The Cap Table Ep. 3 - James Hardiman
========================================
Hi, I'm James Hardiman, and this is The Cap Table.
Affinity put together some questions for us, and we're gonna walk through them today. Tell us about yourself, from your roots in engineering physics to your time at Blackstone and the mission of DCVC in 2026. So I studied engineering and physics at Berkeley because I was really interested in how the world worked, and I wanted to know from first principles what were the rules that governed it and how to actually apply those to make things.
As I went through my career, I really wanted to help other people, take the technologies that they were working on and learn how to commercialize them. So part of doing that was learning how to invest. That's how I ended up at Blackstone. I was only there for a short period of time when I had the opportunity to join DCVC, which was focused on this exact mission, right?
How to help startups that are working on AI or the applied sciences develop those technologies and commercialize them, and that is still DCVC's thesis today. We focus on companies that are applying AI or are built in some kind of science or engineering innovation. How do you balance the hard science of evaluating quantum qubits with the sales job of venture investing?
When you're investing in companies that are developing hard technologies, the entrepreneurs really want investors that understand and appreciate those technologies. Not only do I have to research and understand the technology that they're investing or developing so that I can understand the risks associated with it, but if I'm gonna be a good partner to them in helping them make maybe strategic decisions around how to apply or where to take that technology, I really need to understand it.
And so understanding the technology is actually a really important part of selling myself to companies that have the opportunity to take capital from multiple places. Next question. Atoms versus bits. For twenty years, software was eating the world, but in 2026, the biggest bottlenecks are physical: electricity, water, and biological data.
Are we entering a decade where atoms finally outperform bits for investors? So I'd say that hard tech or physical companies have generated large outcomes for investors for a long time. Semiconductor investing was very successful all the way back to Fairchild Semiconductor. Genentech was a huge outcome venture invested by Kleiner Perkins, you know, decades ago.
So rather than thinking about do things shift between software and hardware, there's a few underlying tenets that I think are always true about venture, and that's about getting big fast and doing so capital efficiently. Now, it was true for a long time that the companies that were going to be most capital efficient and get biggest fastest were software companies.
However, I think the cost of building hardware companies have come down significantly, and the demand for many of the things we were just talking about, electricity, water, et cetera, has given hardware companies an opportunity to also get big fast by servicing that demand. Next question. The energy wall.
Every startup is racing to build a better AI model, but we're hitting a massive electricity bottleneck. If compute is the new oil, does the AI race actually belong to the person who owns the modular nuclear reactor? I would say yes. I think that's true. Value often accrues to the scarcest resource, and in the AI race, there are, for example, many open source models where you might not actually even have to pay a tax to the model developer for running that model.
But the things that will, always require, or will always capture value will be the compute and the energy necessary to run those. Next question. The SaaS-ification of biology. You've called the Noetik GSK deal a dollar sign moment for the industry. How is AI changing the traditional biotech business?
When you think about how drugs are developed, traditionally software has played a discovery role where you might computationally model a drug that gets tested in some Petri dishes, right? Maybe advances to animal models. But most of the value is actually created when that drug is tested in humans and advanced from a discovery candidate to a phase one trial, and then a phase two trial, and a phase three trial.
Traditionally, again, software has played a role in that beginning part, in the discovery business, but the value has been captured in the later stages. Noetik's technology allows drug developers to actually change the probability of success in those clinical development stages. So rather than bringing all of the patients that might benefit from a drug into the trial, they're able to select the ones that are most likely to respond.
And because of that, the efficacy of that drug is improved in that trial. So rather than thirty percent of patients responding, sixty percent of patients would respond if you use Noetik software to select patients as they come into the trial. That is an enormous step up, and it makes the FDA much more likely to approve this drug.
And because of that, I see a future where most clinical trials are run on a subset of patients, where those patients are much more likely to respond than just the general patient within a particular disease. Next question. Quantum on the factory floor. Quantum hype is everywhere, but DCVC recently led a one hundred and sixty million dollar round for Quantum Motion.
Is silicon-leveraged quantum a better bet for institutional money than more exotic architectures? The way I think about quantum today is a religious war. It's still an open question for the experts in the field, which of these architectures is gonna win. If you grew up in a superconducting qubit lab, you see the benefits of superconducting qubits.
If you grew up in a neutral atom lab, you see the benefits of neutral atoms. And I think in this interim period, many of these architectures can win. Some of them are faster, some of them have better qubit performance. And so I think there will be a period where neutral atom computers, ion computers, superconducting computers can all exist, be purchased, be leveraged by customers to actually deliver economic value.
Now, in the long run, will one architecture dominate? It's possible. Is that architecture likely to be silicon-based? Given all of the money that's been poured into semiconductors to date, the billions and billions of dollars, being able to leverage that is a huge advantage for silicon-based architectures.
So I do think in twenty, thirty years, it's possible that the winning architecture is a silicon-based, spin qubit. All right, moving on. The Blackstone versus VC mindset. What is the one private equity mental model that more VCs should adopt, and one VC habit that PE shops should probably unlearn? Private equity is extremely focused on operational excellence, and I actually think VCs could probably benefit by focusing on that more in their diligence and when sitting on the board of companies.
There's a lot of talk in VC around strategy and moats, and one of my favorite expressions is culture eats strategy for breakfast. So what are the things that a company does on a day-to-day basis? What are the systems that they have in place to drive operational excellence? That's gonna be more important than some theoretical strategy or moat that you try to set out.
In terms of reverse, there's probably not a whole lot that PE companies can learn from VCs. We're a pretty loose, shoot-from-the-hip bunch. Next question: The infrastructure handoff. At what point does a breakthrough in physics or biology stop being a venture bet and start looking like a mature, PE-ready infrastructure asset?
As soon as you can describe an ROI. So once you know how much money coming in can produce how much money coming out, that's when I think things move from more of a VC to maybe a PE underwriting model. VC is a lot about uncertainty, maybe that science/engineering risk. We actually don't know exactly how much capital it's gonna take for me to get to a value inflection point.
Once things become underwriteable in terms of a $50 million plant is gonna produce, $100 million worth of output, every year for the next ten years, then you're looking for a different kind of capital other than venture capital. Moving on. Data versus conviction. In engineering physics, data is king.
In early-stage VC, the data often doesn't exist yet. How do you learn to trust your gut on a breakthrough that the math hasn't proven yet? I describe being a good VC as having good taste in art. You kinda know what good looks like. You learn that there are certain kinds of risks or certain kinds of uncertainties that you get comfortable with as a VC.
And there's certain companies that I like that others don't, just with art, there's certain kinds of art I like that others don't. That doesn't mean that those other investors aren't good. They can have good portfolios and good returns, but learning what are the specific kinds of companies that you like and the specific kind of risks that you like, I think, is part of becoming a full stack VC.
All right, next question. There's a lot of talk in VC about wins, but as an investor, what is one miss or failure that taught you more about picking founders than any successful IPO ever could? So one thing you have to keep in mind as an investor is that you're not actually in the company operating. So you really need to trust whoever is at the head of that company, oftentimes the founder or maybe the CEO if a new one has come in. And the judgment that they have and ability to respond to the market.
Because it's really them who are making decisions every day about what to do in response to the market as new things come up. Even being just a little off can often compound over time. If there's a founder who makes the right decision 95% of the time, that means 5% of the time they're making a bad decision.
That compounded day after day can take you off track pretty fast. So next question: What's a piece of data that you value more than almost any other when evaluating a founder for the first time? Probably references. If I could choose one piece of data to have about a founder before making an investment, it would be a reference from a manager, from an investor that had backed them before, or maybe from a C-suite that had worked closely with them in a prior company.
Next question: What's a habit from your time at Blackstone that you refuse to give up even in the world of venture capital? Generally, private equity investors are very disciplined about what the exit opportunity for a company looks like, and I think that's one thing that VCs are less disciplined about.
Throwing around, "Oh, this company can be a $10 billion or $100 billion," and we even heard in some partner meetings a trillion-dollar company. Whereas I think, private equity investors are much more sober about, okay, what kind of revenue is this company going to generate, at what EBITDA margin? And if I look at comparable companies, what does that imply about the value of that company in a publicly traded, shrewd market?
All right, next question. Okay, now we're doing a lightning round. Nvidia or uranium: Which is the scarcer resource for AI in 2030? Assuming uranium means power, then definitely uranium. Building and permitting new power plants is extremely difficult, and we're not gonna have a lot between now and 2030.
Silicon spin qubits: the Holy Grail of quantum or just one of many? One of many for the next five years. 20 to 30 years from now, probably the Holy Grail. All right, next question. The Series A crunch: Is it over or are we just getting started? I think it's gonna be haves and have-nots for a while. If you're working on something AI related, maybe it's gonna be easy.
If you're working on something that's a little less hyped right now, then fundraising is probably gonna be difficult, and I think that's gonna be the case for the next several years. Remote work: Can you build a world-class hardware company over Zoom? No. The expensive mistake.
What is the biggest error GPs are making in AI right now? The biggest mistake GPs are making is thinking that every company can possibly be acqui-hired for five billion dollars. I think this is a moment in time that we're basically reaching the end of, and valuations and outcomes will trend back to rationality.
And so if you're making an investment where you're underwriting that as the bear case, it's gonna be a bad time. All right. Thanks so much to Affinity for having me on the Cap Table. We'll see you on the next one.
