO criticizes Nvidia's "AI moat": The entire industry is motivated to eliminate the CUDA market

Intel launches a new AI chip and confronts Nvidia head-on.

According to media reports on December 15th, Intel CEO Pat Gelsinger announced the launch of the Core Ultra processor, the 5th generation Xeon scalable processor, and the 3rd generation AI accelerator Gaudi 3 at an event in New York.

Gelsinger criticized Nvidia's CUDA (a computing platform developed by Nvidia), stating that its technological moat is shallow and small, and claimed that inference technology will be more important than training for artificial intelligence.

In an interview with the media, Gelsinger said that Nvidia's dominance with CUDA will not last forever.

Gelsinger said, "You know, the whole industry has the motivation to eliminate the CUDA market." He cited examples of companies such as MLIR, Google, and OpenAI, implying that they are shifting towards a "Python-based programming layer" to make AI training more open.

Gelsinger added, "The entire industry has the motivation to bring more extensive technology to wide-ranging training, innovation, and data science."

However, Intel does not rely solely on training; instead, it believes that inference is the right way forward.

Gelsinger stated that when you embrace inference, once you have trained the model, you no longer need to rely on CUDA. This does not mean that Intel will not compete in the training field, but fundamentally, the focus of competition is the inference market, and Gaudi 3 is the product Intel uses to enter the AI inference market.

Gelsinger also took this opportunity to promote OpenVINO, which is a standard developed by Intel specifically for its AI work, and predicted the emergence of a "hybrid computing" world, where some computing occurs in the cloud and the other part happens on personal computers.