The Falcon Shores project will survive Intel’s reorganization

With this, the company slightly nuanced Pat Gelsinger’s previous statement to keep investors hopeful.

Back in mid-September, we wrote about a sentence that could upset the mood of Intel investors, in which Pat Gelsinger practically admitted that it is not yet able to compete in the training phase of machine learning, so in this market NVIDIA, AMD, Google and Amazon is better than them. This is particularly unpleasant from the point of view of the Falcon Shores design, which Intel is releasing later than planned, and not as the originally intended XPU, but in the form of a more traditional GPU.

Since Falcon Shores is considered to be the only product that Intel plans to call the training phase of machine learning, which moves extremely large amounts of money, it was not very lucky to reveal in advance that the upcoming product will not have much of a chance against the four main competitors. This was also felt by the company, so the HPCWire information was recently given to him that the Falcon Shores project will definitely survive the reorganizations and that its fate will develop in accordance with the most recently published data.

All this does not contradict Pat Gelsinger’s words, meaning that the first generation might not be competitive, but it shows that Intel would not let go of one of the fattest markets, so in a few generations they would catch up with the others, which investors it can give hope for him as well.