The Sun in a Bottle
The sun is a fusion reactor. Hydrogen atoms, crushed together by gravitational pressure, fuse into helium, releasing energy that warms the Earth 93 million miles away. This process has powered the sun for 4.6 billion years and will continue for another 5 billion.
For seven decades, scientists have tried to recreate this process on Earth—to build a "sun in a bottle" that could provide limitless clean energy. The fuel is abundant: deuterium from seawater and lithium for tritium breeding could power civilization for millions of years. The waste is minimal: helium and some activated materials, but no long-lived radioactive waste like fission. The safety is inherent: fusion reactions are hard to sustain; they stop if anything goes wrong.
Fusion has been "twenty years away" since the 1950s. This has become a punchline.
But something has changed. In December 2022, the National Ignition Facility achieved "ignition"—a fusion reaction that produced more energy than the lasers delivered to the fuel.¹ Private fusion companies have raised billions of dollars. Advances in superconducting magnets, plasma physics, and AI-driven control are compressing timelines.
Is fusion finally close? Or is it still perpetually twenty years away?
Meanwhile, fission—nuclear power's proven technology—struggles in Western countries but expands in Asia. New reactor designs promise to address the problems that have plagued traditional nuclear: cost, safety, waste, and proliferation.
This chapter examines both: the dream of fusion and the reality of advanced fission. Both could provide abundant, reliable, low-carbon electricity. Both face substantial obstacles. And both might finally be approaching the point where they matter.
2026 Snapshot — Nuclear Power Today
Fission: The Current Fleet
Nuclear fission provides roughly 10% of global electricity—about 2,700 terawatt-hours annually from approximately 440 operating reactors.² This makes nuclear the largest source of low-carbon electricity after hydropower.
The fleet is aging. Most reactors were built in the 1970s-80s. Average age exceeds 30 years. Extensions to 60 or even 80 years are possible but require investment.
Geography varies enormously:
- France: ~70% of electricity from nuclear, the highest in the world
- United States: ~20%, the largest fleet by capacity (94 reactors)
- China: Rapidly expanding, with 58 reactors operating and more under construction
- Germany: Shut down its last reactors in 2023
- Japan: Slowly restarting after Fukushima, with only a fraction of its fleet operating
New construction tells two stories:
- In China, Russia, South Korea, and the UAE, nuclear plants are built on schedule and (relatively) on budget
- In Western countries, recent projects have been disasters: Vogtle in Georgia took 15 years and cost over $30 billion (originally projected at $14 billion); Hinkley Point C in the UK is similarly delayed and over budget³
Fusion: The Research Frontier
Fusion remains a research endeavor, not a commercial technology. But the research landscape is more active than ever.
Public programs:
- ITER (International Thermonuclear Experimental Reactor) in France is the largest fusion project ever. Originally planned for completion in 2020, now targeting first plasma in 2035. Cost has grown from €5 billion to €20+ billion.⁴
- National Ignition Facility (US) achieved ignition in 2022 using laser-driven inertial confinement—different from the magnetic confinement approach most power plant designs use.
- JET (UK) holds records for magnetic confinement performance; it concluded operations in 2023 after decades of service.
Private companies have raised over $6 billion:⁵
- Commonwealth Fusion Systems (MIT spinout) is building SPARC, a compact tokamak using high-temperature superconducting magnets, with plans for a pilot plant (ARC) in the early 2030s
- TAE Technologies pursues an alternative approach (field-reversed configuration) and has operated for over 25 years
- Helion Energy uses pulsed fusion and claims it will have a demonstration plant by 2024 (this timeline has slipped)
- General Fusion is developing magnetized target fusion
- Various others are pursuing different approaches
The milestone to watch: Net energy gain from a fusion reactor design suitable for power generation (not just NIF's laser-based ignition). No one has achieved this yet.
Notable Players
Existing Nuclear
EDF (Électricité de France) operates the world's largest nuclear fleet and is building new plants in the UK and France.
Exelon (US) is the largest US nuclear operator, having spun off from its generation business.
Korea Hydro & Nuclear Power has built reactors on time and budget and is exporting (Barakah in UAE).
Rosatom (Russia) is a major exporter of nuclear technology to developing countries, despite geopolitical complications.
China National Nuclear Corporation and China General Nuclear are building most of the world's new reactors.
Advanced Fission
TerraPower (backed by Bill Gates) is developing the Natrium reactor—a sodium-cooled fast reactor with molten salt energy storage. A demonstration plant in Wyoming is planned.
NuScale has received NRC approval for its small modular reactor (SMR) design, though its first project (in Idaho) was cancelled due to cost increases.⁶
X-energy is developing a high-temperature gas-cooled reactor with pebble-bed fuel.
Kairos Power is building a molten salt-cooled reactor using TRISO fuel.
Oklo is pursuing a very small (1.5 MW) fast reactor.
Various national programs (China, Russia, India) are developing fast reactors, thorium reactors, and other advanced concepts.
Fusion
Commonwealth Fusion Systems is considered the most likely private company to demonstrate net energy gain from magnetic confinement fusion. Their approach uses high-temperature superconducting magnets to build compact, powerful tokamaks.
TAE Technologies has raised over $1.2 billion and has been operating longer than most competitors.
Helion Energy received a power purchase agreement from Microsoft, conditional on actually delivering power.⁷
General Fusion is backed by Jeff Bezos and is building a demonstration plant in the UK.
ITER remains the largest public investment, though delays have frustrated supporters.
The Physics of Fusion
Understanding why fusion is hard requires understanding the physics.
What Needs to Happen
Fusion occurs when atomic nuclei—typically hydrogen isotopes deuterium and tritium—collide with enough energy to overcome their mutual electrical repulsion and fuse into helium, releasing energy.
The requirements:
- High temperature: The fuel must be heated to over 100 million degrees Celsius—hotter than the core of the sun—so particles move fast enough to overcome repulsion
- High density: Enough particles must be present to collide frequently
- Long confinement: The hot, dense plasma must be held together long enough for enough reactions to occur
These three factors multiply together in the "triple product"—temperature × density × confinement time—that must exceed a threshold for net energy gain.⁸
Why It's Hard
Nothing can touch the plasma. At 100 million degrees, any material would instantly vaporize. The plasma must be confined without physical contact.
Magnetic confinement (the leading approach) uses powerful magnetic fields to hold the plasma in place. But plasmas are turbulent and unstable—they constantly try to escape confinement through various instability modes.
Inertial confinement (the NIF approach) compresses the fuel so quickly that reactions occur before the fuel can fly apart. This requires extraordinary precision in the compression drivers.
Materials must survive the harsh environment: intense neutron bombardment, high heat flux, and electromagnetic forces.
Tritium is scarce: Deuterium is abundant (seawater), but tritium is rare. Fusion plants must breed tritium from lithium using neutrons from the fusion reaction itself—a process that has never been demonstrated at scale.
Progress and Challenges
Plasma physics is now well understood. Tokamaks (donut-shaped confinement devices) reliably produce fusion reactions. The Q>1 milestone (more energy out than in) has been achieved at NIF.
The gap is between laboratory demonstrations and power plants:
- NIF achieved Q>1 for the fusion reaction, but the lasers required far more energy than the reaction produced
- Tokamaks have not achieved Q>1 (though JET came close)
- No device has demonstrated sustained energy production suitable for a power plant
High-temperature superconducting magnets may be transformative. They can produce stronger magnetic fields in smaller devices, potentially enabling compact, economical fusion reactors. Commonwealth Fusion Systems is betting its future on this technology.⁹
Fusion: Paths to Power
Magnetic Confinement
The leading approach uses magnetic fields to confine plasma, typically in a tokamak (toroidal shape) or stellarator (twisted toroidal shape).
The tokamak path:
- ITER demonstrates Q=10 (10x more energy out than in) around 2035
- DEMO (a follow-on project) demonstrates electricity generation around 2050
- Commercial plants follow
This timeline is slow—too slow for fusion to help with near-term climate goals.
The private accelerated path:
- Commonwealth Fusion Systems builds SPARC (2025-2027), demonstrating Q>2
- ARC pilot plant demonstrates net electricity (early 2030s)
- Commercial plants under construction by 2035
If this works, fusion could be producing commercial electricity by the late 2030s—ambitious but not absurd.
Challenges:
- High-temperature superconducting magnets must perform as hoped at scale
- Plasma control and materials must work in a power plant environment
- Tritium breeding must be demonstrated
- Costs must be competitive with other low-carbon sources
Alternative Approaches
Stellarators use a twisted magnetic geometry that may offer better plasma stability but is harder to engineer. Wendelstein 7-X in Germany is the leading stellarator experiment.
Field-reversed configurations (TAE Technologies) use a different magnetic geometry that may allow smaller, simpler devices.
Magnetized target fusion (General Fusion) uses mechanical compression rather than pure magnetic confinement.
Inertial confinement (laser-driven, as at NIF) is primarily funded for weapons research but could theoretically lead to power plants.
Aneutronic fusion (using fuels like proton-boron) would avoid the neutron problem entirely but requires much higher temperatures and has worse energy economics.
Trajectory and Uncertainty
The optimistic case: Private companies demonstrate net energy by the late 2020s. Pilot plants produce electricity by the early 2030s. Commercial plants begin construction by mid-2030s. Fusion provides meaningful electricity by 2040.
The pessimistic case: Key challenges (materials, tritium breeding, cost) prove harder than hoped. Timelines slip further. Fusion remains "twenty years away" for another generation.
The realistic assessment: Fusion is closer than ever, but commercial viability is not guaranteed. Even in optimistic scenarios, fusion arrives after solar and storage have already transformed the grid. Fusion may be less a replacement for fossil fuels than a complement to renewables—providing reliable baseload that supports a variable renewable majority.
Advanced Fission: New Approaches to an Old Technology
While fusion remains a research effort, advanced fission offers improvements to a proven technology. Several designs address the problems that have plagued traditional nuclear.
Small Modular Reactors (SMRs)
The concept: Instead of giant 1+ gigawatt plants built on-site over a decade, build smaller reactors (50-300 MW) in factories and ship them to sites.
Potential advantages:
- Lower capital cost per project (though possibly higher cost per kWh)
- Faster construction (factory vs. site-built)
- Better suited to smaller grids
- Learning from serial production
Reality check: No SMRs are yet operating commercially. NuScale's design is NRC-approved but its first project was cancelled when costs escalated beyond what customers would pay. The "factory production" model hasn't been demonstrated—economies come from volume that doesn't yet exist.¹⁰
Trajectory: Plausible for SMRs to begin operating by 2030. Whether they'll be cost-competitive with large reactors or renewables remains unproven.
Fast Reactors
Traditional reactors use slow (thermal) neutrons. Fast reactors use fast neutrons, which enables:
- Burning plutonium and other actinides from spent fuel
- Breeding new fuel from abundant uranium-238
- Potentially "closing the fuel cycle" to reduce waste
Status: Russia operates fast reactors (BN-600, BN-800). China is building them. Western programs have struggled—the US abandoned its fast reactor program decades ago.
TerraPower's Natrium is a sodium-cooled fast reactor with integrated thermal storage. A demonstration plant in Wyoming is targeted for the early 2030s.
Trajectory: Plausible for fast reactors to demonstrate advantages in specific applications (waste burning, fuel breeding). Whether they'll be cost-competitive for general electricity generation is uncertain.
High-Temperature Gas Reactors
High-temperature gas-cooled reactors (HTGRs) use helium coolant and graphite moderator, operating at higher temperatures than traditional water-cooled reactors.
Advantages:
- Higher efficiency (40%+ vs. 33% for water-cooled)
- High-temperature heat for industrial processes
- Inherent safety (the reactor slows down naturally if it overheats)
TRISO fuel: Tiny uranium particles coated with ceramic layers, extremely robust against melting.
X-energy's Xe-100 and Kairos Power's KP-FHR use TRISO fuel in different reactor configurations.
Trajectory: Plausible for HTGRs to demonstrate advantages for industrial heat applications. Commercial deployment likely in the 2030s if current projects succeed.
Molten Salt Reactors
Molten salt reactors (MSRs) dissolve fuel in molten salt, which serves as both fuel and coolant.
Potential advantages:
- Very high temperatures for efficiency and industrial heat
- Inherent safety (molten salt doesn't pressurize like water)
- Potential for using thorium fuel cycle
- Continuous fuel processing and waste removal
Challenges: Molten salts are corrosive. Materials and systems for handling them require development.
Status: Research-stage. No commercial designs are close to deployment.
Trajectory: Wild for MSRs to be commercially significant within a decade. Longer-term potential if materials challenges are solved.
The Western Nuclear Problem
The fundamental problem with nuclear in Western countries is not physics but economics and institutions:
Costs have escalated: Vogtle Units 3 and 4 in Georgia cost approximately $35 billion for 2.2 GW—roughly $16,000/kW. Solar plus storage can be built for $2,000-3,000/kW. Even accounting for capacity factors, nuclear is far more expensive in Western contexts.¹¹
Why costs escalated:
- Loss of construction experience (the US built no new reactors for 30 years)
- Regulatory requirements that increased over time
- First-of-a-kind designs with no learning curve benefits
- Supply chain atrophy
- Project management failures
The contrast with Asia: Korea builds reactors for $3,000-4,000/kW. China builds them for even less. The difference is not physics but institutional capacity.
The question: Can Western countries rebuild nuclear construction capability? Or has that capability been permanently lost to countries that maintained it?
The "Always-On" Value
Solar and wind are intermittent: they produce power when conditions are favorable, not necessarily when power is needed. Batteries help bridge gaps but become expensive for extended periods. This creates value for "firm" or "dispatchable" power sources that can run whenever needed.
What "Always-On" Provides
Reliability: Some applications require power continuously: hospitals, data centers, industrial processes. While batteries can provide backup for hours, some users want sources that don't depend on weather.
Seasonal backup: In northern latitudes, solar output drops dramatically in winter—precisely when heating demand peaks. Batteries economically cover hours, not months.
Grid stability: Spinning generators provide inertia and voltage support. Renewables require additional equipment to replicate these services.
Industrial process heat: Some industries need high-temperature heat that's hard to electrify. Nuclear (especially high-temperature reactors) or hydrogen (produced via nuclear) could serve these needs.
The Competition
The value of "always-on" power depends on what it costs and what alternatives exist:
Natural gas is currently the main competitor for dispatchable power. Gas plants are cheap to build, flexible, and can respond quickly. Their costs depend on fuel prices, which are volatile. Their emissions make them problematic for decarbonization.
Long-duration storage could provide always-on capability from renewables. If iron-air batteries or hydrogen storage become cheap enough, solar plus storage could serve the same role as nuclear.
Overbuilt renewables could provide "always-on" power through sheer redundancy: build enough solar and wind that even in poor conditions, sufficient power is available.
Demand flexibility could reduce the need for always-on supply: shift demand to match supply rather than vice versa.
The Role for Nuclear
If nuclear can be built cheaply, it provides reliable, low-carbon power that complements variable renewables. If costs remain high, solar and storage will dominate, with nuclear relegated to specific niches or countries with particular conditions.
The scenarios:
- Nuclear renaissance: SMRs or advanced reactors achieve cost-competitiveness. Nuclear provides 20-30% of electricity in a deeply decarbonized grid.
- Niche role: Nuclear remains expensive. Existing plants operate until retirement. New construction limited to specific contexts (countries with strong nuclear programs, applications requiring firm power).
- Fusion changes everything: Fusion achieves commercial viability. It eventually displaces both fission and fossil fuels, though this is unlikely before 2050.
Second-Order Effects
Energy Abundance
Both fusion and cheap advanced fission could create a world of energy abundance—electricity so cheap and plentiful that currently uneconomic applications become viable:
Desalination: Unlimited freshwater from seawater
Carbon capture: Affordable removal of CO2 from the atmosphere
Synthetic fuels: Producing hydrocarbons from air and water
Space propulsion: Powering energy-intensive launch systems
Manufacturing: Energy-intensive processes becoming cheap
This is speculative but represents the upside case for abundant nuclear energy.
Proliferation and Security
Nuclear technology carries proliferation risks. Fission involves materials and expertise applicable to weapons. Even fusion facilities might produce materials with weapons applications.
SMR proliferation concerns: More reactors in more locations means more potential diversion pathways.
Fast reactor concerns: Fast reactors can breed weapons-usable plutonium.
Security: Smaller, more distributed reactors might be harder to secure physically.
International frameworks (IAEA, NPT) must evolve to address new reactor types and deployment patterns.
Waste Management
Fission waste remains politically contentious even though technical solutions exist. The US still lacks a permanent repository (Yucca Mountain was cancelled). Finland's Onkalo is the first deep geological repository under construction.¹²
Advanced reactors might reduce waste:
- Fast reactors can burn actinides that would otherwise be long-lived waste
- Some designs produce less waste per unit energy
- But no reactor eliminates waste entirely
Fusion waste is more benign—primarily activated structural materials with half-lives of decades rather than millennia. This is a genuine advantage if fusion becomes practical.
Risks and Guardrails
Safety
Nuclear safety has improved dramatically. Modern reactors have passive safety features that prevent meltdowns without operator action. The probability of a severe accident from a new reactor is far lower than from older designs.
But public perception remains shaped by Chernobyl and Fukushima. A single severe accident could set back the entire industry, regardless of the technical safety of new designs.
AI and automation in nuclear operations could improve safety (better anomaly detection, faster response) or create new risks (cybersecurity, autonomous systems making critical decisions).
Cost Overruns
Western nuclear projects have consistently exceeded budgets, often dramatically. Until a project is completed on time and on budget, skepticism is warranted.
Guardrails: Fixed-price contracts, experienced construction teams, standardized designs, realistic contingencies.
Political Risk
Nuclear projects take a decade or more to build. Political changes can alter policy mid-project—as happened with Germany's nuclear phaseout.
Long-term policy frameworks that survive political transitions are essential for nuclear investment.
Fusion Hype
Fusion has a history of overpromising. Private companies face pressure to show progress and attract investment. Claims should be evaluated skeptically.
The key milestones:
- Net energy gain from a power-plant-relevant design (not yet achieved)
- Sustained operation (not just brief pulses)
- Demonstrated tritium breeding
- Cost projections based on actual construction, not optimistic estimates
Until these milestones are reached, fusion timelines should be treated as uncertain.
The Path Forward
Near-term likely (5-7 years):
- Existing nuclear fleet continues operating with license extensions
- China and Russia continue building new conventional reactors
- SMR demonstration projects advance but don't reach commercial operation
- Fusion companies demonstrate improved performance; none achieve sustained net energy
Plausible (7-15 years):
- First SMRs operate commercially (NuScale, TerraPower, X-energy, or competitors)
- Fusion achieves sustained net energy in at least one approach
- Nuclear provides 10-15% of global electricity (modest growth from current levels)
- Advanced reactors find niches in industrial heat and hydrogen production
Wild (speculative):
- Fusion achieves commercial viability, beginning displacement of fission and fossil fuels
- Energy becomes effectively unlimited and very cheap
- Nuclear (fission or fusion) provides majority of global electricity
- New applications (space propulsion, transmutation of waste) become practical
Nuclear power—whether fission or fusion—remains one of the few proven paths to reliable, low-carbon electricity at scale. Whether it achieves its potential depends on solving problems that have proven stubbornly difficult: cost, construction capability, and public acceptance. The next decade will reveal whether this time is different.
Endnotes — Chapter 11
- Lawrence Livermore National Laboratory announced on December 13, 2022, that NIF achieved fusion ignition—a fusion reaction producing more energy than the laser energy delivered to the fuel target.
- World Nuclear Association tracks global reactor statistics. Approximately 440 reactors in 32 countries produce about 10% of global electricity.
- Georgia Power's Vogtle Units 3 and 4 were originally projected to cost $14 billion and enter service in 2016-2017. Actual costs exceeded $30 billion with operation beginning in 2023-2024.
- ITER Organization provides official project status. First plasma has been delayed from original 2020 target to 2035, with costs increasing from €5 billion to over €20 billion.
- Fusion Industry Association tracks private fusion investment, which exceeded $6 billion as of 2024.
- NuScale announced in November 2023 that its Carbon Free Power Project in Idaho was cancelled due to cost increases.
- Helion Energy announced a PPA with Microsoft in May 2023, with delivery contingent on Helion achieving commercial operation.
- The Lawson criterion defines the triple product threshold for fusion energy gain.
- Commonwealth Fusion Systems demonstrated a 20-tesla high-temperature superconducting magnet in 2021, validating a key technology for their SPARC tokamak.
- SMR cost and schedule projections remain largely theoretical. The NuScale project cancellation highlighted the gap between projections and reality.
- Nuclear cost comparisons vary by metric and location. Direct cost comparisons with solar+storage are complicated by different capacity factors and grid services provided.
- Finland's Onkalo repository is the first deep geological repository for spent nuclear fuel under construction, expected to begin operation in the 2020s.