Aug 8, 2025 · 49m · sourcery
Nuclear Race to Power Superintelligence · Sourcery with Molly O'Shea
Isaiah Taylor · 16m spoken Packy McCormick · 11m spoken Molly O'Shea · 9m spoken JC Btaiche · 8m spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Hosted by Molly O'Shea at the Re-Industrialize conference, this episode of Sourcery features deep-tech founders Isaiah Taylor and JC Btaiche alongside investor Packy McCormick to examine how modular nuclear energy and fusion power are critical to solving AI compute energy bottlenecks and winning the global technological race against China.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Molly holds 21.2% of the talking time here. How this is scored →
Molly as informed peer 3.4 Guest teaching 4.9 Guest disagreement 1.8 Molly pushing back 1.4
speaking balance: gold is Molly, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 17:52 Isaiah Taylor rejects investor narrative on AI and heavy industry
Taylor directly challenges mainstream tech investor assumptions, arguing that nuclear is a physical hedge rather than something AI will rapidly automate.
Hardest push from Molly ▶ 21:40 Molly presses Taylor on Kalshi's low licensing probabilityMolly challenges Taylor with Kalshi prediction market data giving only a 13% chance of US reactor licensing, forcing him to address the specific regulatory distinctions between NRC and DOE.
Biggest teaching moment ▶ 13:30 Isaiah Taylor breaks down total US nuclear waste scaleTaylor educates the host on nuclear waste facts, contrasting 70 years of total US waste fitting in an Olympic pool with 600,000 sedan-sized carbon blocks emitted annually by fossil fuels.
Molly holds their own ▶ 7:07 Molly benchmarks China's 7-to-14 year nuclear head startMolly demonstrates domain knowledge by citing specific industry figures on China's massive multi-year lead in reactor construction.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Molly as informed peer | Guest teaching | Guest disagreement | Molly pushing back | Why |
|---|---|---|---|---|---|---|
| Show Title Card and Guest Intros | 3 | 4 | 1 | 1 | Molly introduces Isaiah Taylor and asks foundational questions about Valor Atomics and the state of American nuclear. Taylor explains the 46-year industry pause since Three Mile Island and the need for modular manufacturing. | |
| Regulatory Obstacles and Executive Order Opportunities | 3 | 5 | 1 | 1 | Molly inquires about regulatory bottlenecks and recent executive orders. Taylor educates on how regulatory R&D cycles stifle innovation and outlines the DOE's prime contractor testing mechanism. | |
| Geopolitical Nuclear Race: US vs. China | 5 | 4 | 2 | 2 | Molly brings informed data noting China is 7 to 14 years ahead in nuclear deployment. Taylor agrees and elaborates on China's repetitive reactor construction and experimental testing, while Molly questions if the US can compete alone. | |
| Sponsor Segment: Brex Financial Stack | 3 | 6 | 2 | 1 | After the Brex sponsor read, Molly asks about geopolitical funding influencing public sentiment and echo-chamber misconceptions. Taylor details European investigations into Russian gas lobbying and breaks down the physical math of nuclear waste volumes. | |
| Sponsor Segment: Turing Intelligence | 3 | 5 | 2 | 1 | Following the Turing sponsor read, Molly asks about physical and cyber defense for reactor sites. Taylor reframes the threat, noting China already leads in IP, and explains the passive safety of Generation IV reactors. | |
| AI Compute Demand and Nuclear Investment Thesis | 5 | 4 | 3 | 3 | Molly pushes Taylor on AGI impacts, industry trade-offs, and Kalshi prediction markets showing 13% odds for new reactor licenses. Taylor counters market assumptions, distinguishing NRC from DOE authorization pathways. | |
| Interview with JC Btaiche & Packy McCormick: Fuse's Mission | 3 | 5 | 1 | 1 | Molly welcomes JC Btaiche and Packy McCormick to discuss Fuse. JC announces their Los Alamos partnership and Nature publications, while Packy explains Fuse's unique near-term radiation revenue model. | |
| Vertical Integration & Powering AI Compute | 3 | 6 | 2 | 1 | Molly asks about Fuse's role in re-industrialization and White House engagement. JC explains the decay of the NIF supply chain, in-house manufacturing of 75% of Titan's 40,000 parts, and the severe $60B/gigawatt compute bottleneck. | |
| Winning the Fusion Race & The Great Differentiation | 3 | 5 | 2 | 2 | Molly asks if the US has been left in the dust by China. JC rejects defeatism by arguing fusion's radical energy density leapfrogs incremental capacity, while Packy emphasizes America's edge in technical talent. |
Statements from this episode (23)
Taylor: Nuclear Industry Has Failed to Mass-Manufacture Reactors in Series
“Nuclear in the last 40 to 50 years has been sort of a one-off industry. You have a plant built here, a plant built there. There were a couple of models and units of reactors built in duplicates around the world. But nobody has really been able to manufacture l…”
Taylor: Valar Testing Prototype in LA, Targets Criticality Next Year
“We have our first, what's called a non-nuclear reactor prototype in LA. We're doing temperature testing on it, pressure testing on it, and we're looking at criticality next year”
Taylor: US nuclear energy has been paused since Three Mile Island
“Nuclear in the United States has been paused for about 46 years. It's been paused essentially since Three Mile Island.”
Taylor: US excels at modular manufacturing over large-scale construction
“We don't do as well at like very large scale manufacturing or very large scale construction. We're much better at modular manufacturing.”
Valar Atomics aims to build thousands of domestic modular reactors
“So that's really what we focus on is building a modular reactor that can be manufactured here and that we build thousands of throughout the United States.”
Taylor: Manufacturing will shift from labor-heavy to energy-heavy over 100 years
“If you think about what's happening over the next hundred years in manufacturing and reindustrialization, we're essentially taking what was an extremely labor heavy industry into being an energy heavy industry, right?”
Taylor: China is commercializing US nuclear reactor designs from the 1970s and 1980s
“They're also turning on innovative types of reactors that were invented in the United States in the seventies and eighties. And they sort of sat on the shelf and we never really tackled them. We ran into a bunch of regulatory roadblocks and then we never turne…”
Taylor: Nuclear is the best climate solution, outperforming solar and wind
“Nuclear is so obviously, so clearly the best climate solution that you could find, right? So if you care about the fact that we're putting more CO two in the, you know, in the atmosphere every single year, and you look at all the options, you know, options tha…”
Taylor: Russia funded German environmental groups to shut down nuclear reactors
“This was sort of a theory for a long time, but it's actually proven now that this did happen in Europe over the last 10 years. So Russian groups funded far left environmental groups in Germany to shut the reactors down in order, you know, to make it so that Ge…”
Taylor: China Does Not Want Valar's IP Because Its Technology Is Ahead
“There's a certain extent to which China doesn't actually want my IP because they have something better already running in the mainland.”
Taylor: Generation IV Nuclear Reactors Cannot Melt Down
“Generation four reactors incredibly passively safe. You basically can't make them melt down this. It's, you know, a pure physics issue. So that's the type of reactor that we build.”
Taylor: Nuclear and Heavy Construction Are Insulated from AGI Disruption
“I actually think that if you're looking for an investment that is protected from AGI, it's something like this, where it's extremely physical. You're bending steel, you're stacking blocks, you're building plants, you're laying concrete. It's going to be a long…”
Taylor: Valar Atomics' Primary Target Customers Are AI Data Centers
“Our customers are going to be data centers, right? They're going to be people who want to actually power the AI.”
Taylor: Nuclear Is a Derivative Bet on AI Growth
“Nuclear is sort of like the derivative to AI. If you believe that AI is going to go vertical, you should probably invest in nuclear, you know, because A, you're going to power it. B, you're going to be very insulated from, you know, the deflationary effects of…”
Taylor: Valar Atomics will seek DOE authorization to split atom next year
“We're going for DOE authorization next year and split the first atom.”
Taylor: Nuclear energy will still only be 8% to 9% of global consumption by 2030
“It'll, it'll still be around like eight or nine percent nuclear. Solar is going to tick up a little bit, but honestly, oil and gas is growing as well.”
Btaiche: Fuse built world's highest pulse power system and published in Nature
“We've built the words highest pulse power system ever called Titan. We fired that for the first time in March of this year and we live streamed it and Publish the results in Nature.”
Btaiche: Fuse partnered with Los Alamos and published pulsed neutron paper
“We also built, and a couple weeks ago, published in Nature another paper on the most efficient neutron delivery system and pulsed neutron system that we also published in Nature, and more timely, just yesterday on July 16th, on the 80th anniversary of the Trin…”
McCormick: Fuse is already generating revenue by selling radiation
“The big difference I would say with Fuse and anybody else is they're already making revenue and that they're not going straight shot for commercial fusion power, but they're selling radiation kind of on the way up towards or down towards, depending on how you …”
Btaiche: Fuse Manufactures 75% of Titan's 40,000 Components In-House
“Titan has 40,000 parts. We make roughly 75% of them in house, and we're continuing to increase that percentage because otherwise we have no way to get to the economic price point.”
McCormick: Fuse recruited the former number two Iranian nuclear scientist
“The number of two, kind of, nuclear scientists in, in Iran who came over here to, kind of, work on, on our side here, and then just, like, everybody that I met who was, like, you know, working on the machines, like top to bottom, just like an unbelievably cool…”
Packy: Deep tech execution inherently creates uncopyable marketing content
“For these companies, there's just, like, actually an opportunity to just show off what you're doing, and, like, that in itself is differentiated because it's so hard to do. That like just actually getting the thing done, like kind of creates all of this unbeli…”
McCormick: Claude's Writing Feedback Is Bad, Regresses Drafts Toward Mediocrity
“I will take my, if I'm, like, stuck in writing an essay, I'll, like, take the draft, I'll put it in Claude and ask for feedback, and, like, the feedback is bad, and it always, like, tries to move me more towards average, and I ignore it, but, like, I just have…”