3 ms·

I think rockets are insanely simple compared to silicon. The amount of research and development it takes to build a light source (hot tin plasma[1]) is extraord

I think rockets are insanely simple compared to silicon. The amount of research and development it takes to build a light source (hot tin plasma[1]) is extraordinary, and fabs are also probably the most complicated manufacturing facilities in the world. Tesla struggled mightily with integrating disparate auto parts manufacturers some years ago. SpaceX and Tesla benefited from being helmed by a crazy person (in their nascent stages), who pushed to break norms in the conventional thinking, either "rockets can't be reused/we must spend $10B on a test campaign[2] before any part leaves the ground" and "EVs aren't cool". I don't think if Musk could design a rocket engine from scratch is relevant, but the strategic design patterns of 1) reduce requirements, 2) remove unused things, 3) simplify/optimize, 4) accelerate cycle times, 5) automate. Those points aren't revolutionary, just a more expanded "go fast and break things." The computers that came out of Silicon Valley in the late 70s-into the 80s were a disruption to the old stalwarts like IBM. Though for silicon maybe I'm just trapped in that pre-SpaceX thinking. [1]: https://phys.org/news/2020-05-exceptional-euv-hot-tin-plasma.html https://phys.org/news/2020-05-exceptional-euv-hot-tin-plasma... [2]: https://www.planetary.org/space-policy/cost-of-sls-and-orion https://www.planetary.org/space-policy/cost-of-sls-and-orion

kazen44 2y ago
> The computers that came out of Silicon Valley in the late 70s-into the 80s were a disruption to the old stalwarts like IBM. Though for silicon maybe I'm just trapped in that pre-SpaceX thinking. another good example of this is how VLSI in particular disrupted the mini-computer/mainframe market and made CPU's cheap enough to put them into smaller and far cheaper machines. This really shook up the old guys who where used to being total system vendors. (IBM, DEC etc). Suddenly, you could get a computer for much cheaper compared to the decade prior, and by the early to mid nineties you had guys like sun eating their lunch completely. IBM pivoted to being a services company and DEC just imploded.
- nunez 2y ago
  AFAIK IBM still completely dominates mainframe, so they're not entirely out of the systems space.
namibj 2y ago
There's another light source that isn't nearly as hard and actually delivers better quality light at far higher electrical efficiency... It's just large and most of the cost is in the large machine that's kept under vacuum. GloFo was working on doing EUV with that (they gave up on EUV), and Japan was or is recently working on a small demonstrator fab with iirc about 10 steppers, starting with the light source machine. That machine btw is called "free electron laser", and utilizes relativistic electrons to match their velocity to free-space photons instead of the slowed-down-by-metal-helix traveling-wave-tube-amplifiers that power most TV satellites. Those work at about 10 GHz or 30mm wavelength, much larger than the 13nm band used for EUV lithography. FELs can have efficiencies in the double digit % if built with energy recovery, btw.

kazen44 2y ago

> The computers that came out of Silicon Valley in the late 70s-into the 80s were a disruption to the old stalwarts like IBM. Though for silicon maybe I'm just trapped in that pre-SpaceX thinking. another good example of this is how VLSI in particular disrupted the mini-computer/mainframe market and made CPU's cheap enough to put them into smaller and far cheaper machines. This really shook up the old guys who where used to being total system vendors. (IBM, DEC etc). Suddenly, you could get a computer for much cheaper compared to the decade prior, and by the early to mid nineties you had guys like sun eating their lunch completely. IBM pivoted to being a services company and DEC just imploded.

nunez 2y ago
AFAIK IBM still completely dominates mainframe, so they're not entirely out of the systems space.

namibj 2y ago

There's another light source that isn't nearly as hard and actually delivers better quality light at far higher electrical efficiency... It's just large and most of the cost is in the large machine that's kept under vacuum. GloFo was working on doing EUV with that (they gave up on EUV), and Japan was or is recently working on a small demonstrator fab with iirc about 10 steppers, starting with the light source machine. That machine btw is called "free electron laser", and utilizes relativistic electrons to match their velocity to free-space photons instead of the slowed-down-by-metal-helix traveling-wave-tube-amplifiers that power most TV satellites. Those work at about 10 GHz or 30mm wavelength, much larger than the 13nm band used for EUV lithography. FELs can have efficiencies in the double digit % if built with energy recovery, btw.