My phd is in the physics of complex systems, so its not exactly this area but the arguments in the engrxiv paper linked in TFA seem robust to me. In my mind there is an abstract question and a concrete question which illustrates the hardness of the problem.

The concrete question is, were there historical examples which showed the feasibility of this kind of endeavor? I dont think so. In fact there is a recent counterexample. Biofuels. They were big in the last hype cycle. George Church, the Tom Knight of Harvard, got involved in Joule, which is now dead. Biofuels was a much simpler pipedream than todays lab grown factory and yet… we are back to designing batteries, which are far simpler than algae.

So after this bubble bursts, I predict that we will be putting our money in making jellyfish tasty. it will be a step up from making yeast or algae tasty for sure.

Another prediction would be uh the second coming of 6502… of amino acid manufacturing.

The abstract question is, is it possible to design a large scale self replicating machine that runs more efficiently than evolution can give us? Again, no historical examples.. but I see the current deep learning “success” as a victory for evolution, not design…

My guess is, if we want to make lab grown meat work we might as well solve the easier problems of immortality and general AI, first.