Lip-bu Tan, brand new CEO of Intel, took advantage of his first important speech in the keynote of Intel Vision 2025 to recognize the errors that have led the company to the worst situation in its history and outline the general plans to recover its position in the semiconductor industry.
“It has been a difficult period for Intel for a long time”he declared so in his inaugural speech at the Intel Vision 2025 conference. «We are left behind in innovation. As a result, we have been too slow to adapt and meet your needs. You deserve something better, and we need to improve«.
Intel Vision 2025, challenges and opportunities
So addressed in front of partners and customers the product lines to debate, from artificial intelligence accelerators to the situation of their foundry business that remains unplug and will be key to Intel’s future. “From my first day of work, my number one priority has been to devote time to customers”he said. “From the first conversation, it was very clear that we have a lot of work ahead”.
AI accelerators
So highlighted the inference opportunity based on CPUS, a concept already explored by Intel, but without the expected results. So expressed its disappointment for the wide range of AI products for data centers that the company has, but that They are rarely mentioned as competitors Serious for NVIDIA and even AMD solutions.
Intel admitted that the sales of their Gaudi accelerators did not reach the income target of 500 million dollars and ruled out its next -generation GPU architecture to focus on a rack scale system that must compete in better conditions with NVL72 NVIDIA systems that use suppliers such as HPE.
«I am not satisfied with our current position. I know that you are either. I have heard the comments high and of course. It’s time to turn page »said the new CEO, promising to offer a competitive AI platformalthough it emphasized that “It won’t happen overnight”.
If Intel fulfills the promises of his CEO, the future Intel systems can exceed the products of the performance competition, they will be sold at a lower Price and consume less energy. If you meet this last point, it will be good news for data centers operators that still try to understand the NVIDIA plan for the monstrous 600 kW rack systems.
Intel Foundry
Samsung caused in 2017 a ‘change of era’ when he surpassed Intel in semiconductor revenue. At that time, the change of Era from the PC to mobility and the exponential business volume increase of the group responsible for Samsung Chips, ended with a era of dominance that had lasted almost a quarter of a century, when Intel became the largest world -semiconductor manufacturer in 1993 surpassing the almighty Japanese company nec.
For Intel It will be impossible to recover that leadership only with the sales of its own chips And a few years ago he announced the creation of an Intel Foundry division that would be dedicated to the manufacture of chips for third parties and that would expand it to any type of architecture beyond X86. I could even agree with TSCM a complicated alliance technically and strategically, but that would solve a good part of Intel’s problems.
Intel’s CEO approached the growing competition of rival chips designers who, combined with Intel’s failure to fulfill their own deadlines for new products, threaten the domain of their processors for servers and also leading desktop computers for three decades.
«We have to offer great quality, you deliver in time without exceptionand we have to have a culture of “first approves” and we are not there “said. «My motto is very simple: promise little and give a lot. That has been my personal seal »He added an always restrained Tan. The Executive promised to detail the manufacturing strategy at the Intel Foundry Direct event next month. Yes mentioned plans to expand the business, assign people appropriate to the position and improve efficiency.
“As we strengthen our Intel brand products, we are equally committed to building a great foundry”he said.
CPUS for servers and PCs
Intel’s products division has suffered delays and cancellations of launches recent years, surely derived from the great problems in the transition to the manufacturing processes of 10 Nm that still crawls.
For example, the XEON scalable processors of 4th generation, with name in Sapphire Rapid code, arrived almost two years late. This meant that AMD EPYC processors were the first CPU based on X86 compatible with the latest industry standards such as DDR5, PCIE 5.0 y CXL.
More recently, Intel implemented substantial cuts in the price of Xeon 6, its latest generation of processors for servers, in what was a significant change of strategy to maintain domain in data centers against AMD and also in front of other companies such as NVIDIA and its accelerators that intend to take the AI market.
If the server market is complicated, in Personal Computers AMD has cut distance in market share both in CPUS for desktops and in laptops. And there is another even greater threat, the growing display of ARM architecture in PCs with Apple at maximum banner, but also with another competition within the Windows platform. An example was that in the initial launch of the Copilot+ Qualcomm PCs was the only partner of Microsoft. Something unthinkable until a few years ago.
Engineering must be protagonist
The new CEO also expressed his desire that Intel return to its roots as a company that prioritizes engineeringa change that believes that it can be achieved by promoting a culture of “startup” obsessed with serving its customers. “Please be brutally honest with us”He said, arguing that Intel needs to do a better job collaborating with customers to design products that meet their needs.
He also said that he believes that Intel should look for growth opportunities in fields such as Quantum computinga field of which he is an expert as director of the startup of Celestial Fotonics AI, whose objective is to replace electronic interconnections with optical networks to accelerate large AI systems. Intel already showed a quantum silicon processor based on spin cubits, a technology that the company believes that can be produced in mass.
Other expansion fields mentioned were those of the Robotics and photonics. The Executive asked the audience to “remain attentive” to the news with the plans for Intel robotics. The company ventured years ago in silicon photonic technologies and last year demonstrated an optical chiplet packed together with a CPU.
Great ambitions those of Intel to recover their privilege position as ‘chip giant’ and good lip-bu start as CEO, for the recognition of errors and their will to listen and comply with customers: «We are here to serve and win your trust. I will not be happy or satisfied until we fulfill our promise consistently, on time, with quality and overcoming your expectations »said in the Keynote of the Intel Vision 2025, an event that must be a point and apart in the company.
