Gelsinger Didn't Lose the Race. He Picked a Strategy That Could Never Pay for Itself.

On November 26, 2024, Intel stood next to the federal government and finalized a $7.86 billion CHIPS Act award to help build fabs across Arizona, New Mexico, Ohio, and Oregon.⁸ Five days later, the man who had bet his company on exactly those fabs was gone. Pat Gelsinger's employment terminated by resignation on December 1.⁵ The timing was almost cruel: the subsidy that was supposed to anchor the strategy arrived the same week the board decided the strategy's architect would not survive to spend it.

The official word was that Gelsinger had 'retired.'⁶ He had not. Reuters, citing a person familiar with the matter, reported the board gave him a choice - retire or be removed - because directors had concluded the plan was not working and progress was not fast enough.⁷ 'Retired' is the word you reach for when the real one is too blunt to print — Intel's press release used it; the board's reported ultimatum did not. The interesting question is not whether he was pushed. It is why a man who had spent thirty years at Intel, returned as its savior, and executed his plan with conviction was nonetheless doomed from the day he took the job.

The two businesses that could not afford each other

Gelsinger took over as CEO in February 2021 with a clear thesis: Intel would stop being just a chip designer and become a contract manufacturer too - a foundry that built chips for other companies, the way TSMC does.¹ Strategically the logic was sound. Foundry is where the durable money in semiconductors increasingly lives, and the United States wanted a domestic one badly enough to write checks for it. But a foundry is the most capital-hungry business on earth. You buy the most expensive machines ever built, you build clean rooms by the acre, and you bleed cash for years before the volume arrives to make any of it pay. The only way that math works is scale - the kind of scale that lets you spread tens of billions in fixed cost across an ocean of wafers. TSMC already had that scale. Intel was proposing to acquire it from a standing start, with money it would have to generate itself.

And here is where the plan met its arithmetic. The business meant to fund all this - Intel's traditional products - was not growing. It was falling off a cliff. Revenue peaked in Gelsinger's first full year at $79.0 billion in 2021, then slid to $63.1 billion in 2022, $54.2 billion in 2023, and $53.1 billion in 2024 - a roughly 33% collapse over three years.²³⁴⁹ Operating income told the same story faster: from $19.5 billion in 2021 to a near-flat $93 million by 2023.²⁴ He was asking a patient who was hemorrhaging to donate blood for an organ transplant.

Look at the bottom two rows of that table. They are the whole story. The business Gelsinger was 'betting the company on' was generating under a billion dollars a year - $952 million in 2023 - and losing money on every dollar of it, a $482 million operating loss that year.⁴ That is not a pivot to foundry. It is a foundry that barely exists, attached to a product engine that was throwing off less and less of the cash needed to build it. By 2023 even the data center segment - the crown jewel of Intel's old empire - was posting an operating loss of $530 million.⁴ There was no longer a healthy core to subsidize the bet.

Why this was a reckoning, not a stumble

The convenient story is that Gelsinger had the right plan and simply moved too slowly - that with another two or three years, the fabs would have filled and the math would have turned. The board itself reportedly framed its loss of confidence in terms of speed: the plan was not working, progress was not fast enough.⁷ But speed was never the binding constraint. Money was. A turnaround has a clock, and the clock is set by how long your existing business can keep paying for the future you are building. When the existing business shrinks a third in three years and the future you are building loses money the entire time, you do not have a pacing problem. You have a self-financing problem - and no amount of urgency fixes a plan that the company cannot afford to complete on its own balance sheet.

This is also why the CHIPS Act money, welcome as it was, could not rescue the structure. The finalized $7.86 billion was real, but it was a fraction of what a competitive foundry costs to stand up - and it came with strings, having already been trimmed from the preliminary 'up to $8.5 billion' figure — senior administration officials stated the reduction was because the separate $3 billion DoD Secure Enclave contract drew from the same CHIPS Act pool.⁸¹⁰ Government subsidy can lower the height of the wall. It cannot change the fact that you are still climbing it on income that is falling. The grant arrived; the runway did not lengthen enough to matter.

The fairest case for Gelsinger

The honest counter is that Gelsinger inherited a catastrophe, not a healthy company to wreck. The manufacturing lead Intel lost happened on his predecessors' watch; he walked into a process-technology hole that took years to dig and could not be filled in a single tenure. By that reading, the revenue collapse was a lagging consequence of decisions made before 2021, and asking any CEO to reverse it inside three years - while simultaneously raising the capital to re-enter foundry - was setting a bar no one could clear. There is truth in this. Gelsinger did not cause the rot; he diagnosed it correctly and chose the one strategy that addressed the root rather than the symptom.

But naming the root cause is not the same as choosing a curable strategy. The deeper objection to the steelman is that there may not have been a self-fundable path at all - and if so, the responsible move was to size the foundry ambition to what the shrinking core could actually carry, or to seek outside capital before the core decayed, not to assume the product business would hold while the fabs were built. Gelsinger bet that it would hold. It did not. The strategy was not undone by bad luck or slow hands; it was undone by an assumption baked into its first slide.

When Gelsinger left, interim co-CEOs and an interim chair stepped in to hold the seat.⁶ The fabs are still rising in Arizona and Ohio; the question Gelsinger could not answer is the one his successors inherit - who, exactly, pays for them while they wait to pay off. He was not the wrong man for an easy job. He was the right diagnostician for a disease whose only honest cure cost more than the patient could earn. The board did not fire a slow executor. It conceded, three years late, that a strategy which cannot fund itself does not need more time. It needs a different strategy - and a CEO willing to admit the first one was never affordable.