Great post Ben. Resonates with my experience of developing and commercializing materials . Gotta nail the product, process technology and qualification of the products in use cases that are willing and able to pay are incredibly difficult. The gold standard is 10 years from bench to commercial product/plant. Bakelite by Baekland, nylon by dupont team and the blue LED by Nakamura are the only 3 examples i know that did this < 10 years. Are you aware of any else ?
I've been working on deep learning at Google for almost a decade. Most recently I was working on Gemini at Google DeepMind. I quit a month ago to take a sabbatical and chase my intrinsic curiosity.
Where do you think my contribution would be highest-leverage in materials science?
It’s hard to agree more with this essay. Advances in materials have enabled every aspect of technological progress. In a big example in my field, the semiconductor industry was birthed out of that initial transistor research from Bell Labs where it quickly led to the formation of Silicon Valley and a juggernaut of an industry that is a not-insignificant proportion of many different countries’ economies.
Better transistors required better performing materials, and not just materials used in the transistors themselves but materials required to pattern them, etch them, and integrate them into useful devices. Purity demands for each material and chemical used thereof drive further innovations. Today, every improvement to our semiconductor devices is nearly entirely dependent upon continuous material and processing advances from raw materials to packaged circuits.
As devices dimensions shrink and more and more transistors are packed closer and closer together, new materials are required. Silicon oxide, for example, is no longer an effective insulator (dielectric really) when it is too thin. Copper can’t conduct or dissipate heat as well when the wires (interconnects) are too thin. I could go on and on… we need new innovations across literally every aspect of this ecosystem.
New materials are required to continue improving everything. Materials and manufacturing are absolutely integral to Progress. I applaud your efforts at trying to bolster the cause.
Great post Ben. Resonates with my experience of developing and commercializing materials . Gotta nail the product, process technology and qualification of the products in use cases that are willing and able to pay are incredibly difficult. The gold standard is 10 years from bench to commercial product/plant. Bakelite by Baekland, nylon by dupont team and the blue LED by Nakamura are the only 3 examples i know that did this < 10 years. Are you aware of any else ?
I've been working on deep learning at Google for almost a decade. Most recently I was working on Gemini at Google DeepMind. I quit a month ago to take a sabbatical and chase my intrinsic curiosity.
Where do you think my contribution would be highest-leverage in materials science?
Figure out how to use AI for scaling, not just discovery!
It’s hard to agree more with this essay. Advances in materials have enabled every aspect of technological progress. In a big example in my field, the semiconductor industry was birthed out of that initial transistor research from Bell Labs where it quickly led to the formation of Silicon Valley and a juggernaut of an industry that is a not-insignificant proportion of many different countries’ economies.
Better transistors required better performing materials, and not just materials used in the transistors themselves but materials required to pattern them, etch them, and integrate them into useful devices. Purity demands for each material and chemical used thereof drive further innovations. Today, every improvement to our semiconductor devices is nearly entirely dependent upon continuous material and processing advances from raw materials to packaged circuits.
As devices dimensions shrink and more and more transistors are packed closer and closer together, new materials are required. Silicon oxide, for example, is no longer an effective insulator (dielectric really) when it is too thin. Copper can’t conduct or dissipate heat as well when the wires (interconnects) are too thin. I could go on and on… we need new innovations across literally every aspect of this ecosystem.
New materials are required to continue improving everything. Materials and manufacturing are absolutely integral to Progress. I applaud your efforts at trying to bolster the cause.