Opinion In the dystopian world of Blade Runner, the killing of rogue superhuman replicants wasn't called that. Rather, they were "retired." Pat Gelsinger is far more human than many of the current crop of cyborg-lite tech execs, and the Intel board is no Harrison Ford, but they "retired" him just the same.
When a CEO leaves their company with no notice, no successor, no transition plan and no public statement beyond "he retired yesterday," some might see that as an assassination.
It was a coup heard around the world. Pat Gelsinger was widely seen as the "if he can't do it, nobody can" turnaround CEO on his appointment in early 2021. He'd previously joined Intel in 1979 aged just 18, where he was hand-reared by the legendary CEO Andy Grove. Lead architect of the 80486 processor while still in his 20s, Gelsinger became Intel's Chief Technology Officer in 2001. In the eyes of many, the job title was superfluous. He wasn't Intel's anything. He was Intel incarnate.
The board disagreed, and passed him over for CEO in 2005 for Paul Ottelini, the first non-engineer to lead the company. With nowhere to go inside Intel, Gelsinger made his surprise move to EMC in 2009, then taking the reins at VNWare, where he proved a real, rather than a virtual, leader. If only that had been true of Intel's subsequent choices.
Then again, like many successful leaders of market sectors they'd created, Intel in the 2010s knew it had just one trick, a strategy of keeping ahead in x86 through the tick-tock of updating process and processor design alternately. If either failed, if the competition caught up, or if the x86 market cooled, there would be trouble. It needed to innovate before one of those happened.
It would be cruel to list Intel's failed innovations, so here goes. Itanium: car crash. Atom, only not a car crash because it didn't run fast enough to dent anything. GPUs: oops. AI: should have worked harder on those GPUs. Optane: why? Wireless, networking, SSDs: worked OK, went nowhere. Silicon photonics and LCD microdisplays: never left the lab.
Even within the mainstream processor lines that fed the Intel giant, there were close calls and mistakes. The Itanium was going to be the high-end architecture, so there was no need for 64 bit extensions to x86, no matter how hard Microsoft asked. AMD disagreed, forcing Intel to grumpily adopt its standard.
Atom was going to be the portable processor of choice, so let's flog off the Arm license. Ever-increasing clock speeds would make up for lack of efficiency, until Pentium 4 and the NetBurst architecture needed so much power for so little advantage that it threatened carbon dioxide as king for global warming. That last mistake was saved by Intel's labs in Israel, who produce the Core design that ran cool and fast.
Core started life as the Pentium-M, a laptop chip that didn't fit into the macho performance model and was thus mostly ignored by the rest of the company. This lack of enthusiasm did not go unnoticed: having rejected the Atom for Arm in its iPhone, Apple didn't see Intel changing its spots and started talking to TSMC about the decade-long project that would lead to its complete de-Intelization.
The trouble with being Mr Intel is that those failures as well as the glory days are part of you. Intel's fears have been realized. Once a leader in design and process, it is now being outpaced by the Arm tribe in almost every sector, let alone by Nvidia in AI, while simultaneously not winning against TSMC in fab. Gelsinger's vision, of revitalizing the company by massive investment in fabs to regain competitiveness and be a foundry business for others, all while rebuilding its design focus, failed to deliver the fast wins that were wanted.
Nor could it. The stuff that came out of Intel over the past four years was mostly the result of the pre-Pat years. His big idea, of an Intel capable of competing and leading in an ecosystem where diversity and rapid iteration have replaced code compatibility and homogenous compute, may have worked in the long term. In the short term, with Intel's share price halving in 12 months and so much uncertainty in the global future of the tech industry, it was a very bad fit for investors. That New Intel would be carrying the legacy of the mistakes of Old Intel for too long for comfort.
It may be a mercy "retirement" for Gelsinger. All of the paths forward for Intel will see the end of anything like the company he gave so much of his life to. Dismemberment into separate foundry and design companies, partnerships on fire sale terms with cash-rich competitors, or just selling off the bits, all seem to be more in keeping with the market zeitgeist. Plus, negotiating the future of the Chips Act billions will mean its new leaders need a lot of contact with the incoming Trump administration. Gelsinger prizes vision, integrity, competence, humility, and a compassionate Christian morality towards others. Code incompatibility isn't limited to CPUs.
We'll know more about Intel's future when it gets a new CEO. Whoever it is will probably see their task as finding a rapid financial fix through deals, generally deadly to engineering companies.
As for Pat, "retired" at 63, there will be no shortage of offers of board memberships and high profile tech execing. Chances are, this won't be how he will wish to use his talents and experience. Watching how his future unfolds may be much better for us than watching Intel's. ®