Helion Energy to demonstrate net electricity production by 2024

[Munchausen]

New video:

https://www.youtube.com/watch?v=oswGdRTyppg

It certainly looks like a serious operation.

[baking]

Running out of 2025 days. I wonder what the issue is?

"Q: It's just that gradual process of just doing more and more tests? A: And every day you add a system, you upgrade, you go to higher power, honestly, some days you break things."

https://www.youtube.com/watch?v=oswGdRTyppg&t=586s

You know, science.

[mvanwink5]

I wonder if the issue is tuning the electronics, massive power in, timed, then massive power back out to the capacitors, **add in plasma dynamics**, fusion reaction timing. (we had 10,000 HP or 7.5 MW, variable speed inductive motors driven by power electronics, key a radio nearby & watch them shutdown, & these power circuits were a commercial product, Helion builds their own power supplies for GW’s & microsecond pulses).

Crazy.

[baking]

I wonder if the issue is tuning the electronics, massive power in, timed, then massive power back out to the capacitors, **add in plasma dynamics**, fusion reaction timing. (we had 10,000 HP or 7.5 MW, variable speed inductive motors driven by power electronics, key a radio nearby & watch them shutdown, & these power circuits were a commercial product, Helion builds their own power supplies for GW’s & microsecond pulses).

Crazy.

Helion recently released a couple of videos describing full-size "bench-top" setups in which they connect a large capacitor bank to a coil to work out potential bugs in the system.

https://www.youtube.com/watch?v=u1R51Z9-TM4&t=213s

https://www.youtube.com/watch?v=XyOL2S3N6Kw&t=456s

[mvanwink5]

Baker, TY for pulling the vids & links. Insight into the reality of Helion Fusion Generator build is helpful in calibrating time expectation. As a guess, the hard engineering is behind Helion by now, but we will just have to wait for the plasma optimization results.

[Skipjack]

Much of the show was background information. My notes from latter part of the interview:
1:41 With FRC, one can recover the input energy at 90 or 95%, and the fusion energy at 80%

that's funny, 80% was actually my BOE/WAG a year or so ago... don't think they've given out a number before

I have said that before.

[Skipjack]

From all I have heard, they are still carefully ramping up power to the compression magnets and doing tests with "simulators" at the same time. They are deliberately careful with ramping up to "15 Tesla". Think about one of those pulsed magnets failing catastrophically! It is not "pretty". So, they are cautious.
My take:
The problem they have with Polaris is that it turned into a large and expensive machine with a lot of parts that are not easily replaced because they had to produce many of them in- house rather than being able to buy them off the shelf. It is sort of ironic because that was what they originally wanted to side- step with their design. But after COVID and the ensuing supply chain problems a lot of things changed (not just for them, btw).
Don't get me wrong here! I am not saying that Helion is on a bad trajectory, but they have certainly not been able to move as quickly as they thought they would be because of that (who is not behind, though?).
IMHO COVID and everything that came with that has made everything worse for everyone and might have cost humanity years (not just in the fusion field). Sometimes I wonder whether it would have been better to just take the deaths and move on (in the long run) and I am high risk person.

[TallDave]

yeah, that makes sense

I guesstimate D-He3 breakeven around 12T given Kirtley's stated B^3.77 scaling

I did ask if they could drop a power emoji when it happens (I don't need details and will take them entirely at their word) but again, if it was my billions on the line I'd delay that announcement as long as possible, preferably not until I was actually delivering power

the Chinese knockoffs are already underway

[Skipjack]

In time for Christmas, here is some new information, I got through the grape wine:
Polaris's capacitor bank is now at 80 MJ.

Trenta did indeed proof energy recovery of both fusion and input energy. However, it was not returned to the capacitor bank, but instead went into resistors. Still hoping for more details on that.
Either way, this fact about Trenta was news, even to me. It explains why they have a certain confidence in their design.

[Carl White]

Grok estimates there's a 15% to 25% chance that Helion will demonstrate net electricity generation in 2026:

Based on the latest updates from Helion Energy and broader fusion industry analyses as of late 2025, the probability that they will demonstrate slight net electricity (i.e., Qe > 1, producing more electrical output than input) in 2026 with their Polaris prototype appears low, around 15-25%.

Key Factors Influencing the Odds:

Progress and Delays: Polaris has been operational since late 2024, running daily fusion operations and tuning as of November 2025. However, Helion has not announced net electricity achievement by the end of 2025, missing their revised target. Reports indicate the goal has shifted to 2026, with ongoing challenges in energy recovery and helium-3 fuel production.

Technical Hurdles: Fusion remains experimentally challenging, with no private company yet achieving commercially viable net gain. Helion's pulsed helical approach shows promise in efficiency (targeting 85% direct electricity recovery), but plasma stability, high-temperature requirements (200-300 million degrees), and component durability have caused setbacks. Skepticism in fusion communities highlights repeated timeline slips across the sector.

Prediction Markets: Crowdsourced markets like Manifold reflect pessimism. Related questions show:

3% chance of breakeven before end of 2025 (likely to resolve NO).

2% chance of breakeven deuterium-helium-3 fusion before 2026.

18% chance of delivering 50 MW to Microsoft by 2029 (requiring net demo as a prerequisite).

19% chance of providing fusion power to a data center before end of 2028.

These low probabilities account for historical fusion delays and Helion's secretive nature, tempering optimism despite their $2.2B+ funding and Microsoft PPA.

Positive Indicators: Construction on the commercial Orion plant (targeting 50 MW by 2028) is underway, suggesting confidence in eventual success. If Polaris tuning yields breakthroughs in 2026, net electricity could follow quickly. Industry investment (e.g., from Sam Altman) and prototypes like Polaris operating at scale provide some upside.

Overall, while Helion is farther along than most fusion startups, the track record of delays in the field (e.g., no net gain announcements despite 2025 promises) outweighs the progress, leading to conservative odds. If new milestones are hit early in 2026, this could rise; otherwise, further slips to 2027+ are plausible.

[Skipjack]

Well, Grok can only go by the information that is publicly available. And even that is not always complete.

[jrvz]

Helion has private funding, and little incentive to share their data with the competition, so Grok doesn't have much to work with. I think they'll succeed, though I am not willing to bet on the timescale. My worries:

- It will take a lot of tuning. And each successive device will take a lot of tuning, slowing production.
- It will be complex, with many interacting parts, and hard to keep in good repair.
- It will take a highly trained team to keep it running, with the appropriate balance (among power production, tritium and He3 production, neutron activation, etc.) and it will take more time to train the next team than to produce the next device.

On the other hand, once they get the first few devices set up, I expect they'll find ways of simplifying the design - less flexibility where they don't need it, with some of the tuning and management moved to the software.

[Skipjack]

Well, experimental machines are always more complicated than the serial produced ones (just for all of the diagnostics alone). Look at the evolution of the SpaceX Raptor engine! They will get a lot simpler and cheaper to build over time.

[baking]

Well, experimental machines are always more complicated than the serial produced ones (just for all of the diagnostics alone). Look at the evolution of the SpaceX Raptor engine! They will get a lot simpler and cheaper to build over time.

Fair, but how long will it take them to get 250X smaller? From 100,000 SF to 400sf: https://imgur.com/a/helions-expanding-f ... rs-raywadu

[Skipjack]

Why does it have to be 400 sf?