Perfecting Imperfect Fusion with Thea Energy’s Brian Berzin

Brian Berzin is the CEO and cofounder of Thea Energy, a fusion energy company developing stellarators using programmable planar coil magnet arrays for commercial power generation.

What led you to founding Thea Energy?

My path to founding Thea Energy started long before I was thinking of fusion systems.  I studied electrical engineering at Lehigh University and picked up a finance degree along the way. During college, I took on a small project and founded a company around it. That was my first dive into building not just technical things but companies themselves. After college, I moved to New York City where I had stints in venture capital and helped set up a multifamily office. Those experiences were invaluable because I got to operate alongside founders and invest in companies, giving me 10 to 50 data points rather than just one.

In 2017, I switched from the finance side to the startup side and joined one of the only fusion companies around at that time. There were really only three fusion companies in North America back then. I worked there from 2017 through 2020, applying the best practices and toolboxes I’d learned from VC and the startup landscape to a fusion startup. That’s where I really caught the fusion bug.

In 2022, I founded Thea Energy with my cofounder David Gates. The New York area has some great fusion programs. Columbia University has an excellent program, and the legendary Princeton Plasma Physics Laboratory is a worldwide leader in fusion research.  David and I had known each other for years through this tight tri-state network, and we started discussing what was then an ARPA-E funded project. Through that project, Thea Energy was born with a theoretical breakthrough that allowed us to look at stellarators without the complicated wiggly hardware.

How does Thea Energy’s approach differ from other fusion companies?

What makes our stellarator approach unique is the 70 years of research that happened before we even founded the company. The stellarator and the tokamak are the two best-studied forms of magnetic confinement fusion. Over those decades, there have been large integrated systems built that demonstrate physics performance in real environments and give us computational tools that are benchmarked against actual system performance.

Where I see challenges with some newer approaches is in proving out new scientific risks. I’m fortunate that I don’t have to sit in front of VCs and explain how long fundamental scientific research will take. That work has already been done for stellarators and tokamaks. It took 70 years. Even with great advancements in computation and technology, I’d bet that new fundamental scientific breakthroughs for exotic fusion approaches won’t just pop up because some VC decides to fund them.

Our focus at Thea Energy is on commercialization. We are taking proven physics and figuring out cheaper, simpler, scalable ways to build the technology. Many companies are working to achieve Q greater than 1, which is a big milestone. But making fusion energy for a fraction of a second isn’t a power plant. A power plant is a real asset that operates for 40 years with a high capacity factor and good economics. That’s what Thea Energy is building toward.

What’s the difference between tokamaks and stellarators?

Both were invented in 1951. The stellarator was actually invented at Princeton Plasma Physics Laboratory, which is pretty cool since we spun Thea Energy out of Princeton 70 years later. Both use toroidally confined plasmas, which are donut-shaped plasmas confined by encircling coils. The main difference is how you create the poloidal field twist needed for a magnetostatic plasma.

The tokamak uses a center solenoid to drive current and energy into the plasma to create that twist. Every slice of a tokamak is symmetrical, making the donut easier to construct, but you have to dump tremendous energy into the plasma to make it work. The stellarator actually twists the entire system. It has a very happy plasma configuration in steady state and without disruptions.

The downside has always been the hardware complexity. If you’ve seen pictures of stellarators, you’ll instantly understand. The French cruller-shaped device requires incredible precision in bending and shaping. That’s why virtually every stellarator in fusion research history has been behind schedule and gone substantially over budget. That’s exactly what we’re fixing at Thea Energy.

How did recent technological advances enable your approach?

Several breakthroughs converged over the past few years. First, high-temperature superconductors are game-changers for magnetic confinement fusion. They allow us to build smaller devices that generate more fusion power, fundamentally changing the economics. Even our approach using hundreds of superconducting magnets in an array wouldn’t have been possible since individual magnets would have been too large with older technology.

Second, computational advances have been crucial. The Simons Foundation ran a project over the past five years that substantially expanded the stellarator design universe. We can now crunch numbers on hundreds of thousands of stellar plasma equilibria. Ten years ago, analyzing one configuration took a month and practically melted a computer. This expanded design space lets us find plasma physics solutions that work with simpler devices.

Finally, modern control systems make our approach practical. We control 450 independent variables, which we can now do with only fancy desktop computers. It doesn’t even require supercomputers. We’re also applying AI and machine learning, and using reinforcement learning models that make our systems more efficient and allow us to tune out defects. This added practicality is completely new to fusion.

How does your planar coil array technology work in practice?

We’re designing our first integrated system right now, and we’ve already proven substantial parts of the componentry through prototyping and R&D. Our arrays of superconducting magnets create really precise magnetic fields. The next step is building this into an integrated device using many of the same technologies that W7-X and tokamaks utilize such as vacuum systems, cryogenic systems, and other standard components.

What’s revolutionary is how we handle imperfection. For the past six months, we’ve been purposefully testing our system’s limits. We’ve swapped in defective hardware, mismounted magnets on purpose, and simulated wear and tear that would happen over 40 years of operation. Through our control system—an array of sensors, feedback loops, and dynamic tuning of every single magnet—we can tune out a ridiculous amount of these imperfections. The hardware and magnetic fields operate at near perfection even with manufacturing defects or degradation over time.

This is the second light bulb moment we had as a company. It’s how the world works in every other industry. Radar operates with phased arrays, modern electronics and computer screens work this way. We’re applying this proven approach to fusion for the first time, and it fundamentally changes how you can construct and manufacture power plants because everything doesn’t have to be perfect anymore.

What makes this approach particularly suited for commercial power generation?

We’re building our first device, called Eos, to make fusion energy using our planar coil stellarator architecture. But here’s the key difference. We’re not doing five-second pulses. We’re targeting 24-hour to 48-hour fusion run at steady state. That’s the game changer and validation of an approach that scales linearly to a power plant.

The platform also enables continuous improvement over time. Take turbulence suppression, for example. The world will be better at modeling and suppressing turbulence in five or ten years, and AI is already doing interesting work here. Every other fusion architecture would need to redesign their device to incorporate those advances. We can just push a software update to a device that’s been running for 10 years and get better performance. It’s like Tesla cars getting longer range through software updates years after purchase. This software layer touching all the hardware represents a paradigm shift for fusion.

What’s next for Thea Energy?

We’ve hit major milestones over the past six months with array performance, control systems, and our ability to tune out variability and defects. We’re nearing 70 people and will be doubling the team. We’ve built a strong group of engineers and experts from commercial industries who think about things as products. Much of our engineering is similar to aerospace, so we’re excited to leverage talent from industries focused on commercializing products, whether rockets or power plants.

We’re also scaling up manufacturing significantly. We’ve pulled together a great team of manufacturing engineers from Tesla, Apple, and SpaceX. Over the next four to five years, we’ll be building that integrated system to show the world we have a power plant-relevant architecture, not just a cool science project. And we’re thrilled to be doing this in the New York area. The talent and drive of the team we’ve built here has been fundamental to our execution.

SPONSORED

HAUS specializes in public relations and creative services for deep tech startups.

Learn more