From Fusion to Octopart and Back

We see plasma every day. It's the stuff glowing in fluorescent light bulbs, the flame of a candle, and the material of stars. Fusion happens when you leave a hot and dense plasma to cook for a long time. Protons knock into each other, overcoming their repulsion and fuse into helium nuclei, releasing energy.

Fusion research. That was my day job back in May of 2006 when we wrote the first lines of Octopart code. I was in physics grad school building a new experiment to study an unconventional scheme for confining fusion plasma.

[caption id="attachment_2226" align="alignnone" width="400"] Me and my unfinished fusion research machine to merge spheromaks into a field reversed configuration, 2006[/caption]

One afternoon, struggling to find an op-amp for a high frequency amplifier, I got a call from my friend Andres. He was frustrated that he couldn’t find a low temperature capacitor for his research at the South Pole. “We should fix electronic parts search,” he said.

Octopart started that spring as a side project for us. We spent nights and weekends figuring out how to connect our Python code to a web server and how to write database schemas. By the end of the summer, we managed to get everything running on a desktop computer I bought at a yard sale for $50. We pointed the DNS entry of octopart.com to my cable modem’s IP address and octopart.com was live.

Fusion powers stars. Gravity pulls gas clouds into a dense ball which heats up over millions of years. Once the density and temperature get high enough, the fusion reactions take off and the star ignites, releasing energy in the form of heat and light. Harnessing fusion on earth would give us a virtually limitless amount of cheap energy with little or no waste products. This kind of research - building a synthetic star - has been going on for about 60 years, making steady progress.

By the end of the summer I started to hear rumors that governmental funding for my fusion experiment was in jeopardy. I missed my long-distance girlfriend, and increasingly I realized that my odds of landing a job as an assistant professor after graduating were small. I thought quitting grad school meant severing my connection to the dream of fusion. I was wrong, but I wouldn’t know it for eight years.

One morning in early summer I told my advisor that I was leaving the program. A few months later I moved away to hack on Octopart full time. Around the same time Andres and I applied to a (then) small program called Y Combinator. Over the previous few years I had read all of Paul Graham’s essays. “How to Start a Startup” felt like he was writing a letter to me, the struggling physics grad student.

Mainstream fusion research centers around tokamaks, elaborate donut shaped machines which compress and heat a plasma in the shape of a ring. An alternative scheme called a Field Reversed Configuration doesn’t need a complicated, donut shaped vessel. It can exist in a simple cylindrical can. While a tokamak can sustain plasmas for many seconds or minutes, FRCs only live for a fraction of a second and are pulsed, requiring advanced electronics.

For the next few years we pushed Octopart through a recession plus all the normal startup hurdles. By 2010 we had established ourself within the electronics industry and figured out how to make money. Fusion faded from my view. I’d occasionally comment on fusion related articles on Hacker News and read articles about fusion in the media.

In the summer of 2014, I saw the headline: “Y Combinator and Mithril Invest in Helion, a Nuclear Fusion Startup." The article’s photograph showed a long can, most definitely not a tokamak.

[caption id="attachment_2227" align="alignnone" width="500"] Helion Energy's Prototype Fusion Machine[/caption]

After some furious clicking I landed on the about page of Helion Energy. I recognized the chief scientist, John Slough - I had read his papers on field reversed configurations when I was in grad school. Amazingly, Y Combinator was funding a fusion company closely related to my mothballed experiment.

A month later I met David Kirtley, the CEO of Helion at YC alumni demo day. I don’t think he was expecting to field technical questions about plasma configurations that day from YC alumni. Nonetheless we had a friendly conversation about some of their machine’s details. I learned that they were using state of the art high voltage solid state switches to pulse their machine. When I told him I co-founded Octopart his eyes lit up. “We use Octopart all the time,” he said.

I smiled. “That’s awesome,” I said.

Follow @swurzel