This episode follows Isaiah Taylor, founder of Valar Atomics, at their first reactor test site in Utah where a bare patch of dirt in September became a 25 MW advanced reactor ready for criticality by July 4th — the first U.S. advanced reactor to make power, built in 9 months for under $100M total raise, demonstrating a SpaceX-style approach to nuclear manufacturing.
Manufacturing nuclear reactors is a SpaceX-style problem
Valar treats reactor manufacturing like Falcon 9 production: a complex vehicle built repeatably, not mass-produced like Teslas.
The reactor size (25 MW) was discovered through manufacturing constraints — using existing supply chains, tooling, and methods — not predetermined.
Scaling from 1 reactor to 10 requires iterative building, not paper design
The only path to hundreds of reactors is turning one on, learning, turning another on, then two, then three — uncovering unknowable supply chain, integration, and operational realities.
Industry defaults to paper iteration because nuclear operations (fueling, criticality, regulatory agreement, security, MC&A) are genuinely hard, and design feels safer.
Historical context: the 1960s–70s EPC model worked because AEC tested 70 prototypes and Navy ran many more; today that test infrastructure and workforce are gone.
Contact with reality reveals unknowable problems
Modular shielding: team thought it impossible, tried pour-in-place (added 2 months of rebar/formwork/pouring), returned to modular and solved 16,000 hard problems — blocks stack without grout or bolts, placed by crane.
“Steel is cheaper than software engineers”: analysis teams cost more than buying/welding steel for real-world tests; often 5–10× cheaper to build and learn.
Nuclear’s “idiot index” (final cost vs. commodity material cost) is in the hundreds; good industrial products are ~10. This reveals an industry formatted with misaligned incentives — too many parties benefit from cost growth.
Vertical integration across the entire stack is essential
Valar owns design, engineering, manufacturing, construction, operations, regulatory engagement, community engagement, and talent — optimizing for cheapest energy, not component margins.
Control rod drive units: vendors quoted 18–36 months; Valar built in-house, ran 40+ major iterations in a helium pressure test stand at nuclear temps/pressures.
Community engagement: team knows city council, neighbors, business leaders by name; hosts cookouts; honestly discusses failure risk — paid dividends in Utah.
Becoming a nuclear company means splitting atoms, not writing papers
Valar forbade calling itself a “nuclear company” until first criticality; before that, it’s a paper company.
“Facility” not “factory” until at least two identical reactors produced — prevents self-deception.
First reactor sized for transportability (C-17 compatible), ease of build, extreme safety, and scalable architecture (pure graphite core, TRISO fuel, helium coolant).
Obsessing over critical path drives speed
Custom software tracks critical path in real time; phrase “critical path” repeated constantly; CEO thinks about it morning/night.
Early assumption: construction was the lead. Reality: building integration (reactor-to-plant connections) and regulatory unknowns (even DOE didn’t know the path) were harder.
Critical path changes weekly; team’s ability to focus on it daily enables velocity.
Nuclear reactors are mechanically simple; complexity is in coordination
Reactor physics simpler than Raptor engine (300 bar) or high-performance V8: pressure tank + graphite moderator + uranium + working fluid. Reactors self-regulate — thermal feedback matches power to heat removal.
Real difficulty: getting 10 things right simultaneously (regulatory, construction, schedule, supply chain, operations, community, talent, safety analysis, security, MC&A) in an industry that forgot how to move fast.
Speed enables safety: faster iteration deploys safer designs (this reactor is orders of magnitude safer than LWRs) sooner.
Designing a Toyota Camry (simple, manufacturable) not a Ferrari (complex, high-performance)
Trade efficiency, power density, size for simplicity and safety — the only path to factory production and $1,000/kW (vs. $7,000–15,000 today).
Factory setting enables continuous “why do we do this?” questioning by fresh eyes; traditional 10-year builds lose that learning to retirement.
Mass replication, vertical integration, and daily improvement drive cost down.
Running through one-way doors with high risk tolerance
Bezos framework: sprint through two-way doors, deliberate on one-way doors. Valar also sprints through one-way doors (e.g., $40M site construction before certainty).
Reading regulations to fundamentals reveals false constraints — many “requirements” don’t exist.
High capital-risk tolerance is a secret weapon; competitors won’t replicate it overnight.
Pulling rabbits out of hats requires a relentless team
C-17 transport demo: 2 days before flight, base loader truck couldn’t handle vessel weight. Steel team designed/fabricated a custom loader truck in 48 hours — specialty rollers, pallet interfaces, motor integration, all-nighters.
Relentlessness is Valar’s most incomparable asset; team runs toward impossible deadlines.
Licensing gap discovered in standup: pulled everyone (accounting, photographer, non-critical staff) into a trailer war room for 4 days (Friday–Tuesday) until resolved.
Culture: hate generalities; demand precise answers or “I don’t know” + joint root-cause chase. “Nothing takes more than 15 minutes — it’s just whose 15 minutes.”
Willingness to do unreasonable things (all-hands war rooms, custom hardware in days) fixes irreducible problems.
Injecting urgency like Elon Musk
Screen shows critical path in big red letters; culture of holistic thinking — everyone asks “what must be true for 24 reactors/day?” and works backward.
Hiring for pace: people join Valar to run fast surrounded by fast movers. Pace is the hardest culture element to change; easier to start a new company than fix a slow one.
Relentless pursuit of 10x cheaper energy guides pivots
North star never changed: make energy 10× cheaper. Philippines plan → Utah pivot in one day when DOE EO14301 offered July 4 deadline.
Team rolls with conditions because goal is hardware experience fast. Will continue pivoting as conditions change.
Evolution: unlocking latent speed in nuclear talent
Many nuclear professionals want to go fast but lacked avenue; Valar unlocks them. Deep industry formatting problems, but huge motivated talent pool.
CEO’s growth: deeper respect for team’s endurance, ability to rethink completely each morning; theory about industry structure confirmed.
Safety from physics, not engineering: no meltdown possible, no runaway reaction.
Reactivity control: U-238 Doppler broadening + graphite thermal conductivity = strongly negative thermal feedback (hotter → less reactive). Control rods for shutdown only; gravity-fed fail-safe.
Decay heat meltdown prevention: graphite absorbs decay heat via high thermal inertia (sublimates at 3,000°C); TRISO fuel survives 2,000°C+; small cylinder = high surface-area-to-volume → passive convection/radiation cools core.
Validated in full-scale thermal test (LA): ran at full temp 1 week, shut off all safety systems — core stabilized at safe equilibrium, cooled over 2 days.
Learning from SpaceX: optimize every choice for scale
“If you prick Valar anywhere, we bleed scale.” Extreme safety enables scale (simpler ops, broader supply chain, simpler manufacturing).
TRISO expensive now but will scale with volume; low power density enables thousands of units. Belief in million-reactor future drives current tradeoffs.
Scale intuition is poor; CEO thinks about millionth reactor daily.
High tolerance for looking dumb unlocks truth
Willing to be the idiot in any room (policy, engineering, construction, White House). 50/50 actually wrong vs. right but early. Painful but superpower — accesses root truth others avoid.
Learned from Palmer Luckey: don’t let hit pieces slide; every CEO is at war, most are in denial. Palmer defended Valar early against “high school dropout” attacks.
Clock time is irreducible; start clocks early in parallel
Two time types: fungible (engineer-hours, hire more) vs. clock time (heat treat, licensing, concrete cure — stops for no one). Money doesn’t compress clock time.
Valar ticking clocks experts said impossible (site prep, licensing, fabrication); resolving in Valar’s favor.
Parallelism forced by clock irreducibility; art to starting clocks without losing focus.
Predicting unknown bottlenecks via end-state thinking + paranoia
High org pace reveals workarounds for apparent bottlenecks. Think backward from million reactors: what must be true? (e.g., wiring under floor now → modular crates for future).
Paranoia: CEO loses sleep over unknown bottlenecks. Painful lessons from first site = company moat (thousands of learnings, not all shareable).
Scaling is organic, continuous process — not a phase
“We’re scaling right now.” Tick rate shrinks: months → weeks → days → hours → minutes. Scaling is organizational DNA, not a design-then-execute phase.
Valar built for this from day one; if you aren’t building reactor after reactor, you haven’t started scaling.
July 4th criticality = rebirth of U.S. nuclear
250th anniversary + EO14301 deadline. Real hardware splitting atoms matters infinitely more than best paper design.
U.S. back in business of splitting atoms; bottleneck on million-reactor future and 10× cheaper energy removed.
