What happened
The New York Times reported in 2024, in a piece carried by The Straits Times, that in 2005 Intel CEO Paul Otellini presented his board with an idea to buy Nvidia for as much as US$20 billion. Some executives believed graphics-chip designs could take on new jobs in data centres. The board resisted: Intel had a poor record of absorbing companies, and the deal would have been its biggest ever. Otellini backed away. One attendee called it 'a fateful moment.'
Intel then backed an in-house graphics project called Larrabee, led by Pat Gelsinger. It consumed four years and hundreds of millions of dollars. In 2008 Gelsinger predicted that 'today's graphics architectures are coming to an end.' Larrabee fell behind schedule and its graphics performance lagged. Intel cancelled it in 2009. Gelsinger still argued in 2024 that more patience and investment might have changed the outcome.
The Times, citing interviews with more than two dozen former managers, directors and analysts, described a pattern: AI projects were started, pursued for years, then shut down when leadership lost patience or the technology fell short. Investment in new chip designs took a back seat to protecting the x86 franchise.
Why it happened
The board declined the Nvidia deal because of Intel's poor record absorbing acquisitions and its size.
Intel put its own hybrid graphics-and-x86 design at the core of Larrabee, which did not work.
Per the Times, the profits from x86 were so large that Intel did not really shift course.
The lesson
Protecting the profitable core while skipping or abandoning bets on the next architecture can cost a market leader the next era. Intel dropped its AI projects when patience ran out.
Aftermath
By October 2024 Nvidia's market value was over US$3 trillion, roughly 30 times Intel's, which had fallen below US$100 billion. Intel's revenue fell more than 30% from 2021 through 2023 and it had recently cut 15,000 jobs. Some big tech companies and bankers were reportedly considering Intel as an acquisition target.
FOLLOW THE EVIDENCE
The sources
- How Intel got left behind in the AI chip boom straitstimes.com