Claim Check: General Fusion’s LM26 compressional-heating result

Verdict: Incremental progress, but still a sub-keV plasma. LM26 showed compression and electron heating, but nowhere near Lawson-relevant performance.

Read my How General Fusion’s reactor will work (or won’t) article for background on what General Fusion is trying to do and how it has performed to date.

What happened

On June 22, 2026, General Fusion announced that its LM26 magnetized target fusion machine had achieved compressional plasma heating.

The pre-peer-review technical paper presents results from the first 11 LM26 compression shots. The best shots showed more than a 3× rise in electron temperature, about a 10× rise in density, and about a 10× rise in poloidal magnetic field during roughly 3× radial compression.

Thomson scattering, the gold standard for electron-temperature measurements, was used, with the best results on shot LMC-11: 0.72 keV; another independent measurement (UV photodiodes) gave 0.67 keV. The ion-temperature result was lower. Neutron-inferred deuterium ion temperatures during compression were about 0.41 keV for LMC-11, with a modest rise near the end of the compression.

What is actually new

This is the first detailed public data from LM26 lithium-liner compression shots: measured liner compression, magnetic-field amplification, density rise, electron-temperature rise, neutron and gamma signals, MHD behavior, and transport modeling. It’s a well-diagnosed and analyzed set of experiments.

The data support the narrow claim that compressional heating occurred, especially for electrons. Importantly, they also show that LM26 came far short of reaching fusion-relevant ion temperatures.

The claim beneath the claim

The claim beneath the claim is that LM26 has validated the central Magnetized Target Fusion bet: form a magnetized spherical tokamak plasma at moderate temperature, mechanically compress it fast enough with a lithium liner, conserve particles and magnetic flux well enough, and let adiabatic compression carry the plasma to fusion-relevant conditions.

If compression is self-similar and particle-conserving, then density should scale like inverse volume:

\[\frac{n}{n_0} \approx \frac{V_0}{V} \approx C_R^3\]

where $C_R$ is the radial-equivalent compression ratio (i.e., the ratio of the initial plasma radius to the final plasma radius).

For plasma heated adiabatically (without energy loss),

\[\frac{T}{T_0} \approx \left(\frac{V_0}{V}\right)^{2/3} \approx C_R^2\]

For a successful adiabatic compression, the 3× radial compression General Fusion achieved in its best LM26 experiments should ideally give:

\[\frac{n}{n_0} \approx 3^3 = 27\]

and

\[\frac{T}{T_0} \approx 3^2 = 9.\]

General Fusion demonstrated about a 10× rise in density, 3× rise in electron temperature, and practically no rise in ion temperature. This is far short of adiabatic compression.

What would have to be true

Achieving closer to adiabatic compression requires some combination of better confinement, higher starting temperature, larger compression, and faster compression. The paper points to higher magnetic field and improved shaping, including negative triangularity, as routes to lower ion heat transport and better performance.

Ion heating must become much closer to adiabatic

The paper’s ion-temperature model has the form

\[\frac{T_i(t)}{T_i(0)}= \left[\frac{V(0)}{V(t)}\right]^{2/3} \exp\left[ -\hat{\chi}_i \int_0^t k(t')^2\,dt' \right].\]

The first term is the desired adiabatic heating. The exponential term is the loss penalty from ion heat transport. The paper finds that an effective ion heat transport of $\hat{\chi}_i \simeq 4\ \mathrm{m^2/s}$ is enough to prevent significant ion heating, while $\hat{\chi}_i \simeq 2\ \mathrm{m^2/s}$ would have produced a stronger ion-temperature rise. In plain English: the plasma leaked heat too fast.

The compression ratio must be judged against the starting temperature

For ideal adiabatic heating,

\[C_R = \sqrt{\frac{T_f}{T_0}}\]

and

\[\frac{V_0}{V} = \left(\frac{T_f}{T_0}\right)^{3/2}.\]

Using the reported starting electron temperatures, LM26 had sufficient compression to achieve 1 keV if it were adiabatic:

\[C_R \approx \sqrt{\frac{1}{0.21}} \approx 2.2\]

or about 10× volume compression.

Reaching 10 keV is much harder

From a 0.21 keV starting plasma, adiabatic heating to 10 keV requires

\[C_R \approx \sqrt{\frac{10}{0.21}} \approx 6.9\]

or about 330× volume compression.

The confinement time must improve relative to the compression time

General Fusion has previously pointed to energy-confinement times exceeding 10 ms as the target-plasma lifetime needed for LM26’s compression program.

In LMC-11, the inferred ion energy confinement time was about 5.3 ms, while the compression time was about 3.6 ms, giving

\[\frac{\tau_{E,i}}{\tau_C} \approx \frac{5.3}{3.6} \approx 1.5.\]

That is enough to see some late ion reheating, but not enough for clean adiabatic ion heating to keV-scale ion temperatures.

What is still missing

The missing result is convincing, fusion-relevant plasma performance. General Fusion needs to show ion heating or it does not have a chance at a fusion energy machine.

Why it matters

After two and a half decades of work, General Fusion has not yet demonstrated compressive heating of ion temperatures above 1 keV, let alone the 10 keV needed for fusion energy. This suggests fundamental flaws in what they are doing and/or how they are doing it. The work presented in this paper is a small step forward from where General Fusion was before LM26. General Fusion is going to need to take a relatively huge leap in plasma performance with the funding from this deSPAC to survive long in the public markets.

Bottom line

LM26 has demonstrated compressional heating, but the published result is still a sub-keV electron-temperature result with small ion heating. It failed to make the plasma behave adiabatically enough for fusion, especially for the ions.