After Intel CEO Swan’s cleanup job, Gelsinger will move the company forward

Bob Swan didn’t fail, he succeeded in righting the ship, but Intel has tapped Pat Gelsinger, who started working there right out of high school, as the visionary to move the company forward.

VMware CEO Pat Gelsinger
Daniel Masaoka

Pat Gelsinger’s return to Intel after a 12-year absence has been greeted positively, with the stock jumping 8% on the news Wednesday, analysts lauding it, and apparently even Intel staff approving, indicating he remained popular there despite leaving the firm in 2009.

Replacing outgoing CEO Bob Swan with VMware CEO Gelsinger, isn’t a sign of failure on the part of Swan, who took over in early 2018. Intel is expected to meet or exceed Q1 revenue and income projections when it reports earnings Jan. 21. The fact that Swan is being given time to clean out his desk—he is staying on until mid-February—says that this is a civil parting, unlike that of his predecessor, Brian Krzanich, whom they couldn’t get out the door fast enough.

Swan’s big failure was the manufacturing process, which was hardly his fault. Intel has struggled mightily to get its fabrication-node size down from 14nm to 10nm in order to improve power efficiency, performance, and price. Meanwhile, foundry giant TSMC is making 7nm processors for AMD, and AMD is eating Intel’s lunch in certain quarters with its highly competitive new chips.

But Swan did a lot of things right. He sold off Intel’s 5G radio business and its NAND business, which were distractions. He acquired Habana, Moovit, and Bearfoot Networks, all closer aligned with Intel’s core competency, and started the XPU project. That’s a solution where multiple chips—CPU, GPU, FPGA, AI—are all tied together through a single API, and whichever chip is best suited to process a task is assigned the job.

Not bad for a finance guy. As the late Intel CEO Paul Otellini—who had an MBA, not an engineering degree—showed us, Intel doesn’t necessarily need an engineer at the helm.

Gelsinger brings vision, something the CEO has to have. (Jen-Hsun Huang at Nvidia has it. Lisa Su at AMD has it. Marc Benioff of Salesforce has it in spades. Tim Cook doesn’t. That’s why Apple has done nothing but iterate and incrementally advance existing products under his tenure.)

While Swan showed it with the XPU strategy, he was more of an operations guy. His task was to fix a company broken by Krzanich’s reign of error, and he did that. He stemmed the talent bleed and fixed morale. He made it so Intel delivered products on time, unhampered by shortages as severe as those affecting AMD. Analyst Rob Enderle put it very well when he told me Swan is the mechanic but Gelsinger is the driver.

Swan’s departure does continue a problem for Intel: its last three CEOs have all walked the plank. Otellini was pushed out for failing to see the mobile market coming and letting Arm eat Intel’s lunch. Krzanich was a nightmare, and now Swan is (politely) shown the door after two years. Hopefully Gelsinger can break that trend, too.

With Intel From the Start

Gelsinger joined Intel right out of high school in 1979. He lacked a degree, so Intel said it would cover his tuition so long as he maintained a B average. While working on the 80286 processor, Gelsinger earned a B.S. in electrical engineering. He got his master’s degree in EE from Stanford while working on the 80386, a product that saved the company.

At the time, most of Intel’s top talent worked on a project to replace the x86 architecture called iAXP432. It was very ambitious and tried to do many things modern CPUs do, such as object-oriented memory and capabilities, garbage collection, multitasking, and interprocess communication. But the chip was a gigantic failure.

Meanwhile, Intel had charged Gelsinger and another engineer, John Crawford, with trying to figure out how to keep the 286 alive as an interim product. They came up with the 80386 and saved Intel’s skin. It’s a reflection of the times that two engineers in their 20s  could alone come up with the 386. New architectures today require thousands of engineers.

At the age of 25, Gelsinger was given the reigns of the 80486 project by Andy Grove as a way to keep him from quitting the company, since Gelsinger wanted to go to Stanford full time to earn his PhD in electrical engineering.

He would go on to become Intel’s first CTO, where divining the future of tech was his job. He created the Intel Developer Forum (IDF, axed under Krzanich), and when AMD was making advances in 64-bit x86 and creating dual-core CPUs, Intel was dismissing it and saying Itanium was the way of the future. Meanwhile, Gelsinger was practically jumping up and down telling his bosses to pay attention to AMD, that they were on to something. He was right, and for a few years AMD was a major competitor to Intel, although nothing like they are now.

Gelsinger was pushed out in 2009 after he was blamed for the failure of Larrabee, an Intel effort to create a GPU that every analyst said was doomed to failure and not his fault. He did a three year stint as COO of EMC before taking over VMware in 2012.

In that time he fought off the Microsoft virtualization threat with Hyper-V, nearly tripled revenue to $12 billion, and oversaw more than 30 acquisitions. The company expanded from hypervisors into networking, cloud, security, containers, and 5G. Glassdoor voted him CEO of the Year in 2019 and by all accounts VMware employees love him.

If there is bad news for anyone it’s VMware. Last year it lost COO Rajiv Ramaswami to become CEO of Nutanix, while senior vice president of cloud management Ajay Singh left for Pure Storage. So VMware is seriously deficient in the C-suite at a time when it was hoping to spin off from its parent company Dell Technologies.

In the mean time I wouldn’t expect any radical news out of Intel for a while, except maybe a return of the IDF. Gelsinger has a lot of catching up to do.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2021 IDG Communications, Inc.

IT Salary Survey 2021: The results are in