From the headlines, it seems that our problems are over.
Fusion power - which has been a distant goal for 70 years - is finally here. Microsoft has it, and we are heading for an era where low-carbon energy is available for all.
"Microsoft to be powered partly by fusion energy from 2028," says Engineering & Technology magazine, an organ of Britain's otherwise-august Institute of Engineering and Technology.
DCD's coverage of the story takes it at face value: "Microsoft has signed a power purchase agreement (PPA) with Helion Energy to get fusion power by 2028," we report, explaining that the cloud giant has signed a PPA for 50MW of power from a fusion reactor that Helion plans to build in five years.
For a story like this, face value is not good enough. Microsoft, Helion, and their investors are playing with us in this press release, and I am willing to bet this announcement will not result in Microsoft getting any fusion power at all by 2028.
Let's start with the deal, and move on to the technology. In power purchase agreements, firms invest money upfront, to fund the provision of new energy from clean sources such as wind farms. On the face of it, that's what is happening here. Microsoft is putting in money now, to get 50MW of power from 2028.
In the past, Microsoft has bought carbon credits (similar to a PPA) from Ontario Power Generation, some of which may come from a small modular nuclear reactor that OPG is building. This week, the company signed a PPA for geothermal energy in New Zealand.
This deal just applies a standard PPA to fusion, according to Helion CEO David Kirtley. He told CNBC that Microsoft has committed to the new technology, and Helion will suffer financial penalties if it can't deliver. "We’ve really put our skin in the game on this too."
But we've only got Kirtley's word for this. Microsoft has said it will buy power generated by a method that has not yet been demonstrated, from a company that has yet to deliver any surplus output energy at all. DCD has asked what penalties Helion will incur if it misses the target, what price Microsoft is paying for the power, and on what terms. "We aren’t able to discuss the details of the contract," Helion told us, "but we can say there are significant penalties to Helion if it is unable to deliver electricity to Microsoft."
There is so much wiggle room there. We have no idea what get-out clauses the deal contains.
That statement, for instance, only promises that Helion can deliver electricity, it doesn't mention fusion. As Rupert Goodwins points out in The Register, when 2028 rolls around, and Helion has to admit that this whole fusion thing turns out to be a little bit tricky, the deal might even allow it to simply buy electricity on the wholesale market and say: "Sorry, we've not got any fusion power yet, would Sir like some solar?"
Before we even look at the technology, consider this: in the real world, the fastest route to more nuclear power is currently supposed to be small modular reactors (SMRs), a new generation of proven pressurized water reactors (PWRs), streamlined for fast permitting and easy production in factories. The closest anyone has got to using SMRs to data centers, Green Energy Partners, hopes to have something running in 10 to 15 years.
With full backing from the UK government, UK SMR player Rolls-Royce hopes to have something running by 2027. In the US, the SMR front-runner NuScale just got regulatory approval, but doesn't expect to have anything running till 2029.
As I mentioned above, Microsoft signed a deal with Ontario Power Generation, a company building an SMR. That PPA will be most likely fulfilled from existing wind and solar, because OPG won't have a working SMR until 2028.
And that's with a working technology. Helion doesn't even have anything working. Permitting will take years - and Helion can't even apply yet.
But, you say, maybe Helion, backed by billionaire investments from the likes of Sam Altman (of OpenAI and Y Combinator fame), has some left-field idea that is going to crack the fusion energy problem. Well maybe, but there is no strong evidence that it has.
The company has been around since 2013, starting out with $7 million in public funding from NASA. It has since relied on private funding, most recently $500 million in 2021, led by Altman. In the private sector, Helion doesn't have to publish much or get its claims verified by third parties.
It uses the field-reversed configuration (FRC), which has been around since the 1960s, and is also used by rival fusion company TAE. Helion promises "pulsed fusion," in which fusion happens in a continuous cycle, with the plasma allowed to expand periodically so Helion can extract the energy.
Instead of the usual reaction between hydrogen isotopes deuterium and tritium, it plans to use a reaction between deuterium and helium-3 (whose ion is known as a "helion"). And instead of the usual energy capture methods, Helion says it can reclaim the energy by simple electromagnetic induction - turning the motion of plasma ions into electricity.
Both those innovations come up against a lot of skepticism from nuclear observers. The D-H3 reaction produces fewer neutrons so it's less radioactive, but it has lower reactivity, meaning that Helion will have to achieve a higher temperature and pressure than the conventional fusion approaches have yet achieved.
Helion also has to produce H3, which is very rare. It says it will also produce it through a separate fusion reaction.
Another problem Helion doesn't seem to have addressed is how to make a plasma of deuterium and helium-3, and prevent the deuterium ions from fusing with each other, creating all the problems of conventional fusion which Helion claims to avoid.
Like all fusion companies, Helion has a history of broken dreams: in 2014, it promised it was three years from fusion. It wasn't.
The big question isn't whether Helion can deliver commercial fusion by 2028 - but why on earth is Microsoft pretending to believe that it can?
The answer is that this announcement provides justification in itself.
Helion has signed the first PPA for fusion. That counts as a milestone, and will encourage future investors to throw more money at it.
The fusion field has a number of similar projects which are essentially playthings of billionaires. To name two, TAE is backed by the late Microsoft founder Paul Allen, while MIT spinout Commonwealth Fusion Systems is backed by Microsoft CEO Bill Gates. Projects like these might have real aspirations, but all too often they are all about the image, not the reality. We've watched schemes like this move on to bigger promises when any current one fails to materialize.
For Microsoft, it's much more cold-blooded. This announcement has generated a lot of positive coverage, and bolstered its image as a pioneer of clean energy. Far too many people will internalize a perception that it's all but running the whole company on fusion power. As Goodwins points out, if the investment turns out to be nugatory, the PR on its own is a very good return.
Fusion promises to create clean energy from water. The whole world will be delighted when this finally occurs.
For now, all Microsoft has managed to do is fuse hype and greenwash - but that's enough to serve its short-term ends.