The Cross-Section Between Fusion and AI

Apr 12, 2024
The Cross-Section Between Fusion and AI

We received quite a few questions this week about nuclear fusion, as well as the cross-section between nuclear fusion and artificial intelligence. I love all the positive energy.

I have some heavy travel coming up over the next two weeks, so I’m afraid that I won’t be able to keep to a daily publishing schedule of my own research in Outer Limits during that time.

But we do plan on publishing with the help of some of my teammates — well-known in the Brownstone/Brownridge universe. We are excited to welcome them back.

Lindsey and I will make sure that you’ll know when we have a guest analyst in any issue of Outer Limits.

I hope you’ll enjoy some added variety at Brownridge Research.

Have a wonderful weekend.


How Far Away is Fusion?

I was wondering if the nuclear fusion reactors will use uranium, as well. Also would they use more or less uranium compared to the fission reactors, if so? Lastly, how far out do you think we are from a fusion company looking for investors? Thanks. — Cy C.

Hello Cy,

Nuclear fusion reactors will not use uranium, or any other fissile material like plutonium. Nuclear fission involves the splitting of a heavy nucleus into two lighter nuclei. Nuclear fusion combines two lighter nuclei together, releasing energy in the process.

Nuclear fusion reactors will most likely use one of the following three fuel sources:

  • Deuterium – tritium (D-T)
  • Deuterium – Helium 3 (D-He-3)
  • Hydrogen – boron (p-B11)

As for the investment side of nuclear fusion companies, we have finally entered the growth phase of nuclear fusion in the private sector. Major funding rounds for nuclear fusion-related companies are happening almost every month now.

The significant breakthroughs that have happened over the last two years have been the stimulus to create interest for investment from institutional capital (i.e. private equity, venture capital, and strategic investors like energy companies).

The other major shift that has occurred in the last 12 months has been a realization that other forms of what is considered to be “clean” energy are not — in fact — clean at all, as they require immense carbon resources to mine and build, have limited lifetime, and are toxic and unrecyclable.

Those who are genuinely serious about clean energy and reducing environmental impact have come to the realization that nuclear fusion (and fission) is the only possible solution for utility scale clean energy. 

This shift has been beneficial, as it has focused more investment capital on exciting nuclear fusion companies.

Most fusion companies are focused heavily on institutional capital due to the large capital expenditures required for research and development in fusion. But I do hope to work with nuclear fusion companies in the context of raising capital through crowdfunding regulations, so that all my readers can participate and contribute to a future full of clean energy.

Speeding Up Fusion

Can AI not speed up the discovery process on Fusion? — Terence J.

Hi Terence,

Yes, absolutely! In fact, I believe that artificial intelligence (AI) is the key accelerant for nuclear fusion.

Arguably the single most difficult technical problem to solve in a fusion reaction is how to maintain and control the stability of a nuclear fusion plasma. A fusion reactor that can maintain a stable plasma — and thus continuously affect fusion reactions — can provide utility scale clean energy.

If a plasma cannot be maintained, then the fusion reactions can only run for very short periods of time — like seconds or milliseconds. That’s fine for research and development, but useless for powering an electricity grid.

This is where AI comes into play. Powerful magnets are used to contain nuclear fusion plasmas. They basically maintain the shape of a plasma and keep it from touching the walls of the nuclear fusion chamber. This is critical because the plasma is at temperatures at — or above — the temperature of the sun. If a plasma touched the walls of the chamber, the chamber walls would melt.

In order to maintain a stable plasma, complex magnetic fields need to be managed in real-time every millisecond. This is something that an engineer or team of engineers simply cannot do manually. This is where AI is so useful. 

Forms of AI are particularly good at optimization problems. In a nuclear fusion reaction, the AI is optimizing for a stable plasma by controlling magnetic fields.

Other areas where AI is coming in handy is in materials design, which includes design of magnets. This is another area for optimization, specifically the materials and the shape of the magnets, thus the shape of the internal fusion chamber.

AI is already having an impact on the nuclear fusion industry, and we’re going to see some AI-related fusion breakthroughs later this year.

Are AI and Quantum Computing Compatible?

Jeff, I've been a reader of your newsletters and research the first before Brownstone and have always found your research, thoughts, and comments both interesting and insightful. Thanks so much. All of the recent discussion about AI is extremely interesting. A few years ago the discussion was about Quantum Computers. Hence my question, are AI and quantum computing compatible with each other or does the way a quantum computer process data too different from the processing of training AI or any other machine learning? Thanks for your work and keep up the great newsletter. — BJ

Hello BJ,

Yes, these two incredible technologies are tightly linked.

Quantum computing is best utilized for highly complex problems that are either unsolvable using classical computing systems… or would simply take too long to solve (e.g. decades).

Complex problems like these tend to benefit from the use of machine learning, and other forms of artificial intelligence (AI). Which is why it makes so much sense for quantum computers to run AI software.

Alphabet (Google) actually has a division called QuantumAI. And in 2022, a new early stage company was spun out of that division called SandboxAQ. It is backed by Eric Schmidt, former Chairman and CEO of Google, as well as the venture capital arm of the CIA, In-Q-Tel. This is definitely a company to watch.

In the short-term, the biggest problem to solve in quantum computing is the noise in quantum computing systems. 

Noise results in errors, which either need to be eliminated or corrected. Once addressed, the industry can progress towards a universal, fault-tolerant quantum computer which… will be able to run AI-software and solve problems that were previously unsolvable.

Clarifications on Ember?

Jeff, your explanation does not make sense to me. We invested in Ember in the hope that if it was successful, we'd receive equity. However, you are defining preferred stock to be equity and it is not. Equity means some small percentage of ownership. Preferred stock bequeaths us no ownership whatsoever. It is normally issued at $25/share and remains around that price. Most folks would define equity as common stock shares and that's certainly what I was hoping for from Ember. I didn't see anything in your explanation that says why we should be happy with preferred shares. What did I miss? — Bob K.
Jeff, thanks for this update [on Ember Fund]. What has me hesitating is that the SAFE is converting to preferred stock, rather than common stock. This, it would seem to me, would greatly limit our upside, is that not the case? Thanks. — Doug
Thanks Jeff! I was very against agreeing to this as were pretty much all of the comments on the Republic update page. As it mentioned that Ember can lower our equity amount that we invested, essentially a down round (at least that's what I read from it). Thanks for your feedback :). I'm wondering about your thoughts on Skybound's new capital raise? Thanks as always. — Mike S.

Thank you Bob, Doug, and Mike for writing in concerning Ember.

Preferred stock is equity, just like common stock is equity. The difference is that preferred stock comes with some benefit(s) that common stock does not have. 

The most common benefit conferred by preferred stock is that they provide a dividend preference over common stock.

This structure is most common in public companies that have both preferred and common stock structures. To continue with this example, if there is enough capital in a company to issue a dividend, preferred stockholders would have the preference of that dividend before common stockholders.

In the case of Ember Fund, I do know that preferred stockholders have a liquidation preference over common stockholders. What this means is that in the event of a sale (acquisition) of the company, preferred stockholders receive a certain amount before common stockholders receive anything. This is advantageous for preferred stockholders.

To be clear, I do not know the specific terms of the liquidation preference for preferred stockholders, as the strategic investment has not yet been completed with Ember Fund.

As for your question, Mike, regarding a possible down round, it is always possible for any private company to have a down round (i.e. a round where new investment capital is raised at a valuation that is lower than the previous round). Valuations are impacted by both company performance/growth, and by the general state of the economic environment (i.e. health of the economy, interest rates, inflation).

Fortunately, Ember Fund has been on an accelerated growth trajectory, and the technology markets, including crypto, have been on fire. Hopefully, things will continue to move in that direction… and the next funding round will happen at a higher valuation, or an acquisition occurs at a higher valuation.

One final comment — I don’t have all the details on Skybound’s new capital raise just yet, but I’ll try to dig into that in the next few weeks to better understand the offering and provide some of my thoughts.

Thanks for a great week! We always welcome your feedback. We read every email and address the most common comments and questions in the Friday AMA. Please write to us here.

Previous Post Next Post