Who is Max Hodak?

Neuroscience researchers at OpenAI and Anthropic have noted it is easier to do neuroscience on AI models than on biological brains, suggesting AI models are now the most productive substrate for neuroscience research.

“Ironically it's probably working on AI. Yeah, I have some I have a couple of neuroscience [clears throat] friends at Open AI and Anthropic who it's like we would joke like, "Oh, you left neuroscience?" Like, "No, no, no. It is just way easier to do neuroscience on the models.”

When supplies are bought in bulk and not attributed to specific experiments, experiments appear to cost nothing, obscuring true R&D costs.

“Therefore, experiments are free.”

Deep tech startups rarely fail because the technology doesn't work; they fail because they outgrow their management systems for physical infrastructure and people.

“It's uncommon that these deep tech companies that fail because the technology doesn't work. They fail because once you end up with this organization of hundreds of people and thousand hundreds of thousands of square feet of physical infrastructure, you haven't built the systems to manage that.”

A single wafer iteration in one BCI prototype protocol costs $40,000, making per-experiment costs easy to underestimate.

“For every iteration of a wafer that we make, in this case, for this protocol, it cost $40,000.”

Science Corp is moving its next retinal prosthesis into animal studies now, targeting human trials next year, with an addressable population expanding from hundreds of thousands to millions of patients.

“The next version, which is going to animal studies now, will be in humans hopefully next year, should expand that to millions.”

Max Hodak on why deep tech companies fail, not because the tech breaks but because management systems can't scale.

“It's uncommon that these deep tech companies that fail because the technology doesn't work. They fail because once you end up with this organization of hundreds of people and thousand hundreds of thousands of square feet of physical infrastructure, you haven't built the systems to manage that.”

Max Hodak on binary funding thresholds for deep tech bets.

“There are definitely some ideas that are worth funding with $50 million or zero dollars, but not $5 million.”

Hodak explains that implant packaging, called tropicalization by European engineers, is the open problem of keeping device internals in and the body out of an implanted device.

“The harder kind of more open problems is what we call packaging our colleagues in Europe call it tropicalization this is the your ability to keep your device in and the body out of an implant that you put in the body.”

Max Hodak explains that AI models and biological brains share measurably similar geometric structure in how they represent concepts, a finding Science Corp uses practically by aligning animal neural recordings with AI model internals.

“If you look at how do how do these AI models represent just like represent concepts and you look at the parts of the brain that represent concepts, you see these you see very similar geometry. That to me was one of the first clues. Like when I really saw that and we use that practically. Like we use that constructively at science. Like that's we know that is that is true because we can get alignments between animal brain neural recordings and AI model internal.”

Hodak uses Anthropic Sonnet's performance on a GPU kernel optimization task as a hiring bar, requiring candidates to beat the model to earn an interview.

“The hurdle for a while was I think it was sonnet's performance if you could beat that then you could get an interview.”

Max Hodak argues that AI models have become the most productive substrate for neuroscience, more tractable than biological brains.

“Ironically it's probably working on AI. Yeah, I have some I have a couple of neuroscience [clears throat] friends at Open AI and Anthropic who it's like we would joke like, "Oh, you left neuroscience?" Like, "No, no, no. It is just way easier to do neuroscience on the models.”

Hodak describes IGEN reviews as a graph-based eigenvector-centrality approach to performance reviews where peer votes are weighted by the voter's own rating, making the system self-calibrating.

“We call this technique IGEN reviews and I've become convinced that this is more or less the right way to do performance reviews.”

Max Hodak on iteration speed as the decisive competitive moat.

“If you can learn one thing every week and there's a competitor that's learning a thing every month, they will they will never matter.”