Intel effectively kills Falcon Shores, the next -generation GPU for high -performance computing and AI workload.
While competitors such as AMD and NVIDIA are gaining status, the release of many disappointed products and historical loss, then Intel tries to correct the course.
Intel Co-CEO MICHELLE JOHNSTON HOLTHAUS said in a call of the company’s fourth quarter, Intel said that “Falcon SHORES is an internal test-chip and use Falcon Shores without introducing it to the market.”
“AI Data Center is an attractive market for us,” said Horsus. “”[B]I am not satisfied with where we are today. We are still not participating in the cloud -based AI data center market in a meaningful way. One of the immediate actions I took is to simplify the roadmap and focus the resources. “
Instead, the focus is in the Jaguar Shears. Horsus was called Intel’s opportunity to “deal with AI data centers more widely to develop system -level solutions in rack scale.”
Holthaus has relieved the expectations of Falcon SHORES last month. She suggested that Gaudi 3, a previously dedicated AI data center chip of the company, is a “repetition” step.
“One of the things I learned from Gaudi is not enough to deliver silicon,” said Holthaus in a revenue call on Thursday. “Falconquers is useful for system, networking, memory -all components.[s]。 However, what the customer really wants is the full -scale rack solution, so it can be reached by Jaguarshs. “
Gaudi 3 is widely seen as Intel’s mistake. In November, the company said that due to software -related issues, the sales of Gaudi 3 could not achieve $ 500 million goals. Today, few major service providers other than IBM are committed to using chips.
Intel faces a difficult battle in the AI Data Center chip space. Rival AMD expects to increase AI chips of about $ 7 billion in 2025, but some analysts can reach $ 195 billion in 2026 years of accounting year.
“When we are thinking about our AI opportunity, my focus is focusing on the problem we are trying to solve, especially the need to reduce costs and increase the efficiency of calculation,” said Holthaus. I mentioned it. “Therefore, all -purpose approaches do not work. You can use the core assets in a new way to see a clear opportunity to promote the most convincing total cost of the whole body.”
Source link
