Microsoft and Helion's fusion deal has an alternative energy
It’s about power and money alright, just not like that
Opinion Let's get the facts out of the way first. Microsoft and fusion energy startup Helion have announced an agreement. Helion is going to provide Microsoft with 50 megawatts of datacenter juice powered by its helium-3 fusion process, starting in 2028.
There we are. Facts done. You can read El Reg's coverage and in-depth look at the background of Helion and fusion in general. It's important to consider claims in such important and high-profile areas, especially when the stakes are so high – not for datacenters in particular but for the whole planet.
It's also important to consider, as the Reg said at the time, that we should take such claims with a "healthy dose of salt." As presented, the deal could be balderdash. Its stated aims and expectations don't seem to make sense. For me, there is little practicable chance that Helion or anyone else will be selling a watt of fusion-derived electricity in five years' time. Even 15 years is long odds.
Helion has yet to demonstrate an exothermic process that can generate more energy than it consumes within its reactor. Even if it did, it then has to show it can run continually. It has yet to detail, let alone demonstrate, how it will create enough excess energy to overcome the inefficiencies of turning that energy into electricity. It has yet to show it can even get hold of the helium-3 it needs, let alone in commercial, sustainable quantities. The tech might not exist. It is very far from being independently verified.
Helion is a well-funded startup with a history of promising improbable milestones that it hasn't delivered. Hardly the first, of course: it would be the first if it succeeded. Yet even if it unveiled a working prototype reactor that demonstrated answers to all of the above problems, it still couldn't hit the 2028 deadline.
You can barely go from blueprint to grid power in five years with an established design of nuclear power plant. Even a gas-fired plant takes two. Plus, both of these are in mature regulatory environments, of which no trace exists for fusion. What, did nobody mention that? You'd have thought Microsoft would be all about regulators by now.
No, sir. No, madam. If you ask me, Helion will not be putting a 100 million degree, neutron-spewing reactor of unknown design anywhere in public by 2028. Even if it has one by then, which it likely won't.
Devil's in the detail
So why has Microsoft done that deal? Does the company know something we don't? Is it just another monumentally stupid big tech deal? Again, hardly the first. No, the deal does have its own logic, just not one with anything to do with powering datacenters.
Let's look at the details of the deal – oh wait, we can't. Is Microsoft investing money up front? Are there get-out clauses that may let both parties walk away with little or no loss? Can Helion fulfill the contract by buying the power on the wholesale market and passing it on? We don't know. We don't know the size of the bet on either side. So it's hard to say how much either side really think the idea's a goer. 50MW of electricity costs maybe $20 million a year if you build your own wind farm, which is hardly high-stakes poker for either company.
Another way to look at the facts is that there's been a deal of unknown commercial importance, potentially none, and a press release with less weight than a child's balloon filled with helium-3. Which is an absolute bargain for what both companies are getting out of it.
Start with Microsoft. It has apparently made a bet on a very important technology, boldly putting its money on something that is widely perceived as having planet-saving potential. That looks good in print, it looks good in the annual report, it looks good if you want to be seen as a forward-thinking, deeply concerned and visionary technology company determined to do its bit. Is this a good return on investment? It surely is if the investment is nugatory: a good move, then.
- Nuclear power is the climate superhero too nervous to wear its cape
- The International Space Station will deorbit in glory. How's your legacy tech doing?
- TSMC and China: Mutually assured destruction now measured in nanometers, not megatons
- Engineers on the brink of extinction threaten entire tech ecosystems
The benefits for Helion are similarly reputational, but more important. Until it has actual products producing positive revenue, it needs investment – and lots of it. With a high-profile vote of confidence from Microsoft, investors will feel more confident in ponying up their cash. That might be really important in the next few years if by some mishap Helion doesn't meet its deadlines. Even if those deadlines include that 50MW supply deal with Microsoft, the company ends up well ahead.
It is of course entirely possible that all of the above reasoning is wrong, that Helion truly believes it can do 10 impossible things before breakfast and Microsoft agrees. Especially in matters of impossible technology companies, it is famously impossible to distinguish justified projection and unreasonable belief, between hope and hype. You have to look for independent verification, for transparency, for detailed answers to hard questions. You have to listen to knowledgeable people with no financial connection to a company to judge the chances of success.
This can be surprisingly hard, even for the professionals. The markets couldn't tell the key difference between Tesla and Theranos until far, far too late. But sometimes the balderdash light on the dashboard is dead on – and for now, it won't be powered by helium-3. ®