A decade ago, if you told a roomful of seed investors that you were building a satellite company, an electric powertrain company, or a new class of autoimmune therapeutic, you'd have gotten a polite nod and a referral to "someone better suited for that." Today, those are exactly the companies showing up in my inbox — and exactly the companies the smartest LPs in the world want exposure to.

Hard tech is having a moment. And it's the most exciting time in my career to be backing the founders building it.

What "atoms, not bits" means at Refactor

"Atoms, not bits" is shorthand for companies that build physical things — molecules, machines, materials, infrastructure — instead of products that live only on a screen. It's how I've described Refactor's investment thesis since I started the firm in 2015.

Atoms-based companies are harder to start. They take longer to build. They cost more capital. They require regulatory approval, supply chains, and pilot facilities. By every conventional measure, they look like worse venture investments than pure software.

But the moats are real. The customers are sticky. The outcomes, when they work, are generational. And the founders who choose to do this work are doing it because they believe it has to be done — not because they're chasing a trend. Those are the people I want to spend the next decade backing.

Why now is the moment

A few things are happening at once that make 2026 a remarkable time to be a hard tech investor.

First, the cost of building physical companies has come down. Reusable launch capability has cut the cost of getting things into space. Modular manufacturing and contract bio-fabs have made it possible to prototype hardware and biology faster than ever. AI is genuinely useful for designing molecules, optimizing chip layouts, and accelerating simulation cycles — not as a replacement for engineers, but as a force multiplier for them.

Second, the United States is rediscovering the importance of building things. Reindustrialization, energy independence, biosecurity, and supply chain resilience are bipartisan priorities now. Capital that used to flow only into software is finally flowing into the critical industries — aerospace, energy, nuclear, defense, advanced manufacturing, critical materials, biology — that underpin a functioning economy.

Third, and most importantly, the talent is here. The best engineers, scientists, and operators I've ever met are choosing to build physical companies. They're leaving SpaceX, Tesla, Moderna, and the national labs to start things. That's the strongest leading indicator I know of for where the next decade of returns will come from.

What this looks like in our portfolio

A few examples from the Refactor portfolio that show what "atoms, not bits" looks like in practice.

Brelle is building electric powertrains for the kind of vehicles that are still stuck on diesel — heavy-duty, off-highway, industrial. The team understands that electrifying a haul truck or a forklift is a fundamentally different engineering problem than electrifying a sedan, and they're solving it from first principles. Every Brelle powertrain that ships displaces a diesel engine that would otherwise be running for the next twenty years.

Astranis is building small geostationary internet satellites that bring broadband to the parts of the world the legacy operators have ignored. They're putting birds on orbit that connect entire regions — Alaska, the Philippines, Mexico, Peru — that haven't had real internet before. It's hardware, propulsion, RF engineering, and orbital mechanics, all done by a team that decided to vertically integrate every piece of it.

Epana is developing next-generation autoimmune therapeutics — going after diseases like lupus, rheumatoid arthritis, and inflammatory bowel disease with a fundamentally new modality. Autoimmune conditions affect tens of millions of people, and the existing therapies are blunt instruments. Epana is building precision tools instead.

Brelle, Astranis, and Epana don't have much in common on the surface. One builds powertrains, one builds satellites, one builds drugs. But all three are run by founders who decided to take on a hard physical problem because they believed the world would be better if it got solved. That's the through-line of every investment we make.

What we look for at the seed stage

When a hard tech founder shows up in my inbox, I'm looking for three things.

A team with deep, earned expertise in the physics, biology, or chemistry of the problem they're solving. Hard tech doesn't reward generalists at the seed stage — it rewards people who have spent years inside the industry they're now refactoring.

A real, defensible moat — proprietary data, a manufacturing advantage, a regulatory pathway, or a piece of hardware that's genuinely hard to copy. Software-style moats don't translate to atoms.

And most importantly, founders who are absolute magnets. Magnets for customers, talent, and investors. Their outlier ability to gather people, resources, and capital to their mission is why VCs always say they like to "Back the best founders". It may sound trite, but really, they're seeking magnets.

The most optimistic time to be backing builders

I'm an optimist about almost everything, but I'm especially optimistic about this. The founders building physical companies right now are working on the problems that actually matter — clean energy, accessible internet, better medicines, secure supply chains, the rebuilding of American industrial capacity. The capital, the talent, and the technology are all aligned in a way I haven't seen before in my career.

More things are possible right now than most people imagine. If you're a founder building something physical and you want a partner who'll show up early, write a real check, and stay close for the long haul, send me a note on the Contact page. All submissions go straight to my inbox.

‍

‍

Building a hard tech company is one of the most punishing things a person can do with their life. You're trying to bend physics, biology, or chemistry into a product, raise capital against decade-long timelines, hire engineers who could be making twice as much at a hyperscaler, and convince a skeptical world that your version of the future is the right one. And you're often doing it on five hours of sleep.

Founders pay for that ambition with their mental health. So do the engineers, scientists, commercial folks, and operators they bring along for the ride. I've watched it happen across five funds.

So, back in 2020, during COVID, I started picking up the bill. For everyone.

What we offer, in plain terms

Every founder I back gets free mental health therapy and coaching. So does every single employee at every seed-stage Refactor portfolio company: engineers, scientists, operators, the whole team. No copays. No paperwork from the company. No hoops.

We do this through our partner Lyra Health, which runs an extended network of more than 5,000 licensed therapists and coaches across the US. Founders and their teams can book sessions for anxiety, burnout, founder isolation, executive coaching, relationship support, grief, addiction recovery — the full spectrum of what humans actually deal with when they're trying to build something hard.

Lyra sends me one bill a month. I pay it out of the Refactor management company. That's it. That's the whole program. Once companies graduate by raising a Series A, I typically ask CEOs to pick up that benefit for their teams.

How the privacy works

This is the part that matters most, and the part founders ask about first.

The Lyra invoice that hits my inbox every month is anonymized. It does not show me who used the service. It does not show me which portfolio company they came from. It doesn't break out sessions by individual or team. I see a number, I pay the number, I move on.

That design is intentional. The minute a founder (or one of their employees) thinks their investor might find out they're in therapy, the program is dead. So I built it so I literally can't find out, even if I wanted to. I'm not the gatekeeper. I'm the check writer.

Why a seed fund covers this instead of waiting for the company to

Most seed-stage hard tech companies — frontier tech, deep tech, the kind of work happening in critical industries like aerospace, biotech, energy, nuclear, defense, advanced manufacturing, and critical materials — don't have HR for the first few years. They have a founding team and a runway. Health benefits get bolted on later, usually around the Series A, and mental health coverage is almost always the last thing added.

That timing is exactly backwards. The hardest psychological stretch of building a company is the first two years, when the team is small, the science is unproven, and the world hasn't caught on yet. The first ten employees of a hard tech company are taking the same existential risk the founders are; sometimes more, because they don't have the founder's equity story to comfort them. Waiting until the Series A to give that team access to a therapist is like waiting until you've crossed the desert to hand out water.

So Refactor covers it from day one. For the founders, and for everyone they hire. It is not charged back to portfolio companies. It does not affect a company's burn. It doesn't show up in a board deck. It is, for the founders and their teams, free.

Why I think every fund should do this

The math on this is genuinely not hard. Lyra's pricing for a portfolio of our size is a rounding error against what we deploy in capital each year. The downside of not having it — a founder spiraling alone, a key engineer quietly burning out, a CEO who doesn't realize they need help until they've already made the irreversible decision — is a fund-returner-killing risk that nobody puts on a risk register.

Also, it's a great way for founders and employees to test out having a professional coach on my dime.

The job of a seed-stage venture investor is to give portfolio teams every unfair advantage I can. Capital is the obvious one. But access to a therapist who's seen a hundred other founders and operators go through exactly what you're going through right now, on a Tuesday at 9pm when you can't sleep — that's the kind of edge that doesn't show up on a pitch deck.

The hard tech Refactor backs will take a decade or more to build. The least I can do is make sure the people building it stay healthy, perform at their peak, and can sleep at night, well at least on the nights they actually want to. 🙂

#BetterCallZal

‍

‍

In March 2025, Causal Labs announced a $6M Seed round. I'm grateful Kelsie Zhao and Dar Mehta let me participate.

The round was led by Kindred Ventures, with participation from Refactor Capital, BoxGroup, Factorial, Otherwise, Karman Ventures, and a strong group of angel investors. But the round isn't really the story. The story is what Kelsie and Dar are actually building.

Causal Labs is building what its founders call a Large Physics Model — a foundation model for cause-and-effect reasoning in the physical world. Their first deployment is in weather prediction and weather control. The longer arc is something more fundamental: the missing piece of intelligence that today's AI systems do not yet have.

What Causal Labs Does

The current generation of AI is extraordinarily good at pattern recognition over text and images. It is much less good at reasoning about physical cause and effect in the real world.

If you've ever wondered why self-driving cars are still hard despite a decade of investment, or why robotics has lagged behind language models, the answer lives in this gap. A model trained to predict the next token can recognize patterns in language. It cannot, on its own, predict what will happen if you push a glass off a table, divert a river, or seed a cloud. Those predictions require an internal model of physics, of causality, of how an action propagates through a system over time.

Causal Labs is building that layer.

The bet — and it's a bet I find compelling — is that the right way to teach an AI system causality is not to scale up text training. It's to train on a domain that is physics-rich, data-rich, and inherently chaotic. Weather satisfies all three. The atmosphere is governed entirely by physical law. There are petabytes of multi-sensor data — satellites, weather stations, balloons, radar — collected continuously. And weather is chaotic enough that prediction errors compound rapidly, which forces any successful model to learn real causal structure rather than surface correlations.

Solve weather, and you have a foundation model for physics. Solve physics, and you have one of the missing pieces of general intelligence.

Who Built It

Kelsie Zhao and Dar Mehta are not weather researchers. They are former safety-critical AI engineers from Cruise, Waymo, and Google Brain — people who spent the last decade building deep learning systems for autonomous vehicles, where being wrong about cause and effect is not a benchmark question, it's a fatality.

That background is the entire premise of the company. The hardest unsolved problem in self-driving is out-of-distribution generalization — what happens when the car encounters something it has never seen before. Today's AI, including state-of-the-art language models, does not handle this well. It interpolates within its training distribution. It does not reason from first principles. The Causal Labs founders watched this problem block deployment timelines at Cruise and Waymo, and they came away with a clear thesis: pattern recognition alone will not get us to systems that act safely in the real world. You need physics. You need causality.

So they left to build it.

Why Weather, Why Now

A few reasons this team's bet on weather is especially well-timed:

The data is there. Weather is one of the few physical domains with the volume, density, and modality diversity required to train a frontier physics model. Satellite imagery, ground stations, radar, balloons, ocean buoys — petabytes of multi-modal sensor data, continuously generated, publicly available. Robotics, by comparison, has nothing like this scale of structured physical data.

The customers are there. Aviation, agriculture, energy, logistics, insurance, and federal/state/local governments all make multi-billion-dollar decisions every day based on weather forecasts. The current state of the art — physics-based numerical weather prediction — is hitting its limits. There is real, immediate enterprise demand for higher-resolution, real-time, hyperlocal forecasts and the decision-support that comes with them.

The climate stakes are rising. Extreme weather events are becoming more frequent and more economically consequential. The infrastructure of human resilience — from disaster response to grid management to agricultural planning — depends on better models. This isn't a speculative future market. It's a market that needs better tools today.

The longer arc is weather control. This is the piece that gets less attention but is, structurally, the most interesting. Causal Labs' eventual ambition is not just to predict weather but to enable safe, steerable interventions: cloud seeding for drought relief, hurricane intensity reduction, wildfire suppression. That's a category of capability that doesn't really exist yet, and it lives downstream of getting the physics modeling right.

A Few Things I Admire About Dar and Kelsie

Sitting on the cap table of a company like this is a privilege, not a victory lap. A few things that stand out about the way they're building:

They earned the right to take this swing. They are not academics parachuting into a flashy problem. They are operators who built safety-critical AI for a decade, watched the field hit a wall on out-of-distribution generalization, and identified the underlying gap themselves. The company is downstream of their diagnosis, not the other way around. That's rare.

The thesis is structurally contrarian, in the right way. Most of the AI ecosystem is building on top of LLMs. Kelsie and Dar are building underneath them — the physics layer the language layer is missing. They saw something most of the field hasn't yet, and they're acting on it.

They've structured the company sensibly. Weather forecasting and decision support is a near-term enterprise business with paying customers. The physics foundation model is the long-term platform play. They're aware that one funds the other, and they're not pretending otherwise.

Safety is a first principle, not an afterthought. They built the company around safety from day one, drawing on their AV experience. In a category where the long arc involves intervening in real-world atmospheric systems, that posture isn't optional — it's what makes the whole thing buildable. I trust them to be careful.

A Closing Thought

Most of the AI conversation right now is about scale: more compute, more parameters, more tokens. Causal Labs is one of a small number of companies betting that the next breakthrough won't come from more of the same — it'll come from teaching machines something they don't currently know how to learn. Cause and effect. Physics. The structure of the world.

If they're right, the next decade of AI looks very different from the last one. I think they're right.

‍

‍

A small but historic thing happened in our portfolio this month: The Biological Computing Company (TBC) — formerly known as BBB — closed a $25M Seed round led by Primary Venture Partners, with participation from Builders VC, Refactor Capital, Wonder Ventures, E1 Ventures, Proximity, and Tusk Ventures.

But the round isn't actually the headline. The headline is what TBC has quietly become: the first company ever paid by a customer for biological compute.

Not a grant. Not a sponsored research agreement. A real commercial transaction, for compute work performed by living neurons, paid for by a customer who needed the answer.

If that sentence sounds dry, sit with it for a second. It is, in a literal sense, the beginning of a new computing substrate going to market.

What TBC Actually Does

TBC connects living neurons — yes, real biological neurons — to electrodes, then uses them as a computing substrate alongside modern AI systems. The team's first product has demonstrated a 23x retained improvement in video model efficiency: real-world data gets encoded into living neurons, the neural activity gets decoded into richer representations, and those representations map onto state-of-the-art AI architectures.

This isn't neuromorphic design. It isn't "brain-inspired" silicon. It's biology, doing the computation directly.

The premise underneath the company is one of the cleaner first-principles arguments in deep tech: the human brain delivers roughly an exaflop of computational power on about 20 watts. To get the same throughput out of NVIDIA's Blackwell, you'd need approximately 120,000 watts. Biology is a million times more energy efficient than the best silicon we've ever built. For decades, neuroscientists have tried to translate the brain's functionality into digital systems. TBC is doing the inverse — letting neurons compute on their own, and integrating the result into the AI stack.

Who Built It

The founders are Alex Ksendzovsky (CEO) and Jon Pomeraniec (COO), both neurosurgeons trained at the University of Maryland and the University of Pennsylvania. They've been working on the underlying problem together for over two decades — through medical school, surgical residencies, and research roles — before founding the company in 2021 after a research breakthrough in the brain's computational capabilities.

There's a reason this company exists, and why now. You needed two people with this exact pairing of clinical neurosurgical experience and a long obsession with computation to even attempt it. Alex and Jon are, as Primary's Brian Schechter put it well, founders who are brilliant, on a mission, and wildly different.

Why This Matters for Hard Tech and Deep Tech

For most of the history of computing, "compute" has meant silicon. The entire stack — chips, datacenters, the AI infrastructure boom of the last few years — has been a story about getting more out of transistors.

But there are problems silicon is not the right tool for. Some of them, like the energy and efficiency demands of frontier AI models, are becoming more acute every quarter. The cost of training is going up. The power draw is going up. The thermal load on datacenters is going up. Something has to give.

Biological compute is one of the more credible answers to that pressure. It sits squarely at the intersection of deep tech, frontier tech, and the kinds of critical industries — AI infrastructure, advanced manufacturing, energy, defense — where the next decade of value creation will happen. It's a category that didn't exist as a commercial market until very recently.

The first paying customer is the hinge moment. Until you have one, it's a thesis. After you have one, it's a market.

I Wrote the First Check

I led TBC's pre-seed round and wrote the very first institutional check into the company. That's not a marketing line — it's literally true. There was no other institutional investor on the cap table when I signed.

This is the part of the job I actually live for. The pre-seed call is the one where the founder is describing something that, if true, is enormous, and where the diligence is mostly about whether the people in the room are the right ones to bend reality toward the vision. There's no traction to point to. There's no comparable company to benchmark against. There's just the founders, the substrate of the idea, and a decision.

I said yes early. I'm glad I did.

What's Happened Since

Watching TBC operate has been, honestly, one of the more remarkable things I've seen in a decade of investing across hard tech.

The team has hit nearly every milestone they laid out in the pre-seed deck — and in several cases, they've hit them ahead of schedule. They've recruited an absurdly strong technical bench. They've moved from research demonstration, to a working product, to revenue from a paying customer for biological compute, to a $25M Seed led by one of the most thoughtful infrastructure investors in the country.

That sequence — from pre-seed thesis to commercialized first product — is one almost no deep tech company executes cleanly. It usually takes longer. It usually breaks somewhere in the middle. TBC just kept moving.

Why This Was a Refactor Investment

People sometimes ask me what makes something a Refactor deal. TBC is a useful case study.

We invest at the intersection of computation and the physical sciences. We back hard tech and deep tech across critical industries — aerospace, critical materials, energy, nuclear, defense, advanced manufacturing, and the bio-compute layer that increasingly sits underneath all of them. We lead or co-lead seed and pre-seed rounds with $1–2M checks for 5–10% ownership. We do this roughly 20 times per fund.

The structural choice that lets us be useful in moments like the TBC pre-seed is that I'm the only employee at Refactor. There's no associate triaging the meeting. There's no investment committee debating whether the thesis fits a slide template. When a founder building something genuinely new shows up, the person hearing the pitch is the same person writing the check, and the decision can happen in the same week.

For TBC, that mattered. Alex and Jon needed a partner who could move at the speed of the conviction, not at the speed of a process.

What I'm Watching Next

The first paying customer is the beginning, not the end. The interesting question isn't whether TBC has a customer — they now do. The interesting question is how fast the second, fifth, and twentieth come.

If biological compute behaves the way most platform technologies behave, the curve is non-linear. The first customer is the hardest. The next handful come from the gravity created by the first. Then, at some point, the category stops being a curiosity and starts being infrastructure.

I think TBC is closer to that point than the outside world realizes.

‍

Today is Refactor's 10th anniversary.

Why "Refactor"

The word comes from software. To refactor is to rebuild something from the inside without breaking what works. That's what I wanted the firm to back. Founders rebuilding industries we'd given up on. Chemicals, energy, manufacturing, biology. The physical world.

A Decade In

Four funds. 100+ companies. Six unicorns. One employee.

The Bet, Restated

Atoms matter. Execution matters. And the most important companies are still the hardest ones to build.

Onto the next decade.

‍

‍

I spent a few days at a conference this week with the founders and managing partners of several large venture firms. Funds in the $500M to $3B range.

I came away more convinced than ever that being solo is the right model for me.

The Stat That Stopped Me Cold

I asked the same question to a handful of GPs: where does your daily attention actually go?

The answers were strikingly consistent. Between 50% and 75% of their daily email, Slack, and texts are internal. Their own team. Not founders.

Half to three quarters of their best hours, going to managing the people who manage the firm.

The Math of a Solo GP

I'm the only employee at Refactor. No partners, no associates, no investment committee, no chief of staff.

That means my inbox looks different. The vast majority of what I read and write every day is to and from founders. Prospective founders pitching me. Portfolio founders working through a hire, a board deck, a tough quarter.

When you remove the internal layer, the math changes. The hours other GPs spend running their firm, I get to spend with founders (and my agents, of course heh).

Different Models, Different Tradeoffs

Large firms have real advantages, and many founders thrive with that infrastructure. Different model, different tradeoffs.

But for the founders I want to back, building hard things in the physical world, what they need most isn't a platform. It's a partner who picks up the phone.

That's the bet I made in 2015, and ten years later, it's the bet I'll continue to make.

#BetterCallZal

‍

‍

Today, I'm thrilled to announce Refactor 5. A $50M fund focused on hard tech seed.

Ten years. Five funds. One employee with a single mission: backing the "hard but not impossible".

Some VCs and LPs call me "Mr. 50" because I keep raising $50M funds like clockwork. That's the playbook, and I'm not changing it. Concentrated portfolio. Disciplined fund size. Direct access to founders.

100% of Refactor 5 came from existing LPs. That kind of loyalty is rare, and I never take it for granted.

The Bet

The bet from day one was simple. Back founders building hard things in bio, energy, aerospace, manufacturing, and the physical world. Atoms, not bits. Things that matter for the long arc of human and planetary health.

Ten years later, the bet paid off.

Thank You, Founders

To every founder I've backed across funds: thank you. You are the reason this firm exists. Backing you is the best job I've ever had.

Thank You, LPs

To our LPs: thank you for betting on me when I went solo and for sticking with me through five funds. The 100% re-up rate on Fund 5 means more to me than any markup ever will.

How Refactor Works

I'm the only employee. No partners, no associates, no investment committee. Just me, the founders, and a phone that's always on. Across five funds and multiple SPVs where I've poured in capital into a small number of companies, that's $300M in AUM. Small funds, concentrated portfolios, fast decisions, direct access. When a founder needs me, they get me. Not a junior. Not a scheduling link. Me.

That's the whole product.

Refactor Is Open

If you're building something hard but not impossible, in bio, energy, aerospace, defense, nuclear, manufacturing, robotics, or anywhere else in the physical world, I'd love to hear from you.

You know what to do.

#BetterCallZal

‍

‍

During a Series A fundraise, founders text and call me constantly. I love it. Serving as a thought partner, giving rapid counsel. That's the job.

One particular founder called me maybe 10 times in a single day. We were negotiating terms with a couple of potential leads and about to make a decision on who to go with. On one of those late night calls, he says to me (I'm paraphrasing): "You know, I called you so many times today that I just decided to change your name in my phone to Better Call Zal. You know, like the TV show."

I laughed and laughed. It was so perfect. It's how you pronounce my name (rhymes with Saul). And hopefully, it's the tagline for every Refactor founder.

They know they can pick up the phone, shoot me a text, and I'll answer immediately. Mostly. If not, they'll get a ring back in minutes. Waiting until the next day, or even a few hours, doesn't cut it in our world. Things move fast. Founders are in desperate need of immediate help as they work through a problem or an opportunity. I want to be available for them.

Building trust, transparency, and hopefully lifelong friendships. It's why I do this. And why I do this solo, which lets me spend nearly all of my time with founders. Keeping my fund size the same makes fundraising with LPs faster, so I can focus on the job at hand. I've had the same loyal LP group for a while now. That helps tremendously.

So if you're a founder building something hard, and you want a partner who actually picks up the phone, you know what to do.

#BetterCallZal

‍

‍