Berkeley Lab

“Chips have a lot of memory, but no imagination”

Supriya Jaiswal of Astrileux Corporation gave a refreshing talk today on a start-up’s effort to tackle a ten year industry problem. Jaiswal explained that we live in the era of zettabye computing – we use 10 nm chips (what your fingernail grows in 1 second!) and we’ve manufactured more chips than there are stars in the Milky Way!

Despite this growth, the industry didn’t deliver on e-beam lithography, a common chip making technology. According to Moore’s Law, we should actually be using 5 nm chips by now, but as Jaiswal cheekily pointed out, “Moore’s law was…no more.” The main lithography tool that is now considered the next generation tool for chip production is extreme ultraviolet (EUV) lithography, yet EUV faces the problems of insufficient light source power, inadequate photoresist sensitivity, and mask defectivity. To combat this, Astrileux’s main priorities are making new materials for the EUV light souce, scanner tool, and photomask – this could increase the efficiency of EUV by orders of magnitude and reduce the power consumption. But how can a start-up compete with industry?

Jaiswal’s talk, while not technical, answers this question with three wise suggestions. Firstly, she claims that innovation requires solving the right problem and winning on science, rather than engineering. In the end, billions of dollars spent on state-of-the-art equipment cannot beat the potential of a good scientific idea. Secondly, because money is usually tight, start-ups need to find redundancies and leverage the system, the people, and the equipment that the company has access to. Finally, it is crucial to build a strategic network before building the company.

“Chips have a lot of memory,” Jaiswal quotes, “but no imagination!”