Max Hodak on How Startups Build Speed

1. Key Themes

Speed Is the Ultimate Competitive Moat, and It Comes From Boring Infrastructure

Max Hodak's central thesis is that iteration speed separates winners from losers far more than technical brilliance. The compounding effect of learning faster than competitors is mathematically decisive.

"If you can learn one thing every week, and there's a competitor that's learning a thing every month, they will never matter." [00:18:38]

"Speed determines success and failure. And speed is determined by infrastructure. This is driven by really boring sounding things like how well do your purchasing and recruiting and spending processes work. This is as important as how well do you understand the object level technical content of the thing that you're building." [00:19:08]

Purchasing Systems Are Life-or-Death for Deep Tech Companies

The procurement problem is underrated by almost every founder. The failure mode isn't overspending on individual items — it's the lack of attribution and budgeting at the team level, which makes every experiment feel "free."

"If nobody knows how much an experiment costs, like every time you grow up a new cell line or every time we make a new probe in the fab, how much does that loop cost? Nobody knows. Therefore, experiments are free. It doesn't cost dollars. It costs media. And media comes from the fridge." [00:06:39]

"You can get yourself into trouble and... if you get them wrong, you'll end up spending $5 million a month and feel like you have very little control over it." [00:21:56]

Hiring Process Requires Rigorous, Systemic Infrastructure — Not Ad Hoc Judgment

Hodak outlines a four-step hiring process with company-wide voting at the top of funnel to prevent bottlenecks, distributed phone screens, AI-resistant homework, and high-conversion on-sites. Without a defined process, recruiting becomes a total time sink.

"A wrong answer for sure is not having something that you do very religiously as a company... You can easily spend almost all of your time recruiting if you're not doing it efficiently." [00:10:00]

"The system picks out seven or eight current employees that it thinks look something like their backgrounds, and it pings them all for votes... we can distribute the voting across a lot of the company for this initial review, which is essential because if you're doing anything cool by the time you get a couple years into it, that top of funnel is overwhelming." [00:11:29]

EigenReviews: A Continuous, Graph-Based Alternative to the Annual Performance Review

Hodak has developed and refined over six to seven years a PageRank-inspired system for continuous performance feedback that distributes judgment across the whole company and surfaces issues without the trauma and latency of annual HR cycles.

"Every couple weeks, every four to six weeks... people around the company get pinged with a question... knowing how this person turned out, would you vote again today for their hire? It's the same question we use on the initial voting." [00:16:43]

"Your vote should be weighted more highly if everybody else has rated you highly. And the astute may notice that this looks a lot like the original Google algorithm, PageRank, which is an idea called eigenvector centrality... We call this technique eigenreviews. And I've become convinced that this is more or less the right way to do performance reviews." [00:17:10]

Building Internal Software ("Helix") Is Now Justified and Strategically Essential

Historically, internal tooling was too expensive to build. Agents and AI coding have changed that calculus. Companies that build bespoke internal software create compounding operational advantages that off-the-shelf ERP systems cannot match.

"SpaceX and Tesla internally have a pretty giant piece of software called Warp Speed that runs a lot of their manufacturing and R&D processes. And so when one company grows up around a harness fit to it, it can be very powerful... The fact that you can vibe code this now makes it a reasonable thing to think about." [00:38:07]

"Everything that you can do in the company is a button somewhere in the software. We call it Helix... because this extends all the way through to manufacturing, where we have every step that happens in the lab in the database, we can correlate all of this through and get this information." [00:07:36]

AI's Biggest Immediate Impact in Deep Tech Is Regulatory Compliance, Not Just Coding

While coding improvements are well-known, Hodak identifies regulatory navigation — identifying applicable standards and building compliance evidence tables — as an underappreciated AI killer app for regulated industries.

"This thing can take many, many months historically. AI has totally transformed it. I mean, we can very quickly look up all the standards. We can very quickly generate the evidence tables. And I think that the combination of AI and regulation is a better fit than people think." [00:37:10]

Deep Tech Companies Fail at Organizational Execution, Not Technology

The primary cause of death for deep tech companies is the inability to manage the human organization at scale — not technical failure. This is an underappreciated insight in a world that obsesses over science risk.

"It is uncommon that deep tech companies fail because the technology doesn't work. They fail because once you end up with this organization of hundreds of people and hundreds of thousands of square feet of physical infrastructure, you haven't built the systems to manage that and it becomes unwieldy. And then you can't connect strategy to execution." [00:19:37]

BCI Is Primarily a Longevity and Healthcare Story, Not an AI Story

Hodak reframes BCI as an engineering-first approach to medicine that bypasses the hardest unsolved biology problems and delivers uniquely large, reliable effect sizes — making it a longevity play disguised as a niche hardware story.

"You turn on a deep brain stimulator and a patient goes from not being able to hold a cup of water to being able to write cursive in, like, 10 seconds... when you deal directly with the brain as a computer, not only do you not have to solve some of these really hard biology problems that are just beyond humanity's capabilities, but you get these results very... pretty readily." [00:40:56]

Startup Culture Is an Oral Tradition That Must Be Inherited, Not Reinvented

Successful startup cultures are rarely discovered from first principles. They propagate through networks of people who worked together at other high-performing organizations, making the choice of where to work early in one's career extremely high-stakes.

"It is relatively uncommon that startup cultures get rediscovered entirely from first principles. Usually they're passed down as oral traditions because there's a founding team that worked at another company, which worked at another company. So they inherited it." [00:30:43]

2. Contrarian Perspectives

Raise Twice What You Think You Need for the Experiment — Then Actually Run the Experiment

Most founders raise what seems reasonable or what they think investors will accept. Hodak argues you should price the actual experiment, then double it — because underfunding means you get an ambiguous outcome and never actually find out if your idea works.

"There are definitely some ideas that are worth funding with $50 million or $0, but not $5 million. You won't run the experiment. It'll be a really frustrating experience. You'll get an ambiguous outcome... figure out what it costs to actually run the experiment. Raise twice that." [00:43:43]

Deep Tech Companies Should Push for Profitability Far Earlier Than They Think

The conventional wisdom for ambitious deep tech companies is to prioritize the mission and fundraise as needed. Hodak argues that even with a decades-long roadmap, pursuing revenue and sustainability early is critical — both financially and for investor perception.

"No matter how big of a problem or big of a vision it feels, you do need to think about how do you get to revenue so that not just you can do it forever, but then you'll be valued on your long-term roadmap, not valued on your probability of dying. And it really opens up another set of investors that wouldn't be relevant otherwise." [00:45:09]

The Best Engineers Are Now in Industry, Not Academia — and the Gap Is Widening

Hodak challenges the assumption that academia retains the deepest expertise. When a field truly starts working, industry wins on resources and iteration speed so decisively that it attracts the best talent.

"20 years ago, the best computer scientists were at CMU and Harvard... and now they're at Google and Apple and OpenAI. And the best rocket scientists are at SpaceX and Blue Origin and others. So when a field really starts to work, industry can just marshal such larger levels of resources and can just move so much faster." [00:25:13]

You Cannot Delegate Your Judgment as CEO — and That Is a Feature, Not a Bug

Most startup advice is about seeking counsel and building consensus. Hodak argues the opposite: to achieve truly exceptional outcomes, you must make decisions that make sense only to you, even when you are entirely alone in that view.

"As the CEO, you must always make decisions that make sense to you, no matter how much momentum or inertia the alternatives seem to have... The successful companies are the exceptions. By becoming an average, that is not good enough." [00:20:31]

Building Internal Software at the Seed Stage Is Now the Right Call

Conventional wisdom says early-stage companies should buy, not build, software tools. Hodak argues that AI coding agents have flipped this, making bespoke internal systems a reasonable and strategically superior investment even at the seed stage.

"Historically software has been so expensive. You would have had to buy it. And that's what everybody did for a long time. That was, I think, a worse world and that world has changed. And so now there are better options available." [00:38:36]

3. Companies Identified

Science Corporation

A medical device startup founded in 2021 by Max Hodak and four co-founders from Neuralink. Its primary product is a photovoltaic retinal prosthesis — a wirelessly powered implant that restores vision to patients blinded by rod and cone degeneration. The device completed major clinical trials, was featured on the cover of Time magazine, and has results published in the New England Journal of Medicine. It holds a CE mark (European approval) and is running trials in six countries.

"Our main product is a retinal prosthesis. It's a chip that's implanted under the retina in the back of the eye to restore vision to patients that have gone blind due to loss of the rods and cones in their eye." [00:00:37]

"We are running clinical trials in six countries, now with an approved medical device in Europe and clinical trial results in the New England Journal of Medicine." [00:53:22]

Neuralink

Brain-computer interface company founded by Elon Musk. Mentioned as the origin institution for Max Hodak and four of his five Science co-founders. Cited as an exemplar of excellent judgment culture worth apprenticing at.

"I spent five years running a company for my CEO at Neuralink. I was working with someone who has empirically excellent judgment." [00:29:20]

SpaceX

Cited twice: once as an example of industry surpassing academia in rocket science, and once as a cultural model for integrating visionary "true believers" (committed Martian colonists) with serious technical professionals. Also cited for building Warp Speed, a large internal software system.

"About 20% of that company is what you might characterize as committed Martian colonists. And 80% are serious engineers... You need both of those cultures to be really successful long term." [00:52:24]

"SpaceX and Tesla internally have a pretty giant piece of software called Warp Speed that runs a lot of their manufacturing and R&D processes." [00:38:07]

Anthropic

Cited as an example of a best-practice employer (employees aren't hassled over equipment costs) and for designing an innovative AI-resistant hiring homework — GPU kernel optimization with a performance hurdle benchmarked against Claude Sonnet.

"Anthropic had a really interesting take on the AI-resistant homework where they've had a couple tasks where... it's the GPU kernel optimization. What is the minimum number of cycles you can get it down to? And this is naturally adjusting. The hurdle for a while was Sonnet's performance. If you could beat that, then you could get an interview." [00:13:53]

Y Combinator

Mentioned as a canonical example of an organization that has built proprietary internal software that makes the whole operation work distinctively well.

"YC famously has a lot of internal software that I think really makes YC work." [00:38:07]

Facebook (Meta)

Mentioned as a company that invested heavily in internal tooling and derived significant efficiency gains as a result.

"Facebook also very famously invested heavily in internal tools and now has a lot of efficiency from that." [00:38:07]

Tesla

Co-cited with SpaceX as a user of Warp Speed, the large internal manufacturing and R&D software platform.

"SpaceX and Tesla internally have a pretty giant piece of software called Warp Speed that runs a lot of their manufacturing and R&D processes." [00:38:07]

Greenhouse

Named as the commercial applicant tracking system Science used before replacing it with their internal Helix system. Identified as a bottleneck because it couldn't support distributed company-wide voting at the top of the funnel.

"We previously had used Greenhouse. Greenhouse required us to have a small number of people as a bottleneck at that first funnel stage. Replacing that with software, we were able to explore voting mechanisms and fairly detailed voting mechanisms." [00:38:36]

4. People Identified

Max Hodak

CEO and co-founder of Science Corporation. Former co-founder and president of Neuralink. Started his career nearly 20 years ago doing BCI research as an undergraduate at Duke. Architect of the eigenreview performance system and the Helix internal software platform. Synthesizes deep technical knowledge across neuroscience, microfabrication, and organizational systems.

"I'm the CEO of a company called Science... I've spent most of my life working on brain computer interfaces. This is almost 20 years ago now. I started my career as an undergrad working in a lab at Duke." [00:00:06]

Elon Musk (implicit — "my CEO at Neuralink")

Cited as an exemplar of empirically excellent judgment — specifically the ability to make high-stakes calls correctly under uncertainty, without waiting for feedback.

"I was working with someone who has empirically excellent judgment. Like we could get into trouble together and there'd be all... something would happen and there'd be two possible solutions that would make sense. And I'd go to him and say, is it option A or is it option B? He'd look at it and be like, oh, it's definitely option B. The problem would never occur." [00:29:47]

Paul Graham

Founder of Y Combinator. Cited for his observation that cities emit distinct cultural "vibes" that shape the ambitions of the people in them.

"It was Paul Graham that wrote a long time ago that you get vibes in different cities. The vibe in Cambridge, Massachusetts is you should be smarter. The vibe in New York is you should be wealthier. The vibe in San Francisco is you should be more powerful." [00:54:23]

5. Operating Insights

AI-Resistant Homework Design: Optimize for Pareto Frontiers, Not Correct Answers

The homework step of a hiring process is critically undermined by AI if it tests for known-answer problems. The solution is to design tasks that don't saturate — where improvement is open-ended and the output collapses to two or three numbers that can be plotted. Anyone who genuinely beats the Pareto frontier stands out automatically, regardless of what tools they used.

"Our favorite types of homeworks are things that don't saturate, have a very high ceiling, and are naturally scorable to two or three numbers that we can put on a plot. So that when we get responses to homeworks, we can just plot them all. And it's very obvious when someone has really beaten the Pareto frontier and we otherwise don't care whatever AI models they use." [00:13:23]

The Minimum Hiring Funnel Has Exactly Four Steps — And the On-Site Must Convert at 25%+

Hodak argues there is no more efficient way to gather enough information to make a hiring decision than: (1) company-wide voting on application, (2) distributed phone screen testing judgment, horsepower, and agency, (3) take-home homework, (4) full on-site interview. The on-site must convert to offers at at least 25%, or the process is burning too much team time.

"By the time you get to the interview, it is important that you have, from there, a reasonably high, at least 25% conversion to an offer. Because otherwise, it's just you're going to waste too much of your time doing on-sites for employees that don't convert." [00:14:21]

Use Markov Chain Monte Carlo Dropout on Graph-Based Reviews to Detect Voting Cartels

When running graph-based performance reviews, applying dropout over thousands of iterations and examining the distribution of resulting scores reveals hidden voting cliques: a single population produces one peak; cartels produce multiple peaks, flagging the sub-graph for manual investigation.

"We apply Dropout, where we'll run a thousand iterations, where we'll randomly remove some percentage of the edges each iteration. And then when you look at the distribution of scores that you get out of that, if you see additional peaks... this is a clue that there could be voting cliques that need further investigation." [00:17:39]

Spend Review Must Happen at the Budget Level, Not the Transaction Level

Approving individual purchases as they come in is the wrong abstraction. The correct intervention point is setting team-level budgets so that employees can make tradeoffs within their allocated resources. Transaction-level review is both too slow and too expensive in opportunity cost.

"Spending review has to come earlier. You have to have some concept of budgeting. It's not really about even just the payment rail of buying a thing. It's how do you understand the bucket of money that you have?... I don't want to be making the $1,500 power supply versus $3,000 power supply tradeoff. They need to understand the resources that they have so that they can make tradeoffs within those other resources." [00:04:44]

When Stuck, Force Action — It Produces Information That No Amount of Analysis Can

Hodak draws on physics to make the point that action is information-generating in a fundamental sense. When a company is stuck in a local minimum, the answer is to do something — even a decision as difficult as removing a high-performing individual who is the wrong fit — because it unlocks new possibility space.

"Action produces information... whenever you exert action into the universe, that creates information, like in a very fundamental sense... I've seen situations where the company is stuck in a deep local minimum and there's someone who is great in many ways, but it's just the wrong fit for what that company is at the time. And removing them, even though they individually are very strong, unblocks the company and allows it to kind of enter a new phase." [00:22:50]

6. Overlooked Insights

Cross-Domain Biology-Electronics Co-optimization Is a Structural Advantage of Small, Integrated Teams — and It Has Direct Analogies in Any Deep Tech Company

Hodak mentions almost in passing that Science's protein engineering group was able to make their opsins more light-sensitive, which reduced the power requirement per LED, which allowed more LEDs to be placed on the implant — converting an intractable electronics problem into a solvable biology problem. This was only possible because both groups were part of the same small team that could hold the whole problem in their heads. Large interdisciplinary centers with siloed experts "ship the interfaces of their departments" — i.e., their output is constrained by where the organizational boundary sits, not where the physics boundary sits. This is a non-obvious structural insight applicable to any hardware-biology-software stack: small integrated teams can relocate the bottleneck across domains, while large siloed organizations cannot.

"Our protein engineering group has been able to develop much more sensitive, like much better proteins for some things that we need to do, which has allowed us to relax some electronics requirements... we've turned this electronics problem into a biology problem that allowed us to relax those constraints. You don't get that as much when you have these interdisciplinary centers where there's like one group focused on one thing, there's another group focused on another thing." [00:33:28]

Conformal Coating / Next-Generation Implant Packaging Is a Wide-Open Materials Science Investment Opportunity

In a single throwaway paragraph, Hodak identifies what may be the most underappreciated bottleneck in the entire neural implant and broader implantable device industry: packaging — keeping the device intact and the body's immune response out, over multi-decade timescales, without a titanium box. There is no adequate commercial solution today. He explicitly names it as an open field for materials scientists. Given that cochlear implants, pacemakers, deep brain stimulators, retinal prostheses, and next-generation BCIs all face the same problem, a breakthrough conformal biocompatible coating would be platform technology with enormous TAM across all implantable devices.

"Having this next-generation packaging, some type of conformal coating that we can use to protect the implant that is not degraded by the body, is also not harmful to the body, and is resistant to all of the ways the body will try and kill it, that material science is a very open-ended field. If you're interested in material science, that is a thing that we need progress in." [00:49:44]