Jim Keller's startup wants to mass-produce chip fabs from Texas — a semiconductor expert calls it smoke and mirrors
Atomic Semi rebrands as Fab2 and moves to Texas with a wild pitch: a factory that builds small semiconductor factories. In the thread, a self-described semiconductor expert reads the same news and sees another wunderkind startup that dissolves on contact with reality.
Fab2 now runs three sites — a 120,000 sq ft Austin HQ, a 30,000 sq ft Lockhart "fab fab," and the original SF garage fab — chasing the idea of mass-producing small chip fabs rather than chips, co-founded by Jim Keller and DIY-fab pioneer Sam Zeloof.
A semiconductor insider is unimpressed: no fundamental manufacturing advance, likely electron-beam lithography that suits prototypes, not volume — and why not just verify in simulation?
The pitch: a factory that makes factories
Tom's Hardware reports that Atomic Semi — co-founded by legendary chip architect Jim Keller and DIY-fab pioneer Sam Zeloof — has rebranded as Fab2 and shifted operations to Texas after about four years in California. The concept is deliberately recursive: not a fab that makes chips, but a "fab fab" meant to mass-produce small semiconductor fabrication plants. Fab2 now spans three sites — a 120,000 sq ft Austin HQ for research and production, a 30,000 sq ft Lockhart building housing the fab-fab itself, and the original 25,000 sq ft "garage fab" still in San Francisco.
It's the kind of hard-tech swing HN loves. The thread supplied the cold water, from someone claiming the relevant credentials.
The insider who isn't buying it
"I'm a semiconductor expert and without seeing the details, this seems like yet another smoke and mirrors wunderkind startup that will disperse once practical considerations are made. As far as I know, Mr. Zeloof has not made any fundamental advancements in any steps in manufacturing." — a self-identified semiconductor-industry expert
A second commenter got specific about the likely process — and why it may not bridge to real production:
"The article … doesn't explicitly state, that they're going to be using electron beam lithography. Makes sense for their low volume … you might as well just do verification in simulation rather than sidetrack through Fab2." — a commenter familiar with lithography and process-porting
And the plain governance question the announcement invites:
"I don't get it. How is Jim Keller running a brand new, hard tech startup while being CEO of Tenstorrent at the same time?" — a commenter on the founder's dual role
The bet, stated honestly
Fab2's premise is that the bottleneck in chips isn't the wafer, it's the number of fabs — so make fabs cheap and plentiful. It's a genuinely different theory of the industry. The thread's expert isn't attacking the ambition; he's asking for the one thing a rebrand and three buildings don't provide: a single fundamental manufacturing advance that makes a small e-beam fab worth building instead of simulating. Until that shows up, "fab fab" is a thesis, not a factory.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
He earned a first-class CS degree on Coursera for £17k — a PhD holder in the thread calls his own 11 years a waste
Lex Toumbourou finished a Goldsmiths BSc by distance, through Coursera, in under four years for about £17,000. The comments split hard: a Bachelor-Master-PhD holder calls university a social club, while hiring managers insist the degree is the filter that keeps your résumé alive.