David Kirtley, Founder & CEO of Helion Energy

Carolyn and I are here with David Kirtley, founder and CEO of Helion Energy, building fusion power plants, and I’m thrilled we finally get the full story after missing you at the YC Retreat.

Let’s dive into how you actually build a fusion business.

Give us Fusion 101, please.

Fusion is what powers stars: light isotopes like deuterium and helium‑3 fuse into helium‑4 and a proton, releasing energy via that famous mass‑to‑energy trade, and our goal is to harness it on Earth for safe, low‑cost, always‑on electricity.

So we’re talking about clean electricity that feels almost unlimited?

That’s the mission, not free but designed to be cheaper than anything else because the fuel is everywhere and the machines can be built efficiently.

Simple physics check: the energy comes from the act of fusing, right?

Exactly, deuterium plus helium‑3 makes helium‑4 and a proton, the products weigh less than what you started with, and that mass deficit is the energy we capture as very hot, energetic particles.

Amazing, but what’s the brutal part?

Getting there means about one hundred million degrees and roughly a thousand atmospheres in one place, hotter than the sun’s surface and pressures like the Mariana Trench, which is why it’s taken decades to crack.

Isn’t that what those old nuclear plants did?

Those were fission plants splitting heavy atoms; fusion is the opposite and avoids the long‑lived waste and meltdown modes people worry about in fission.

Say the sun and trench line again because it stuck with me.

To make fusion here, we heat the fuel to about ten times the sun’s surface temperature while holding pressures higher than the deepest ocean trench.

How did you decide to go after something this hard and not give up?

I grew up wanting to solve energy, saw academic approaches missing the path to a product, pivoted into space propulsion to build fast, then brought modern power electronics, fiber optics, and rapid iteration back to fusion, and we’ve now run seven prototypes with Polaris operating and improving.

I heard Sam Altman basically hand‑recruited you and pushed YC—true?

He visited with a stack of textbooks, dug into our tech and iteration speed, then said YC would help us move faster and build a scalable business.

I told him his take—that you were just building without fancy decks—really resonated with me, does it with you?

Totally, we were building, measuring, and iterating in a small warehouse with machines in every state of assembly, and now Polaris fills a much larger facility.

Early on, what were you focused on before YC sharpened things?

It felt messy across materials, plasma, fusion, and propulsion, and YC forced the hard focus on a single product—compact generators you can ship and deploy.

Weren’t you skeptical you needed YC while building power plants?

We were, but it rewired us to cut side quests, invest in manufacturing, and ask nonstop if each action speeds deployment.

That focus lesson is huge.

Demo Day also taught us to show the vision with one striking image—the real fusion glow and the world it could unlock—then we took months to diligence and kept the cap table tight with patient partners like Mithril and Capricorn.

Did you deliberately filter out investors who might get impatient?

Yes, we chose a few who did deep technical diligence and had the patience for big builds and the move to hundreds of thousands of square feet.

So YC was worth it.

Absolutely, and we’ve kept that founder‑mode habit of getting in the weeds, hiring for the same mindset, and rejecting bureaucracy.

Did you have to grow into leadership as the company scaled?

For sure, we drifted into process at times and had to return to front‑line leadership, and while people say it doesn’t scale, it does if each area you leave behind runs in founder mode.

How often are you tempted to chase the faster, easier money?

All the time—we could sell helium‑3 for medical imaging today—but we say no to stay on the trillion‑dollar mission, and having long‑term investors gives us the room to do that.

Fusion’s been “twenty years away” forever; what makes your three‑year Microsoft timeline believable?

We broke ground and are targeting 2028 because Moore’s law, fiber optics, and distributed compute finally let us control in tens of nanoseconds, handle massive data, and switch one hundred gigawatts cleanly, and Polaris is built, fusing, and being tuned.

It sounds like you put in tons of energy to get more out—what ratio are you targeting?

We focus on efficiency, recovering around ninety‑plus percent of input electricity so fusion needs roughly one and a half to two times the input, instead of chasing hundreds of times like others.

If something slips before 2028, what’s the likeliest culprit?

Surprises are guaranteed, so we pre‑build knobs and capacity to handle a few percent efficiency shortfalls, supply or permitting delays, and we insisted on penalties because accountability matters.

Did regulators ever feel like a dead end you couldn’t route around?

We engaged early, proposed shielded vaults with no one in the room, and treat hazards like high voltage and routine x‑rays or neutrons the way hospitals manage particle accelerators, with remote handling where needed.

Here’s Paul’s question: can you build hardware as perfect as your simulations?

Holding forty‑foot machines to within one thousandth of an inch is brutal, but laser trackers turned a month of alignment into hours, and in timing the real world bites—fiber lengths, device turn‑on, and the speed of light force hardware corrections beyond the model.

Do rivals worry you?

Demand from AI and EVs is so huge that our real competition is time, though we’re thoughtful about what we publish.

If fission cleans up its image, does it compete?

Timelines rule: we won a determination of non‑significance for our Orion plant so we could move dirt immediately, and we’re pushing toward gigafactory‑style assembly and rapid deployment instead of one‑offs.

What do you do for fun when you can breathe?

Hockey dad weekends and Halloween animatronics at home, though my nine‑year‑old sometimes cringes.

It feels like we’ll look back on today as the moment before abundant power; do you feel the weight of that?

The need has exploded with EVs and data centers, so success isn’t one historic plant but a business that can ship gigawatts per day around the world.

Who actually buys your electrons?

Microsoft’s power goes to the grid via a utility and a marketer, while Nucor will take a much larger plant behind the meter, and homes will still be served by utilities as we place modular generators near load and on industrial sites to avoid new transmission.

AI companies must be calling nonstop.

They are, and Microsoft’s nuclear and power teams have tracked our progress for years as we advanced the tech.

We’re taking you up on the field trip—I can’t wait to see Polaris, and I’m so energized after this conversation.

Same—complex ideas made understandable, and now we’ve got a road episode to plan.

I can’t wait for everyone to hear this one; see you on the next episode.