Fusion Energy in 2026: Which Private Companies Are Closest, and What Does 'Closest' Actually Mean?

Fusion energy has been 30 years away for 70 years. The joke is old. But the reason it persists is instructive: the physics is genuinely hard, the engineering is genuinely expensive, and "progress" in fusion can mean very different things depending on what you measure.

Here's what the numbers actually show in 2026.

## NIF Ignition: What It Proved and What It Didn't

On December 5, 2022, the National Ignition Facility at Lawrence Livermore National Laboratory announced **ignition**: a fusion shot that produced 3.15 MJ of fusion energy from 2.05 MJ of laser energy delivered to the target. For the first time, fusion output exceeded fusion input in a controlled experiment.

The physics is real. Ignition means the plasma was hot and dense enough for fusion reactions to propagate faster than energy losses — the plasma sustained itself briefly.

What NIF ignition did **not** prove:

- **Wall-plug efficiency**: The 2.05 MJ of laser energy delivered to the target required approximately **300 MJ of electrical energy** from the grid to power the lasers. Net energy gain from the electrical grid perspective: deeply negative
- **Repetition rate**: The NIF facility fires one shot every several days to weeks. A power plant requires millions of shots per second
- **Tritium breeding**: NIF uses tritium as fuel. Tritium is rare, artificially produced in nuclear reactors, and costs approximately $30,000 per gram. A fusion power plant must breed its own tritium from lithium blankets surrounding the plasma — NIF has no such system

> ⚡ The NIF result proved the core plasma physics works. It proved almost nothing about whether fusion electricity generation is practical. Those are entirely different questions.

## ITER: The 35-Nation Machine

**ITER** (International Thermonuclear Experimental Reactor) in Cadarache, France, is the world's largest fusion experiment — a €22 billion machine involving 35 nations. Its goal: demonstrate **Q=10**, meaning 500 MW of fusion power from 50 MW of heating power.

ITER uses the **tokamak** design — a toroidal magnetic confinement chamber. The superconducting magnets (Nb₃Sn niobium-tin compound cooled to 4 K) create a magnetic cage that confines plasma at 150 million °C — ten times hotter than the sun's core.

**Timeline reality**: ITER's first plasma was originally scheduled for 2025, but faces significant delays. Current projected timeline: **first plasma in 2027 at the earliest**, with deuterium-tritium (DT) experiments producing Q=10 plasma in the mid-2030s.

The delays are engineering, not physics:
- Vacuum vessel sector manufacturing defects discovered during assembly required partial disassembly and repair
- Superconducting magnet manufacturing quality issues (primarily from European suppliers)
- COVID-19 disrupted international component delivery schedules

ITER will not generate electricity. It is an experiment. Its purpose is to validate the plasma physics at fusion-relevant scale, provide tritium breeding blanket testing data, and train the engineers and operators needed for DEMO — the demonstration power plant that follows.

## The Private Sector Landscape

The urgency of climate change and declining cost of private capital has driven unprecedented private investment in fusion since 2020. Here are the companies with credible engineering progress, not just capital raises:

**Commonwealth Fusion Systems (CFS) — SPARC Tokamak**

CFS's differentiator is **high-temperature superconducting (HTS) magnets** using REBCO (rare-earth barium copper oxide) tape. HTS magnets at 20 T (tesla) field strength are physically possible at the scales required for compact tokamaks. In September 2021, CFS demonstrated a 20 T HTS magnet — the strongest superconducting magnet of its type ever built.

Why this matters: magnetic confinement power scales as B⁴ (fourth power of field strength). Doubling field strength from 10 T to 20 T increases fusion power by a factor of 16 for the same plasma volume. This allows SPARC to be roughly **1/65th the volume** of ITER while targeting similar fusion gain.

CFS targets SPARC **first plasma in 2025–2026** (experimental device, not power producing), with DEMO (the ARC power plant) targeting first electricity in **2030**. These timelines are aggressive. The HTS magnet technology is real and validated. Whether plasma physics scales as predicted at SPARC's parameters is what first plasma will reveal.

**Helion Energy**

Helion uses a **field-reversed configuration (FRC)** — a different magnetic geometry than the tokamak. Their Polaris device (7th generation prototype) has reached plasma temperatures of approximately 100 million degrees Celsius. They have a signed power purchase agreement with **Microsoft** for delivery of fusion electricity by **2028** — the first commercial fusion contract in history.

The Microsoft contract includes a penalty structure: Helion pays Microsoft if electricity isn't delivered by 2028. This is not vaporware — it is a legally binding commercial agreement with financial consequences.

Skepticism is warranted: Helion's FRC approach has never demonstrated net energy gain. The jump from Polaris (demonstration device) to commercial generation is enormous. The 2028 date is almost certainly not achievable. But the company has $2.2 billion in funding and the most aggressive commercial timeline of any private fusion developer.

**TAE Technologies**

TAE uses a beam-driven FRC design and is funded to over $1.2 billion. Their near-term pivot: using their plasma control technology for **medical proton therapy accelerators** to generate revenue while fusion development continues. Commercial fusion target: **2030s**.

## The Q-Value Confusion That Every Fusion Article Gets Wrong

The **Q-value** in fusion physics measures plasma gain: fusion energy output divided by heating power input. ITER's Q=10 target means 10× more fusion power than heating power.

But this is **not** the Q-value that determines whether a fusion power plant makes commercial sense.

The **wall-plug efficiency** chain for a fusion power plant:

1. Electricity from grid → heating systems (neutral beam injectors, RF heating): ~50% efficiency
2. Fusion plasma → thermal energy in blanket: ~80% efficiency
3. Thermal energy → steam turbines → electricity: ~35% efficiency

Working backwards: to generate 1 GW of net electricity, you need plasma Q of roughly **30–50**, depending on specific design parameters. ITER's Q=10 target, while a genuine physics milestone, is well below the minimum for a commercial power plant.

> ⚡ ITER's Q=10 demonstrates scientific feasibility. Commercial power generation requires Q>30 minimum. No fusion device in history has achieved Q=1, let alone Q=30. The gap between ITER's target and commercial viability is still a factor of 3–5 in plasma performance.

CFS's ARC design targets Q≈25 in steady-state operation. If achieved, this would be the first device close enough to commercial viability to inform real power plant economics.

## Timeline Reality

The honest timeline for fusion electricity on the grid:

- **2025–2027**: SPARC first plasma, Helion Polaris successor testing — proof of concept devices
- **2030–2033**: First demonstration devices attempting sustained positive plasma gain (Q>1 wall-plug)
- **2035–2040**: First grid-connected demonstration plants (if 2030–2033 devices succeed)
- **2040–2050**: Commercial deployment at scale, if economics are validated

This is not a pessimistic forecast — it is what the engineering timelines support if each step succeeds on the first attempt. Major technical surprises (positive or negative) are possible.

## The Bigger Picture

Private fusion investment has done something that decades of government-funded fusion research could not: it has created genuine engineering competition with real commercial deadlines. The HTS magnet breakthrough at CFS is real. Helion's commercial contract creates accountability. The race to first plasma is being run with a different urgency than ITER.

Whether fusion makes commercial sense before the grid has already decarbonized with solar, wind, and batteries is the right question. Fusion's window of commercial opportunity is 2040–2060. Missing that window doesn't make the physics wrong — it makes the technology late. The engineering is worth watching closely, with clear eyes about what the Q-values actually mean.

Fusion Energy in 2026: Which Private Companies Are Closest, and What Does 'Closest' Actually Mean?

// COMMENTS

ON THIS PAGE