Nuclear Fusion: Why 2024 Was Different, and What Still Has to Work

## The Standard Joke

"Fusion power has been 30 years away for the last 70 years." In December 2022, the National Ignition Facility (NIF) achieved ignition — producing more energy from fusion reactions than the laser energy delivered to the fuel. This happened again in 2023, with higher yields.

This was not electricity-on-the-grid fusion. But it was a genuine scientific milestone that hadn't been crossed before.

## What Fusion Is and Why It's Hard

Fusion powers stars. Two light atomic nuclei — deuterium and tritium — collide at high enough speed to overcome electrostatic repulsion, fuse, and release energy.

Energy per reaction: enormous relative to fuel mass. Deuterium comes from seawater. Tritium can be bred from lithium. No CO₂. Waste products are helium and fast neutrons.

The problem: temperatures of around 100 million degrees Celsius are required. Containing plasma at those temperatures is the central engineering challenge.

## Two Main Approaches

**Magnetic Confinement (Tokamak)**: A donut-shaped magnetic cage confines plasma so it doesn't touch any surface. ITER, being built in France with 35 nations contributing, is the largest tokamak ever. Designed to produce 10x more energy than it consumes (Q=10). First plasma ~2025-2026; full D-T experiments ~2035.

**Inertial Confinement (NIF)**: Lasers compress a tiny fuel pellet so rapidly that fusion reactions occur before the plasma expands. NIF's 2.05 megajoules of laser energy produced 3.15 megajoules of fusion energy — net gain at the fuel level. But the lasers consumed ~300 megajoules of electricity. Overall efficiency remains deeply in the red.

## Private Sector Acceleration

40+ private fusion companies, over $6 billion in private capital (2023):

- **Commonwealth Fusion Systems (MIT spinout)**: High-temperature superconducting magnets enabling smaller, cheaper tokamaks. SPARC device targeting Q > 2; commercial plant (ARC) in the 2030s.
- **Helion Energy**: Targets direct electricity extraction from fusion. Microsoft has a power purchase agreement contingent on delivery.

## What Still Has to Work

**Tritium breeding**: Reactors would breed their own tritium by bombarding lithium with neutrons. Works in principle; not demonstrated at scale.

**Materials under neutron bombardment**: Fast neutrons damage structural materials over time. Long-duration materials are a significant open problem.

**Wall-plug efficiency**: The total system needs Q > 10 at the system level to be economically viable.

## Timeline

The most optimistic credible timelines (CFS, Helion) suggest demonstration plants in the 2030s. "30 years away" may finally be inaccurate — in the right direction.

Nuclear Fusion: Why 2024 Was Different, and What Still Has to Work

// COMMENTS

ON THIS PAGE