← All scenarios

Spaghetti Scenarios · Growth trajectory

Large-scale operational nuclear fusion

Humanity manages to generate electricity from nuclear fusion, the reaction that makes the Sun shine, in power plants that run commercially and at large scale. It would be a decisive step towards a Type 1 civilization, one that harnesses all the energy of its planet.

Fusion already releases more energy than its fuel receives. But no machine turns it into electricity yet.

Gain of the best experiment

4.13 times

the laser energy: 8.6 MJ of fusion from 2.08 MJ on the fuel (April 2025).

▲ 11 ignitions since the first, in 2022

Source: NIF ↗

Longest plasma

22 min

an experimental machine in France held its plasma (February 2025), a world record. In China, another sustained its high-performance mode for 1,066 s.

▲ Two world records in 2025

Sources: CEA ↗ · Hefei Institutes of Physical Science ↗

Cumulative private investment

US$14.24bn

raised by fusion companies since 2021; US$4.48bn in the last year alone (to July 2026).

▲ Seven times more than in 2021

Source: Fusion Industry Association ↗

Schedule of the big international experiment

2039

is when ITER will start operating with its final fuel (deuterium and tritium). The 2016 plan said 2035.

▼ 4 years later and €5bn more expensive

Sources: ITER Organization 1 ↗ · ITER Organization 2 ↗

More energy out than in

Fusion energy released in the best shots of the world's largest laser, in the United States, compared with the energy the laser delivers to the fuel. Since 2022 fusion has exceeded the input.

Laser energy on the fuel2.05 MJ
Dec 2022 · first ignition3.15 MJ
Feb 20245.2 MJ
Apr 2025 · record8.6 MJ
Jun 2026 · most recent7.9 MJ

To fire, the facility draws about 300 to 400 MJ of electricity from the grid: a power plant would need a gain of around 100.

Sources: NIF ↗ · Lawrence Livermore National Laboratory 2026 ↗ · Lawrence Livermore National Laboratory 2023 ↗

Private money for fusion

Cumulative private investment in fusion companies, according to the report the industry association publishes every July. Move the slider to see each year.

048121620212022202320242025202614.2
  • Cumulative private investment14.2US$bn in 2026

For 2021, 2022 and 2024 the source gives 'almost' or 'over' that figure. In the last year (to July 2026) 56 companies raised US$4.48bn, a record.

Sources: Fusion Industry Association 2021 ↗ · Fusion Industry Association 2022 ↗ · Fusion Industry Association 2023 ↗ · Fusion Industry Association 2024 ↗ · Fusion Industry Association 2025 ↗ · Fusion Industry Association 2026 ↗

What is missing for a power plant

How many times the input energy is multiplied. The white line marks the next threshold; the end of the bar, what a plant would need.

Laser: record gain (2025)4.13
A test plant with no net electricity demand would need a gain approaching 15; a power plant, around 100.
Tokamak: best Q achieved (1997)0.67
ITER is designed for Q = 10: 500 MW of fusion from 50 MW of heating.

Sources: Lawrence Livermore National Laboratory 2023 ↗ · Lawrence Livermore National Laboratory 2025 ↗ · ITER Organization 2026 ↗ · ITER Organization 2024 ↗

How long the plasma lasts

A power plant would have to run non-stop; experiments are still measured in minutes.

China · 2023 (high-performance mode)403 s
China · Jan 2025 (high-performance mode)1,066 s
France · Feb 2025 (world record)1,337 s · 22 min

The French one was a hydrogen plasma at about 50 million degrees; the Chinese one, in its high-performance mode. They are not directly comparable.

Sources: Hefei Institutes of Physical Science ↗ · CEA 1 ↗ · CEA 2 ↗

The road to the first power plant

  1. 2022First ignition: the US laser obtains 3.15 MJ of fusion from 2.05 MJ on the fuel.
  2. 2023JET shuts down after its final record: 69 MJ in 5 seconds with 0.2 mg of fuel.
  3. 2025Duration records in China (1,066 s) and France (1,337 s). The laser reaches 8.6 MJ, 4.13 times the energy it receives.
  4. 2026ITER installs 6 of the 9 modules of its vacuum vessel. Fusion companies raise US$4.48bn in a year, a record.
  5. Today · October 2026
  6. 2027The SPARC machine (US) aims to produce more energy than it consumes. China plans to complete its BEST machine.planned
  7. 2028Helion plans to deliver energy from its fusion plant to Microsoft.planned
  8. Early 2030sARC power plant in Virginia (US): about 400 MW of electricity to the grid.planned
  9. 2034 · 2039ITER starts research operation (2034) and, in 2039, operation with deuterium and tritium, the fuel of a power plant.planned
  10. 2040United Kingdom: first operations of the STEP prototype. A prototype power plant (DEMO) would be built by then.planned
  11. 2nd half of the centuryElectricity generation from fusion, according to the International Atomic Energy Agency.planned

Sources: NIF ↗ · GOV.UK 2024 ↗ · Hefei Institutes of Physical Science ↗ · CEA ↗ · ITER Organization 2026 ↗ · Fusion Industry Association ↗ · Commonwealth Fusion Systems 2026 ↗ · Chinese Academy of Sciences ↗ · Helion Energy ↗ · Commonwealth Fusion Systems 2024 ↗ · ITER Organization 2024 ↗ · GOV.UK 2025 ↗ · IAEA ↗

Large-scale operational nuclear fusion

Probability
Possible
Timeframe
50 years
Confidence
Medium for the scientific progress; low for the timeframe
Classification
Technological

The values from the original scenario table are kept; what that table does not define is proposed by the research, because the scenario does not have its own profile yet.

Signals to watch

Proposed by the research: the profile does not define its signals yet.

  • Gain of the best experiment
  • Longest plasma
  • Cumulative private investment
  • Schedule of the big international experiment

Possible human action

  • Sustained public and private funding
  • International cooperation on the big experiments
  • Solve the engineering: gains close to 100 and continuous operation
  • First pilot plants connected to the grid