Posted on Jan 24, 2021 4:24 PMUpdated Jan 24, 2021, 4:40 PM
So nothing would have to change for everything to change? Pat Gelsinger, the former chief technology officer of Intel who returns to the helm of the world leader in microprocessors after a decade spent at EMC and VMWare, could not however be satisfied to reverse the formula of the “Cheetah” if not to reverse the table. Even before taking office on February 15, Intel’s third CEO in three years warned that by 2023, ” majority »Semiconductors would continue to be produced in-house (compared to 80% currently) and that the objective remained to regain technological leadership from Taiwanese TSMC. The former architect of one of the chips which made the glory days of the firm of Santa Clara, the i486, has put another in the ears of the investors: Intel is part of the ” national heritage From the United States, a rather clever way, in the Biden era, of dismissing activist Dan Loeb in his goals. The financier demands the split of the ten “fab” of the group in the name of the sustainability of ” America’s access to the latest generation semiconductors “. Gelsinger is trying to buy time, but his appeal in Washington has paid off on the stock market for the loss of two-thirds of the gain since the announcement of his arrival. Intel is now worth four and a half times less than TSMC, NVidia and AMD combined, which it equaled just five years ago. The flea is waiting for giant boots.
Although Intel has always had a reputation for using internal promotion to run the business, Pat Gelsinger will be the first semiconductor engineer to become general manager of Intel since 2005. His predecessor, Bob Swan, was a graduate. in business administration, and the boss from 2013 to 2018, Brian Krzanich, was a chemical engineer. You have to go back to Craig Barrett (1998-2005, doctor of materials science) to find a specialist in miniaturization. Intel’s board of directors now has four engineers (out of ten members), but only one expert in these miniaturization issues, which the group stumbles.
Critics of Intel criticize the obsession of its engineers to pursue the trajectory of power and density of integrated circuits desired by Moore’s Law, to the detriment of processor performance adapted to the needs of its customers. The law established in 1965 by Gordon Moore, one of the co-founders of Intel, which wants the number of transistors on a chip to double every two years, tends to be exhausted because it affects the physical limits of the atom. NVidia boss Jensen Huan proclaimed it “ death ” several times. Last summer, Intel engineers estimated they could further increase the number of circuits by 50 in the next ten years, 20 times less than the pace predicted by Moore’s Law. In fact, the latest advances from TSMC and Samsung Electronics on Intel have resulted in less power gain.