Gelsinger’s Intel Goes Global

by Jean-Louis Gassée

The US semiconductor manufacturing industry, which once reigned supreme, has lost a step to companies in Taiwan and South Korea. Intel’s new and rehired CEO might have plans to make the US a prime semiconductor manufacturing competitor - again.

On March 23rd, Intel held an on-line event titled Engineering the Future in which CEO Pat Gelsinger laid out his vision for the company (see my take in the April 11 Monday Note, Intel 2.0 Reboot). Among other lofty goals, Gelsinger promised a return to Intel’s more competitive days when the Santa Clara company reigned supreme in the world of microprocessor chips that run in the heart of Windows personal computers and, a little later, servers:

“We’re bringing back the execution discipline of Intel. I call it the Grovian culture that we do what we say we will do. That we have that confidence in our execution. That our teams are fired up. That we said we’re going to do x, we’re going to 1.1x, every time that we make a commitment. That’s the Intel culture that we are bringing back.”

We’ll note Gelsinger’s castigation of Intel’s recent performance, the implication that company has gone soft. He openly yearns for what one can’t help call the hard-ass custom that prevailed when Andy Grove, the author of Only The Paranoid Survive, was in charge. Grove’s book subtitle, “Exploit the Crisis Points That Challenge Every Company”, might well be a guiding motto for his younger disciple. (See also The History and Culture Gelsinger Inherits.)

In his March 23rd message, Gelsinger told the world that Intel would dedicate $20B to two new chip foundries in Arizona, the first components of an independent division called Intel Foundry Services. The IFS foundries are tasked with fabricating chips for any architectures and companies — including Intel’s competitors — a 180-degree turn for a company that once turned down the opportunity to fabricate the ARM chip that powered the first iPhone.

Gelsinger also told us that Intel would abandon its outdated Immersion Lithography (IL) fabrication method and commit itself to the Extreme Ultraviolet (EUV) process favored by Taiwan Semiconductor Manufacturing Company (TSMC, now the world’s largest chip maker). Not only had the IL process caused repeated delays in the production of competitive x86 chips, it was probably the reason why Intel was unable to regain Apple as a customer after the magnitude of the original iPhone sin became obvious.

Healthy and frank as this aggiornamento is, the modernization entails more than just building a new manufacturing unit — itself a lengthy and costly endeavor — it also requires adopting semiconductor design rules to fit the EUV process. The net result is that new EUV chips competitive with TSMC output won’t come out until the end of 2022 or early 2023. Intel 2.0 is in motion but will need time to produce results — which is why the new CEO made references to contracting some of its CPU chips to “other sources”, most likely TSMC, something unthinkable under his predecessors.

Four months later, we hear more about Intel’s plans. According to a July 16th Wall Street Journal article (behind its paywall), Pat Gelsinger is considering buying semiconductor fabrication company GlobalFoundries for $30B. This, on top of the $20B committed to Intel Foundry Services, would raise Intel’s bet on its manufacturing future to $50B.

As of this writing, the GlobalFoundries rumor hasn’t been confirmed by Intel or any other reliable sources. The 2.1M hits Google reports when asked to search for “Intel buys GlobalFoundries” appear to be based on the sole WSJ story! But even if it’s true, the deal could be torpedoed by antitrust considerations unless geopolitics prevail in a context where China insists that Taiwan is a Chinese province.

Moving past that uncertainty, there’s a curiosity — almost an irony — to Intel buying GlobalFoundries. The company was formed in 2009 as a spinoff from AMD when the latter company realized it could no longer compete by fabricating its own CPU chips. The Mubadala Investment Group (an Emirati state-owned organization) bought the spinoff, and gave it its immodest name. Having discarded GlobalFoundries because its technology had become obsolete, AMD then adopted TMSC as its fabricator. And now Intel is rumored to be interested in this company that offers less than state-of-the-art chip production.

There’s no doubting Pat Gelsinger’s brain power, so there must be a plan behind the purported acquisition. Perhaps Intel wants to merge GlobalFoundries into Intel Foundry Services in order to create a full-range fabrication giant: New ISF chips for top-of-the line products and GlobalFoundries technology for less demanding applications such as automotive, aerospace, industrial, telecommunications, home automation, entertainment, retail, office automation, and so on… Smartphones and wearables might be added to the list sometime later.

In this plan, Intel 2.0 would make a powerful complement to the design prowess of companies such as Nvidia, with its GPU and Machine Learning chips and, of course, Apple and its home-designed Apple Silicon CPUs powering its phones, tablets, personal computers, and wearables.

All of this speculation could justifiably be called the product of a fertile imagination…but whose imagination, mine or Pat Gelsinger’s? Intel’s rehired chief (he was once the company’s first CTO) is an extraordinary individual with a combination of intellect, and mental and spiritual energy that makes him stand out among his peers. For more on this, I promise you won’t waste your time if you read the second half of my April 11 Monday Note that starts with [one addition mine]:

“Raised on a farm in the heart of Pennsylvania Amish country, Gelsinger earned an associate degree from Lincoln Tech in New Jersey when he was 18 and immediately began work as a quality-control technician at Intel, a company he would stay with for 30 years [and leave for 10 years]. While holding his Intel job, he earned a bachelor’s degree in electrical engineering from Santa Clara University, followed by a master’s degree from Stanford.”

Under Gelsinger, Intel 2.0 could leave behind its years of disappointing production quality, slough off the old Wintel portmanteau like an ophidian shedding its skin, and vie with TSMC for the top position in the general semiconductor fabrication space. Intel could make US Semiconductor Manufacturing Great Again.

— JLG@mondaynote.com