Intel’s strategy for the future is to go back to 2005. It was then that Paul Otellini, the CEO of Intel, turned down the opportunity to be the manufacturer of the chip that would be at the heart of the iPhone. The result was that Apple went to Taiwan Semiconductor to produce that chip. In 2005, Intel was the world’s most valuable semiconductor company. It was riding on the PC wave that started in the 80s. This wave was about to peak, but Intel could not recognize this. Intel missed every significant opportunity that has come along in this century, with the possible exception of providing the microprocessor to power internet servers in the cloud. And that too is vanishing.
Intel’s decision not to manufacture the chip for the iPhone caused Apple to move to Taiwan Semiconductor as its manufacturer. It gave that company the scale to build a world-class semiconductor manufacturing capacity and effectively changed the nature of the semiconductor business, separating design capabilities from manufacturing. Companies that manufacture semiconductors for other companies are called foundries. Foundries operate on low margins. Intel was designed for high margins. It did not want to be a foundry.
Since Pat Gelsinger returned to Intel a year ago to become its CEO, he has decided to recreate the Intel of his youth, the Intel that led the world in semiconductor manufacturing. But to do that in a world no longer dominated by the PC and therefore the Intel X86 architecture, Intel would also have to become a foundry. To accomplish this requires significant investment in factories. As a result, Intel has announced investments of tens of billions of dollars to build new factories in the USA and Europe.
Right now, demand for semiconductors leads supply. But when that happens, all the major semiconductor companies invest heavily in production. I think that Samsung, Taiwan Semiconductor, and Intel each will all build-out for 60% of the capacity needed. The total manufacturing capacity could exceed 180%. The result of this will be plunging prices and lower and lower margins.This has always been the nature of the semiconductor business. Intel was able to escape this because of the success of the PC and its propriety hold on the X86 architecture. I believe that of the three companies mentioned, Intel will suffer the most.
Pat was a brave man to take over the helm of Intel. Pat sees himself, it appears, as the Steve Jobs of the semiconductor world, bringing back Intel to its glorious past. But to paraphrase, “I knew Steve Jobs and you sir are not Steve Jobs .”Jobs did not reach back into the past. Instead, he created a new future in which Apple could dominate. I see no evidence of that for Intel.
It saddens me to write this. Intel was good to me in many ways. After I left in 1999, I offered my advise to the CEOs various times, but they didn’t take me up. I like to think it might have made a difference, but we will never know.
So it is natural to say, “so if you are so smart, Avram, what should Intel do”? I would probably suggest doing the opposite of what Pat is doing. I would retrench for a while. I would build up my cash reserves and use that to invest in new applications of semiconductors. I might even see if I could take the company private to not deal with the stock markets quarterly judgments. I would think more and do less.
If you are interested in reading about my 15-year experience and intel, please read my book, The Flight of a Wild Duck.
One good thing that come of all this is that the U.S. and other countries reduce their reliance on China and China influenced countries for semiconductors. Or do I not understand who controls the bulk of semiconductor manufacturing?
China has little influence at the moment but things could change if they took over Taiwan. Korea is independent. But both Taiwan Semi and Samsung will be doing manufacturing in the USA. USA government policy may be helpful to Intel but not helpful enough.
I love the wisdom of this “I would think more and do less.”
Who the hell is Avram Miller? And why does he not think of the 4004 and the 8008 as being the pivotal points in the last 100 years of computing? Sure — they were mostly from Federico Faggin, but he was paid by Intel at the time. So, surely, Intel isn’t a “cloud” company? Anyways — let’s all move forward to RISC-V — argument is moot.
Roy, thanks for your comments. With all due respect, I am not sure I understand them. For sure the development of the first commercial microprocessor was a pivotal event for the computer industry. and for the world. Can’t figure out how the that or the rest of your comments relate to my post.