Intel

Intel is the textbook firm that beat the innovator's dilemma - it gutted its own margins with the Celeron in 1998 and won. Then it lost mobile. The popular lesson is that it stopped being paranoid. The real lesson is that the second test wasn't the same shape as the first.

Intel is remembered for missing mobile out of blindness. It wasn't blind - it sold its ARM business for $600M to protect fat x86 margins, exactly as the textbook says you should. Then the forecast it trusted was off by 100x.

In 2017 Intel could have bought a 15% stake in OpenAI for $1 billion. It walked. The popular version blames one CFO's AI skepticism - but the deal had a second killer: Intel's own data-center unit refused to build chips at cost.

Everyone calls it an $8.5 billion grant. The final award was $7.86 billion, paid out only as auditors verify construction — and Intel had drawn just $2.2 billion against an $18.8 billion net loss and a CEO shown the door. This isn't a rescue. It's a probation.

Intel's IDM 2.0 was sold as a bold pivot. It was a rescue plan — and a late one. The same overreach that buried 10nm in 2015 and 7nm in 2020 came back for 18A, the very node meant to save the company.

Intel paid Dell to shun AMD, and the money grew from 10% of Dell's operating income in 2003 to 76% in early 2007. Everyone remembers regulators punishing the scheme. The €1.06B EU fine was thrown out entirely, and Dell's $100M penalty was for lying about the money — not for taking it.

Intel called it a retirement. The board called it a binary: retire or be removed. Pat Gelsinger's foundry bet was bleeding $5.84 billion in a single quarter — and the smartest chip operator on the board had already walked out in protest months earlier.

In August 2000 Intel was worth $509 billion - the most valuable company on earth. By December 2024 it was worth $104 billion, behind even AMD and ARM. The iPhone story everyone tells is the wrong one. The real culprit was the model that once made Intel unbeatable.

In 2024 Intel's revenue fell just 2%, to $53.1 billion. Yet its foundry lost $13 billion, it cut the dividend and 15% of its workforce, and the board forced out the CEO it had hired to save it. The top line is the autopsy's red herring.

IDM 2.0 was sold as a trillion-dollar foundry comeback. The real numbers: roughly $95 billion of capex over four years, a $13 billion foundry loss in 2024, a CHIPS award quietly cut to $7.86 billion, and a CEO the board pushed out.