November 12, 2025
From IPO to Oh‑No
What happened to Transmeta, the last big dotcom IPO
The Linus Torvalds startup that wowed Wall Street — then ghosted the market
TLDR: Transmeta, the last big dotcom-era IPO, gambled on a software-driven CPU and couldn’t keep up with Intel and AMD. Commenters argue whether it was aimed at desktops or dotcoms, debate links to Apple’s Rosetta, and ultimately hail it as a bold misfire that still influenced modern chip ideas.
Transmeta, the last swaggering IPO of the dotcom era, has the tech commentariat split between nostalgia and snark. One chorus says the company was mostly famous for hiring Linus Torvalds, then vanishing into the silicon mist. Another camp argues the real star was its bold “Code Morphing” trick—software that translated everyday PC instructions on the fly—an idea some claim echoed in Apple’s Rosetta and Nvidia’s Denver CPU. Cue debate over whether Transmeta was a misunderstood pioneer or just a shiny IPO with a slow chip.
The drama heats up when folks challenge the story itself: “Selling to dotcoms? Nah,” say skeptics, insisting Transmeta aimed at desktops, not web startups. Others pile on the performance tea: Crusoe and Efficeon promised magic, but ran slower than Intel and AMD, turning “Code Morphing” into “Code Mourning.” Commenters also zoom out to the macro vibes—post-9/11 investor chills, fewer IPOs, shrinking deal sizes—and frame Transmeta as a perfect storm of hype, timing, and tough physics. One thoughtful voice applauds the moonshot, noting that even if the bet lost, lessons fed into modern chip and GPU designs. Between the “IPO to RIP‑O” jokes and hot takes, the crowd lands on a spicy truth: Transmeta was both ahead of its time—and out of time.
Key Points
- •Transmeta raised $273 million in its Nov 7, 2000 IPO, often cited as the last major dotcom-era tech IPO.
- •Despite the narrative, PayPal completed a notable $70.2 million IPO in Feb 2002, and IPO activity fell sharply post-9/11.
- •Transmeta pursued x86 compatibility via a software translation layer, contrasting with AMD’s hardware-based translation.
- •The Crusoe CPU delivered Pentium III-like performance but about 30% less efficiently; IBM manufactured it.
- •Efficieon (2004) peaked at 1.7 GHz versus Pentium 4’s 2.4 GHz and struggled amid the industry shift to 64-bit and AMD’s Athlon 64; manufactured by TSMC and Fujitsu.