As a man of faith, outgoing Intel CEO Pat Gelsinger will perhaps be familiar with the verse from the book of Ecclesiastes that reads: “It is better not to promise anything than to promise something and not do it.”
On Monday it was confirmed that Gelsinger had paid the price for a perceived failure to fulfill the many promises he has made since taking charge of the venerable US chipmaker in 2021. Intel said in a statement that its CEO, who first joined the company as a teenage technician back in 1979, was retiring and leaving the business with immediate effect.
Gelsinger had given no indication he planned to leave the role he views as “the honor of his lifetime,” so it’s probably safe to assume his ‘retirement’ was not entirely voluntary - indeed multiple reports suggest he jumped before he was pushed after disagreements with the Intel’s board over the company’s future direction.
Pat Gelsinger: A man with a plan
It is a far cry from the heady days of February 2021, when Gelsinger was hailed as the returning hero after he rejoined the chipmaker as CEO following spells away at EMC and VMware. An engineer in the tradition of Intel icons Gordon Moore, Robert Noyce, and Andy Grove, he promised to return the company to its former glories with a five-year strategy he dubbed IDM 2.0 (IDM standing for integrated device manufacturing), which would see Intel manufacturing its own cutting edge chips as well as supplying components to third parties.
Gelsinger’s strategy seemed ideally suited for the moment the industry found itself in. With the world emerging from a global chip shortage, which led to widespread shortages of electronic equipment and stopped production lines in car plants for weeks at a time, governments started offering billions of dollars in subsidies to shore up semiconductor supply chains. Intel secured $7.9 billion under the US CHIPS act (Gelsinger was part of the committee that advised President Joe Biden on the make-up of the act) to help fund the build-out of new chip fabs in Arizona and Ohio, as well subsidies from the European Union under its own chips legislation as part of a $36 billion spending spree planned for the continent.
Four years on, putting these plans into action has not proved easy. Intel has so far failed to narrow the manufacturing gap on TSMC, which remains unmatched when it comes to churning out the world’s most advanced chips, and has made limited impact in the AI space, watching on while Nvidia has cleaned up. Critics say the company has taken its eye off the ball in its core PC market, and in the data center, where Intel has ruled the roost for years, things are not looking rosy, with chief rival AMD narrowing the gap between the companies and the hyperscalers building their own custom silicon based on rival firm Arm’s architecture.
Gelsinger has continually promised his shareholders that the good times were just around the corner, but things came to a head in August when Intel posted a quarterly loss of $1.6 billion and announced plans to lay off 15,000 employees. The CEO said at the time that 2024 would be the “trough” during which Intel tackled its technology deficit, but it seems this argument fell on deaf ears.
Can Intel still win the AI war?
The events of recent months will no doubt color the judgments on Gelsinger’s time at the helm of Intel, but it is worth noting that he took over a rapidly sinking ship that was already trailing TSMC and had been beset by years of delays to its manufacturing processes, with its seven and 10-nanometer process nodes arriving months later than planned.
While some colleagues have praised Gelsinger for restoring Intel’s “engineering mindset,” affecting cultural change at a business of such scale and history, in a notoriously cyclical and capital-intensive industry that itself was undergoing a rapid reshaping as the AI revolution took hold, was always likely to require time and patience.
Indeed, consider the AI market. Here, Intel may yet find success thanks to the open nature of its oneAPI developer platform, which enables code to be built that runs on any type of hardware. This contrasts with Nvidia’s proprietary Cuda platform, which is compatible only with the vendor’s GPUs.
Speaking to DCD earlier this year, Dr. Paul Calleja, director of Cambridge University’s Research Computing Services division, which provides compute power to scientists across the UK, said he believes openAPI “will eventually win out” over Cuda, adding that his department already has clients “who use Nvidia but don’t want to develop on Cuda because they don’t want to get locked in.” Intel has lost the AI battle, but could end up winning the war, particularly as the market moves from focusing on training to inference, where CPUs - Intel’s bread and butter - may play a greater role.
For now, the company will head into 2025 searching for a new CEO and with a decision to make about whether to stick with its current strategy or take a less risky path in a bid to stem its losses. But if its board is tempted to retreat to safer ground, it may want to consider the case of another US technology stalwart, IBM.
The IT giant appeared to be in terminal decline until it took radical action in 2020, completely jettisoning its historically lucrative managed services division to focus on other areas. Four years later, the company’s previously drooping share price has hit a record high.
Whether Gelsinger could have eventually turned Intel around will remain a mystery, but having never been given the opportunity to complete his grand plan, he might argue he was handed an impossible job. Those being forced to look for new roles themselves as a result of his actions are unlikely to have too much sympathy.
