Substrate, a San Francisco startup, raised $100 million to develop a potential rival to #EUV photolithography, in which the Dutch juggernaut ASML has a monopoly valued at 360 billion euro. To this day, ASML's #EUV machine is the only tool capable of producing highly advanced #AI chips at scale. The U.S. dropped the ball on #lithography long ago, but #Substrate sees an opportunity and even dreams of building its own fabs to replace TSMC. The goal: for "America to regain its rightful place as the leader in semiconductor production". For its high ambitions, Substrate is light on detail. It uses X-ray lithography with a wavelength of less than 4 nanometers. That allows it to print finer details on a chip, even smaller than EUV's 13.5 nm wavelength. It's not the first time X-ray lithography has emerged as an option. The semiconductor industry (especially IBM and Philips) researched it in the eighties and early nineties, when DUV, with a 193 nm wavelength, was about to run out of steam. All players, including #ASML, dismissed X-ray lithography because it requires very expensive mask materials and 1:1 scaling. This means the masks have to be extremely accurate, which makes them even more expensive. The mask also needs to be in close proximity to the wafer, which complicates the process. Other alternatives, like e-beam direct write, were also considered too slow for mass production of chips. That’s why EUV #photolithography turned out to be the most economically viable option for the foreseeable future. Even though it's a extremely complex machine with a light source that uses exploding tin droplets to create EUV light, it is the only tool that is fast enough for modern fabs. Substrate says it has built a tool that 'can meet the throughput of a leading-edge fab'. Chip production is not only about resolution, as former board member Frits van Hout explained in '#Focus, the ASML Way'. In any chip fab, adding value and speed of production is key. “Keep on stamping, is all we hear.” As a sidenote: ASML is researching the use of X-ray for inspection and metrology, not for printing chips. The company stopped its research on free-electron lasers (FELs, which are capable of creating shorter wavelengths) in 2015, when EUV turned out to be feasible. The New York Times https://lnkd.in/e2J9AB9i The Wall Street Journal https://lnkd.in/eif-Cq3r Financial Times https://lnkd.in/ev9B-PtP | 13 comments on LinkedIn