Intel’s CEO Pat Gelsinger has left the building – specifically Intel’s headquarters: the Robert Noyce Building in Santa Clara, California. On December 2, Intel issued a press release stating that:

“…CEO Pat Gelsinger retired from the company after a distinguished 40-plus-year career and has stepped down from the board of directors, effective Dec. 1, 2024.”

This is not the normal way to announce the retirement of a high-tech CEO. Normally, such things are pre-announced, not announced after the fact. The announcement’s appearance a day after the fact indicates that things are not normal at Intel. They’re extremely abnormal. The announcement is pure BS, conjured up and supplied by Intel’s marketing machine, which appears to have pulled this particularly banal and vapid sentence out of its corporate bunghole.

Intel’s stock price is down by more than 60% since Gelsinger took over and Nvidia’s stock just replaced Intel’s in the Dow Jones Industrial Index. That must be a low blow to the company. Intel’s many woes are quite public, and, as a friend said to me recently, the stock price is down more than 50% and it happened on Gelsinger’s watch. Someone has to take a fall for these events, and that someone is Pat Gelsinger, even though I don’t think it’s his fault. Fair warning: I’m not an objective observer here.

I joined Intel in April 2019 and stayed for nearly 2.5 years before retiring. I worked in marketing for Intel PSG (Programmable Solutions Group), the organization formerly, and now, once again known as FPGA maker Altera. At the time I joined, Bob Swan was the interim CEO. Swan was a nice guy and a numbers guy and he took the CEO position when it was vacated by Brian Krzanich. Swan had been Intel’s CFO. He wasn’t a product or semiconductor process guy. Pat Gelsinger became Intel’s CEO in February 2021. His tenure as Intel CEO overlaps my stay at Intel by about six months, but I’d met Gelsinger back in 1999 when we shared a limousine at some Intel event. During my last half year at Intel, I saw Gelsinger begin the Herculean effort to turn the company around.

From my perspective, turning Intel around was going to be a daunting project. As a member of Intel PSG’s marketing group, I observed how everything in the company – including FPGAs – was slaved to Intel’s processor marketing. FPGAs were nothing if not connected to a host processor, specifically a host x86 processor, more specifically, an Intel x86 processor. Intel corporate marketing viewed all of Intel’s products through the lens of its PC client and server processors. However, in the real world, FPGAs are typically not slaved to host CPUs except in the narrow field of heterogeneous data center computing. That’s how Intel PSG and its FPGAs ended up in the DCAI (Data Center and AI) group. The FPGA marketing from Intel was a real Frankenstein’s monster as a result of this peculiar perspective on FPGAs. Intel AI and GPU efforts encountered the same sort of corporate bias.

I have to admit that I did not see Gelsinger properly address this hobbling sort of marketing during my final months at Intel, but I was encouraged when Intel spun out Intel PSG and restored the Altera name in February 2024. However, jettisoning Altera was just part of clearing the decks of extraneous enterprises that aren’t processors or true processor adjacents such as Gaudi AI chips or Xe GPUs. Intel remained and still remains a very CPU-centric company, because that’s where the company makes most of its money.

When Gelsinger took over the CEO slot at Intel, he had not one but two Herculean tasks to accomplish. The first such task was to recover Intel’s semiconductor processing technology leadership, which had been lost to TSMC and Samsung over the past couple of decades. When Intel started in 1968, it was the era of “Real men have fabs.” In the 1960s, all semiconductor companies had captive fabs, and success required both process acumen and semiconductor design leadership. Intel developed both in short order. There were no fabless semiconductor companies until the 1980s. The fabless hadn’t been invented until then, although even in its early days, Intel would design ASICs for customers as a means to fill the fab. That’s how and why Intel developed the first commercially successful microprocessor, the 4004. It was an ASIC designed for Busicom, a Japanese calculator company.

Today, the story is significantly different. Intel’s biggest rivals, Nvidia and AMD, are fabless semiconductor vendors. (The reality is that Nvidia is no longer a semiconductor company. It’s a systems company that also makes and sells semiconductors, however, that’s a different story.) Fabless is in, largely because TSMC’s drive and energy have made the fabless semiconductor model a winning idea. When you’re the foundry for the world’s top semiconductor makers, you’ve got to be good. As long as Intel’s TD (Technology Development) group was creating only semiconductor process nodes for Intel processors, it was not being driven by the same competitive pressures that forced TSMC to work harder and longer than its foundry competitors to crank out new process nodes. Gelsinger needed to fix that situation, and he did so by declaring that Intel would become a competitive foundry.

Gelsinger’s second Herculean task was to regain design leadership for the company’s bread and butter: x86 processors. AMD under CEO Lisa Su really stepped up its game in this arena, once the company divested itself of its fabs with the creation of GlobalFoundries. With the onset of AMD’s fabless era, it could go to the leading semiconductor foundry, TSMC, and instantly reap the benefits of the world’s most advanced semiconductor nodes without needing to expend corporate resources on developing those nodes. Meanwhile, Intel continued to ride the dual horses of semiconductor process leadership and x86 design leadership.

To keep Intel whole, Gelsinger needed to bring Intel’s technological prowess up to world class leadership in both. As part of his plan to accomplish this feat, Gelsinger took pilgrimages to various governments worldwide, seeking the funds to upgrade Intel’s fab capabilities around the world at a time when these governments had become extremely interested in investing in fab capacity and capabilities that did not reside in Taiwan, a stone’s throw from an increasingly belligerent People’s Republic of China. The timing could not have been better for Intel and Gelsinger’s plans.

However, building fabs is slow work. It takes two to four years to build a new fab and facilitize it, and Intel wasn’t going to build just one new fab, it was going to build a bunch of them, all over the world. Meanwhile, a recent comment by interim co-CEO David Zinsner suggests that Intel’s board of directors was not happy with Intel’s progress on the product side under Gelsinger. Intel’s appointment of Michelle Johnston Holthaus as co-CEO and CEO of Intel Product would tend to confirm that conclusion. Intel’s latest offerings have failed to ignite the kind of excitement and sales that Intel shareholders want and expect, and the board of directors must certainly want more attention paid to that side of the company. As recent events have shown, the stock market doesn’t give today’s public corporations years to turn things around. It punished Intel’s stock price because of a lack of perceived progress on both the product and process sides.

Intel’s December 2 announcement of Gelsinger’s retirement is a sad end to Gelsinger’s reign as Intel CEO, a position he’d wanted for at least a quarter of a century. Frankly, I don’t know where Intel will find a CEO who can do better than Gelsinger, who can turn the immense ship of Intel around fast enough to suit the shareholders. One name being floated is Lip-Bu Tan. I spent 18 months working for Cadence while Lip-Bu was CEO, and he’s an excellent choice. He’s the best-connected CEO in the business, able to call a wide range of semiconductor and systems corporate CEOs on speed dial from one of the two mobile phones on his belt. He’s not a process or product guy, but he’s certainly connected.

Other candidate names have also been floated in the past few days, but none of them bleeds Intel blue like Pat Gelsinger. My own pick might be Steve Sanghi, who spent 30 years transforming Microchip into a semiconductor powerhouse, starting with nothing but the ashes of General Instrument Microelectronics. However, the semi-retired Sanghi just retook the CEO position at Microchip after the company had some disastrous quarters. He’s up to the task, but he might not want to take on the burden just at this moment, for more than a few reasons. Coincidentally, late last week and a day or two after I wrote this paragraph, Intel announced that Sanghi had joined Intel’s Board of Directors.

One can speculate on the reason for Gelsinger’s sudden “retirement.” I’m not a betting man, but I’d put the full buck I’m willing to bet on a fight between Gelsinger and the Intel board of directors on the question of splitting Intel into a semiconductor foundry, to compete directly with TSMC, and a product company, to compete with AMD and Nvidia. Such a move is often perceived as a way to increase shareholder value. As a technology historian, I can say that cutting up a company for its components often leads to disaster, starting with the inevitable talent exodus as competitors recruit the best people from the ailing company’s talent pool. The one successful split I can think of is AMD. My poster children for the disastrous sort of outcome include:

HP – which splintered into HP Inc (PCs and printers), HP Enterprise (servers and corporate services), Agilent (life sciences), and Keysight (electronic instrumentation)

Motorola – which was split into a semiconductor company that withered on the vine and was snapped up by NXP, Motorola Mobility – which consisted mostly of the mobile phone and cable product groups, and Motorola Solutions – which produced equipment for governments and corporate enterprises.

Gelsinger left the building on December 1. Intel has quickly named interim leadership while it seeks a new, permanent CEO to take over a very troubled company that has long neglected the self-care it sorely needed to stay competitive. As an Intel alum and retiree, I wish Intel and Gelsinger well. As a technology historian, I will watch history unfold.