To Bet on Fusion or Fission?

Managing Editor’s Note: Before we get to today’s AMA, I just want to quickly remind you that Jeff’s “Project MAFA” briefing is coming up…

Next Wednesday, July 16, at 8 p.m. ET, he’s revealing the details of President Trump’s new “MAFA” initiative that Jeff believes could kick off a “Golden Century” for America.

The Trump administration, Wall Street, and Silicon Valley are all pushing it forward. That’s why Jeff is hosting this urgent briefing revealing the details – and the name of a stock he’s watching that may soar as “Project MAFA” takes root – next Wednesday evening.

Go here to automatically add your name to the guest list, then read on for today’s AMA…

This week was definitely one for the history books.

On Wednesday, NVIDIA became the world’s first $4 trillion company.  And as I write this morning, it continues to trade above that remarkable valuation.

It’s such a remarkable story, considering that the company was worth just $8.4 billion when I was pounding the table for people to buy it back in the first quarter of 2016.  And now, this fiscal year, NVIDIA is not only worth more than $4 trillion, it will also generate more than $100 billion in free cash flow.

It is one of the greatest companies of all time.

And it’s also a flagship signal for the entire artificial intelligence (AI) industry.  It’s pretty obvious that things haven’t been slowing down.  They’ve been speeding up.

And there is so much more growth ahead of us.

Have a wonderful weekend,

Jeff

Google’s “Bet” on Fusion

Jeff & Co,

Although my knowledge of Commonwealth is mostly limited to what you guys have written here (and in the past). I think fusion is the future, esp. given that all these data centers coming along are so power hungry.

Any chance you guys can also opine on why Google chose Commonwealth vs. any other provider?

For instance, TAE I had thought offered the “cleanest” overall platform using the most abundant “fuel” available. So I assume it’s their overall design that Google wasn’t interested in, or maybe at the end of the day, it was less viable or somehow less scalable?

Just curious if you have any additional insights in this regard?

Thanks.

– Graham L.

I have been an enthusiastic advocate for nuclear energy since I saw documentaries on the subject in the late 20-teens highlighting the work of Alvin Weinberg and his team at Oak Ridge National Laboratory.

As such, I’ve been optimistic that small modular reactors would soon (3-5 years) become the primary provider of electricity for our AI-led future. However, after reading the July 7 Bleeding Edge article, Google’s Secretive Investment in Fusion, it appears fusion may be progressing so quickly that it passes SMRs as the go-to future energy source to the point that 4th-generation nuclear companies such as Nuscale and Oklo become irrelevant.

I would like to hear Jeff’s in-depth perspective on this topic.

– Thom M.

Hi Graham, Thom,

Thanks for writing in, gentlemen. For readers who missed it, Graham and Thom are referencing The Bleeding Edge – Google’s Secretive Investment in Fusion from earlier this week.

In this issue, I dug into the recent deal between Google and Massachusetts-based nuclear fusion company, Commonwealth Fusion Systems (CFS). As I wrote…

[The] financial details of the deal weren’t provided, but my best guess is that this is at least a $500 million deal and likely a $1 billion+ transaction (in terms of how much Google invested in Commonwealth)…

What has been made public, however, is a fantastic development:

• Google has invested in Commonwealth Fusion Systems at an unknown valuation.
• Google has agreed to a power purchase agreement (PPA) for at least 200 megawatts of power from Commonwealth’s first full-scale ARC nuclear fusion power plant to be built in Chesterfield County, Virginia.
• Google has the option, perhaps even the first right of refusal, to purchase power for future Commonwealth nuclear fusion plants.

Naturally, the fact that one of the largest data center companies in the world, Google, has both invested in Commonwealth and entered into a power purchase agreement in advance is clearly an indication of Google’s belief in Commonwealth and its approach to nuclear fusion.

But Graham, that’s not Google’s only “bet” on nuclear fusion.  Alphabet (Google) has directly invested in TAE Technologies in 2016, 2021, and most recently this June in a follow-on round where TAE raised $150 million.

As with Commonwealth, this is a clear indication that Alphabet sees a potential path towards commercialization for TAE Technologies and its nuclear fusion technology… As do I.

TAE’s approach is different from others as it uses a field-reversed configuration (FRC) approach to its nuclear fusion reactor.  The FRC approach is interesting because it uses both magnets and the fusion plasma’s own magnetic field to control the fusion reaction.

Historically, without advanced technology, it was impossible to maintain a stable plasma with this approach.  But advanced magnets and TAE’s unique approach to beam neutral atoms precisely into the magnetic field enable TAE to maintain a stable plasma.

Also worth noting, to your point, is that TAE uses hydrogen-boron as a fuel (p-B11) for both its abundance and non-radioactivity.  That enables TAE to produce clean energy with no radioactive waste, whereas most nuclear fusion reactors do produce small amounts of short-lived radioactive waste, which is very easy to manage safely, unlike nuclear fission reactors.

If you’d like to dig a bit deeper on TAE Technologies, you can check out my issue of Outer Limits – The Cleanest Approach to Nuclear Fusion.

Which brings us to Thom’s question: Will nuclear fusion win out over the fourth-generation small modular reactors (SMRs) that use nuclear fission?  This is an exciting topic.

Until very recently, politics have intentionally held back the entire nuclear industry and the employment of very safe fourth-generation nuclear fission reactors (SMRs) through regulations designed to make it almost impossible to innovate and advance clean energy using SMRs or other forms of nuclear fusion.

At the same time, technology was advancing quickly with regard to nuclear fusion technology.

Magnet technology was improving, as were the fusion reactor designs, and the advancements in artificial intelligence as they apply to maintaining a stable plasma were a critical missing piece needed for commercialization.

During the last few months, with the new administration, economic and regulatory support for nuclear energy has flipped 180 degrees to a positive, proactively supportive stance.  This swing to support nuclear energy has resulted in a confluence of incredible potential for both advanced nuclear fusion designs and nuclear fission designs.

And when institutional capital sees both economic policy and regulatory policy that support the nuclear energy sector, it invests heavily in the space.  And that’s precisely what we are seeing now.

I can’t imagine a better time.  All of the best potential designs can be pursued for both SMRs and nuclear fusion reactors.  Several of them will be proven to be both safe and effective methods of producing grid-scale clean energy.

So, regarding your hypothetical, Thom, it’s actually simple… it’ll come down to economics.

The nuclear energy technology that will win large shares of the marketplace for clean energy production will be reactor designs that are most efficient at producing clean energy.

This will be a balance between the capital expenditures required to build a reactor facility and the operational costs required to maintain a facility.

The framework that most people familiar with the industry will state correctly that SMRs will be more expensive to build, but there is very little technical risk, as nuclear fission has been around for decades.

Nuclear fusion designs, however, will likely be much cheaper to build, with an even smaller footprint, but have high technological risk.  People say this because there are currently no nuclear fusion reactors providing energy to a grid.

My perspective is that this has become a very close race, and nuclear fusion companies are innovating and building faster than the fourth-gen nuclear fission companies.

It’s exciting because we’re going to see this once-in-a-lifetime event happen where several companies will be commissioning their new nuclear energy plants feeding clean energy to the grid, all within the next few years.

And we’ll be able to see which technologies are the most economical for commercial clean electricity production.

And once that has been determined, these new reactors will be scaled as quickly as possible around the world.

The Magnet Market

Jeff Brown,

Could you highlight what companies are manufacturing “magnets”, being used in fusion, vs. EVs, versus robots? Would these companies be a good investment as they could ramp with Optimus and fusion?

 – Heather M.

Hi Heather,

I love the way you are thinking.

Naturally, the market for the massive magnets used in nuclear fusion is very nascent as all companies in the sector are working on custom-designed prototype reactors.

Commonwealth actually assembles its own magnets. But it procures the key component, rare-earth barium copper oxide tape (REBCO), which its magnets are made of.

I know that Commonwealth has been working with Faraday Factory, a Japan-based company, to acquire the magnetic tape, as well as others that are prominent in the industry.   Fujikura (Japan), SuperPower (a division of Furukawa in Japan), and THEVA (Germany).

Needless to say, this is a highly specialized industry that requires extremely high-quality manufacturing skills. So, it’s no surprise to see Japan so prominently represented, as well as one German company.

It’s important to note that the kinds of magnets used for nuclear fusion will be very different than those used in humanoid robots for actuators.  And it’s also worth mentioning that the humanoid robot market is also nascent.

A number of companies in China manufacture magnets for the legacy robotics industry.  There is also a congregation of magnet suppliers in Japan, like Proterial, Shin-Etsu Chemical, and TDK Corporation.

But the reality is that the robotics industry is about to change radically.  At a high level, the shift is from pre-programmed robots for specific applications to general-purpose intelligent agentic robots.

I fully expect to see new entrants and changes in the supply chain for magnets for these new intelligent robotics applications, specifically for the humanoid robots.

And yes, there are going to be some fantastic investment opportunities in this area in the months and years ahead.

CoreWeave Acquisition

What are your thoughts regarding CRWV, Applied Digital, and CORZ?

– Mel J.

Hi Mel,

Near Future Report senior analyst Nick Rokke actually wrote about CoreWeave (CRWV) and Core Scientific (CORZ) earlier this week in The Bleeding Edge – Did CoreWeave Acquire a Gold Mine?

For those who missed it… in a $9 billion all-stock agreement, AI infrastructure company CoreWeave has acquired Core Scientific – a Bitcoin miner with a somewhat significant number of fully operational data centers wired to the grid and ready to run.

That part of the deal is significant because if you miss that, you might think this was some sort of crypto investment… but no, this is all about CoreWeave’s strategic move to acquire what all artificial intelligence (AI) companies are hungry for – power.

Here’s Nick…

Hyperscalers like Microsoft, Google, and Amazon are spending hundreds of billions of dollars this year building out new AI-scale data centers. But there’s a catch…

Even if they break ground today, most utilities are sitting on multi-year backlogs – some as long as four years – just to get power hooked up.

That’s why CoreWeave’s move makes so much sense. The bottleneck in AI compute now isn’t really GPUs. Production has drastically ramped up over the past couple of years. But the bottleneck now is the electrical infrastructure to power these GPUs.

This merger is a power grab. And it’s just more proof that the AI data center boom is just getting started.

If you missed the issue, I highly recommend giving it a read. Nick dives into CoreWeave’s history with Core Scientific and the significance of this deal during this time. It isn’t just CoreWeave… every AI-focused company is solving for scale and working towards ensuring sufficient AI infrastructure that’s wired, powered, and ready to deploy.

And as I’ve written extensively in The Bleeding Edge, when companies can’t get enough power from the grid, they simply haul in generators powered by natural gas or propane to produce the needed energy as a short-term solution.

This is exactly what xAI has done in South Memphis, where it has built its data center that’s making so many breakthroughs with Grok – its frontier AI model.

As for Applied Digital (APLD), it has a background similar to Core Scientific’s.  It was nothing but a shell company back in 2020.  In April 2021, it changed its name to Applied Blockchain, Inc., raised capital, and became a crypto mining company.

But like so many other crypto mining companies suffering from the anti-crypto regulatory stance of the last four years, it pivoted quickly to use its data centers and know-how to provide cloud services to be used for artificial intelligence.

Applied Blockchain officially changed its name to Applied Digital in November 2022 to focus on high-performance computing services, primarily targeted at artificial intelligence.

Announced just last month, Applied Digital will be providing CoreWeave 250 megawatts of computational horsepower to support CoreWeave’s artificial intelligence services.  Basically, CoreWeave is leasing computational horsepower, and thus energy, from Applied Digital.

I haven’t done a full analysis on Applied Digital because a few things jumped out at me that are of concern.  Applied Digital only has $68 million in cash and almost $1 billion in debt.  It is deeply unprofitable and will have a negative $604 million in free cash flow this fiscal year.

It is in a very capital-intensive business, and it has almost no capital.  It must raise a large amount of cash, quickly, or it will go bankrupt.  And despite these dire realities, it is currently trading at 13.26 times current fiscal year sales.

I can only assume that the management team at Applied Digital is scrambling right now trying to raise at least $1 billion to fund the next 12 months of operations.

That’s all for this week’s AMA. We’ll be back at it next Friday with a new round of reader questions, so feel free to write to us right here with any questions or comments.

Have a great weekend.

Regards,

Jeff