I’m Suspicious

This is one of those “damn it, I am not qualified to write about this…” topics that I want to write about. Therefore, I encourage any of you who feel you are more qualified to correct me accordingly.

In my news-feeds for the last month, I’ve noticed a smattering of odd postings about a fusion energy project from MIT.

I say they are “odd” because they’re really optimistic, i.e.: “we may soon have all the free energy we want!” kind of stuff. Which does not fit well with what I know about controlled fusion and the engineering problems that come with it. The New York Times has had several references to the MIT reactor, and MIT’s publicity engine appears to be spinning hard enough that maybe they could power Cambridge by hooking a generator to it.


The New York Times
Publication Date:


In a series of new papers, MIT researchers provide evidence that plans to develop a next-generation compact nuclear fusion reactor called SPARC should be viable, reports Henry Fountain for The New York Times. The research “confirms that the design we’re working on is very likely to work,” says Martin Greenwald, deputy director for MIT’s Plasma Science and Fusion Center. 

[SPARC is a trademark of Sun Microsystems, now a part of Oracle, btw]

That sounds … weird. Word-hashful. “researchers provide evidence that plans to develop a next-generation compact nuclear fusion reactor called SPARC should be viable” wait, someone wrote a paper saying “this outta work!”?

I thought that ITER [iter] and the US’ national implosion doodad were also “this outta work” type stuff. I mean, you don’t spend hundreds of millions of dollars to build a great big tokamak with the expectation that it won’t work, right? I thought that the whole point of “controlled and sustainable fusion is really hard” is that it’s hard from an engineering standpoint and that it’s fine and good for a theoretician to say “this oughtta work” but it turns out that the components can’t hold up to operations, or something like that.

I read Charles Seife’s Sun In A Bottle [amzn] a few years ago, and I came away with the overall impression that researchers have explored a lot of options and they variously seem attractive for a while but otherwise haven’t worked. Obviously, if someone can build a self-sustaining fusion reaction, that would be a game-changer for humanity, so it’s a Big Deal and we should take it seriously. It seems odd, though, that MIT is suddenly swanning in and saying “we got this!” when they haven’t even got a 3 story tall mass of giant electromagnets, or anything remotely like it.

If you google for ‘MIT fusion reactor’ you’ll find a stream of hits from MIT’s press site and other places that picked up the story, all around September 29, 2020 then continuing for a while after that. Sure looks like a publicity blitz, to me. But it also looks more like the kind of publicity blitz that accompanies junk science attempting to attract investors to a new kind of perpetual motion engine. There have been a few of those, too. It seems that every so often there’s a publicity blitz for a new rocket engine that will let humanity get to the stars (if it only worked and if we could just spend a few $100mn scaling it up) or perpetual motion catalytic something-or-other.

The MIT SPARC press releases include cool computer renderings of something that looks a lot like a tokamak:

Maybe I’m cynical but pointing to a 3D rendering and saying “that’ll work!” is OK for blacksmithing and woodworking but I’m not so sure if it scales up to nuclear engineering.It’s also interesting to see the various follow-on articles that stem from the initial MIT press releases:

  • Scientists say their compact reactor could be up and running in just 10 years
  • Nuclear fusion reactor could be here as soon as 2025
  • MIT: Revolutionary Nuclear Fusion Reactor on Track to Operate by 2030

Back in the day, we used to giggle about MIT press releases predicting artificial intelligence breakthroughs, and worrying about runaway nanotechnology turning the planet to a hellscape of muck.* This sounds like another one of those.

Out of curiousity, last night I spent some time re-reading chunks of Sun In A Bottle and stumbled across the fascinating story of relocated Nazi scientist Ronald Richter, who set up shop in Peron’s Argentina [wik] [iter] claiming to have solved the problem of sustainable fusion at around the same time that the rest of the world was just managing to make uncontrolled fusion explosions.

Has the requisite amount of stainless steel

This is all that’s left of Richter’s project.

It sounds like he had a good scam going for a while, but eventually it didn’t work. There’s a story that’s relevant to this, about a sultan who had a sooth-sayer come that told him he could teach the sultan’s horse to sing. The sultan, being no fool, told the sooth-sayer that he had one month to teach the horse to sing and, if it didn’t, he’d be impaled. The sooth-sayer got the sultan to agree to a year, plus fine lodging and meals. The sooth-sayer went to work cheerfully each day, saying “maybe it will learn to sing.”

Stainless steel, check. Orange jumpsuits,, check. Round thing, check.

I’m not an idiot and I’m aware that research scientists at places like MIT are able to spin off their research into commercial applications, and MIT gets a big chunk of their action. I believe that those companies can also do independent fund-raising, i.e.: sell stock in the venture for equity. If that’s why this looks like marketing bullshit, it’s because it probably is; they’re trading on MIT’s good name.

------ divider ------* Everyone knows that runaway nanotech would most likely turn the planet into a mass of jiggly pink silicone penises. Because that’s how tech titans think.

In the course of farming links for this, I checked out ITER’s page, which has some really cool footage of someone forging a massive armature out of what looks like 8 feet by 2 feet of yellow hot steel. It’s very impressive but I have no idea where something like that fits in a tokamak, other than on the webpage about one.


  1. says

    There are numerous privately (or military) funded fusion research projects at the moment. MIT’s is just one of them.

    MIT seems to think that the use of new high-powered superconducting magnetics along with better computer modelling will allow them to produce energy with a small tokamak. (Note that the picture is a visualisation, not a photograph.) Whether they’re right or not we can’t say yet.

    There’s also General Fusion in Canada, working on generating fusion inside a rotating sphere of molten lead, First Light Fusion in the UK, working on an inertial confinement fusion system, TAE Technologies in California, working on a linear accelerator system (I think), and several others.

    Will any of them succeed before ITER in demonstrating net energy generation? I’ve no idea.

  2. says

    Scientists say their compact reactor could be up and running in just 10 years.

    Well, there’s an improvement. For the last 60 years they’ve been saying that nuclear fusion is “just 20 years away”. Now they are down to being able to say, for the next 60 years, that nuclear fusion is “just 10 years away.”

  3. says

    One of my former interns works on SPARC and so I’ve had to chance to visit the old fusion reactor at MIT (SPARC isn’t built yet).
    When I was asking him what was fundamentally different about SPARC vs ITER he handed me a length of a weird looking wire and explained about the latest in high powered magnets with low temp superconductors.
    Now I worked in radiology (in the software side) at GE and Kodak before that for almost 20 years so I have some small amount of familiarity with high powered magnets. When I started at Kodak the highest powered MRIs were 1.5 Tesla and that is still commonly used in clinical settings, but you can get a 10.5 Tesla human rated MRI today or a 20 Tesla non-human rated MRI, that level of power was just science fiction in 1998.
    ITER was designed with 11.8 Tesla magnets, see: https://www.iter.org/mach/Magnets
    SPARC is being built with 12 Tesla magnets, see: https://www.psfc.mit.edu/files/psfc/imce/research/topics/sparc/MITSPARCbrochure.pdf
    Think of it as Moore’s Law for magnets.

    Go search what the Lockheed Skunk Works is up to these days. You might be surprised…

  4. says

    Patrick Slattery@#3:
    Good points – From reading about how difficult it is to control plasma, it sounds like an improvement in sensor and processor speed could solve some problems. E.g: a Segway was not possible 10? 20? 30? years ago. If they have more powerful and faster (is that even a thing?) magnets it might be possible to turn plasma control into a software problem. At which point someone’s gonna say “why not use node.js?” ;)

  5. says

    That’s a lot of what my friend there works on, he’s a DevOps engineer (Worked at Docker for a while) and has been helping them convert their test workloads to containers and Kubernetes so that they can scale out more easily.

  6. billseymour says

    … it might be possible to turn plasma control into a software problem. At which point someone’s gonna say “why not use node.js?”

    Marcus wins the Internet for today.

  7. DrVanNostrand says

    In fairness to ITER, it still ought to work. It’s just a massive construction project, and like most massive construction projects it’s taking longer and costing a lot more than anticipated. But it’s not like there have been any technical failures, or impossible to fabricate components. Steady progress is still being made.
    As far as the “national implosion doodad” goes, knowing several people that work at NIF, it was generally considered a long shot to produce sustained net-positive power output. It’s primarily a tool for fundamental research, particularly research about nuclear reactions that are otherwise impossible without performing test explosions that have been banned. There may have been some hope that they’d learn something after they built it that would make it work as a power source, but unlike the massive ITER tokamak, they didn’t have any concrete plan for generating sustained power.

  8. Who Cares says

    @Paul Durant(#1):
    ITER does not generate electricity so it will never breakeven. ITER is a testbed for materials, controls and physics. You might get people to breathlessly report that ITER is generating X MW but that is/will be thermal energy and not electrical output. Other then that 2030 to 2035 before a theoretical conversion (40% efficiency) of thermal output would be bigger then the total amount of energy needed to run ITER.

    Also ignore the reports that state that only 50 MW of power is needed for ITER since that is just the initial plasma heating and not the entire energy input budget of ITER. The claim that they expect a factor 10 return for the first runs is not incorrect, with the added caveat that you must only use that 50 MW for initial plasma heating, but that is thermal energy so not yet converted to electricity. The more realistic numbers I have gotten my grubby mitts on is that the first runs take around 300 MW for all of ITER not just that 50 MW to do the initial heating and a return of 675 MW of thermal energy, which would require an 45% conversion efficiency. Not an unreachable number seeing that a combined cycle plant can get between 50% & 60% but that is after optimizing so for the first few runs a 35 to 40% is more realistic. And at 40% that means you end up with putting in 30 MW more then getting out of it.
    It is DEMO (yes a lovely backronym there) the plant that comes after ITER that is supposed to show commercially viable fusion and this electricity generation. Note that it is very well possible that ITER gets converted to DEMO.

    @Marcus Ranum(#7):
    Difficulty is also dependent on the type of reactor.
    As one of the people who was working on the microwave waveguides for ITER (came in to consult the ex-prof who was the owner of the startup I was working at at the time) complained:
    A tokamak is easy to build but a pain to control the plasma (the microwaves were needed to ‘massage’ unstable areas of the plasma among other things).
    A stellarator is a pain to build but due to the form but it is easy to control the plasma.

    Just let that sink in. They considered ITER easy to build and it is six years behind schedule (or should that be only six seeing the size and number of parties involved). Yes first test runs with fusion were supposed to have started in 2019. That was four years from start of the reactor vessel starting to be constructed (note constructed not assembled) not the ten it is semi officially now.

    And that a stellarator makes it easy to control the plasma is I guess relative to a tokamak.

  9. Allison says

    If they have more powerful and faster (is that even a thing?) magnets it might be possible to turn plasma control into a software problem.

    If only it were so simple.

    AFAIK, the only “control” you have over a plasma’s shape is via the external magnets, since you can’t have anything inside the plasma. The basic problem in magnetically confined plasmas is that the plasma generates its own magnetic field which basically bends and twists the net magnetic field, mostly in ways you don’t want, like directing the plasma into the wall of the vacuum vessel. Increasing the toroidal magnetic field dampens some such instabilities, but others can happen regardless of the strength of the field. Moreover, tokamaks work by inducing a strong toroidal current, which also generate magnetic fields, and the currents flow (approximately) along the magnetic field lines, which of course are getting bent by the plasma’s fields.

    There’s a reason they describe magnetic confinement as “containing a blob of jello with a few rubber bands.”

    That was at least the situation 40 years ago, but judging by the mock-up, it hasn’t changed.

    There’s also a problem with the terminology. “break-even” is frequently defined leaving out stuff like the cost of powering the magnets and the cost of any other external heating devices.

  10. cvoinescu says

    Marcus, the website does say it’s a tokamak design, the difference being much better superconducting magnets. According to Patrick Slattery, the field intensity is not a huge improvement over ITER, so maybe the magnets are smaller for roughly the same field, which allows for a smaller doughnut and/or a more favorable geometry?

    […] has been helping them convert their test workloads to containers and Kubernetes […]
    We’re doomed.

  11. cafebabe says

    In 1957 the ZETA project in the UK announced to the press (not to peer review!) that they had achieved controlled fusion with their “pinch” plasma device. Then they checked their calculations one more time and have worn egg on their faces ever since. By 1961 I was working in a nuclear science lab with one of the guys who had been on that project. He was a plasma physicist and still claimed that the approach was sound and they would achieve net positive power within ten years.

    Well, we’re still waiting.

    Apart from the plasma confinement problems which even the huge expenditures of the ITER experiment are still struggling with there are material availability issues and shielding challenges. Although fusion reactions do not produce fission fragments, all the reactions that are within reach of current technology produce huge neutron fluxes and thus great amounts of radioactivity in the containment system. There are aneutronic reactions, but these require much higher plasma densities for ignition.

    There are a number of project based around “tricks” of various kinds to avoid doing plasma containment. IEEE Spectrum gave an overview of five of these in the January issue this year.

    I switched careers in 1972, so I won’t claim any current expertise in this business, but for a measured but negative view from someone who is real expert I can recommend this article from the Bulletin of the Atomic Scientists.

  12. Who Cares says

    The description you give is for when trying to control the plasma in a tokamak. A stellarator is actually designed to work with the plasma to contain it. Thing is you could build a tokamak with WWII technology (yes that is exaggerated) where the first successful stellarators basically required computers capable of running CAD and physics simulations.

    And that engineering difficulty shows.
    There are hundreds of tokamaks in the world, my local university probably has the capability to build one if there was a reason for them to do so.
    While there might have been a hundred plus attempts at producing a working stellarator but less then half a dozen that actually manage to have a plasma capable of fusion. Well 2 that I know of these days and both produce ridiculously stable plasma (compared to a tokamak).

  13. DrVanNostrand says

    @14 cafebabe

    That 2nd link was great. I’m still all in favor of continuing fusion research, but it’s always valuable to keep in mind the very real practical hurdles. And those are just the “known unknowns”. Imagine all the “unknown unknowns”.

  14. Numenaster, whose eyes are up here says

    “a Segway was not possible 10? 20? 30? years ago.”

    Ahem. My Amazon order history says I ordered mine 17 years ago, on March 7, 2003.

  15. lorn says

    I don’t wish to cast any shade, or cause any loss of funding, or enthusiasm because a cheap, clean, reliable source of energy would solve a lot of problems. I’m all for it. That said, I’m over 60 and still remember any number of pronouncements from people in lab coats standing in front of drawings about fantastic new energy sources. Hydroelectric, geothermal, solar, nuclear, and fusion (both cold and hot) and granted some work well enough to handle real needs but none of them have been all they were cracked up to be. There have always been costs and technical issues that got overlooked in the gee-whizz phase. So far power that is too cheap to meter, and doesn’t poison the environment, is still a pipe dream.

    So far the story has been: ‘fusion is the energy source of the future’ … and always will be.

    Perhaps it’s another fallout from this administration. I’m tired of the constant diet of over-promising and under-delivery. I knew Don was a crook and didn’t vote for him but damn this administration couldn’t pour piss out of a boot if instructions were written under the heel. And they are determined to making sure nothing makes them look bad by being more effective than they are. The result of 40 years of Reaganomics and four years of Don the con is: I’m all out of optimism.

    I’ll jump on the fusion bandwagon a couple of years after they start wiring houses to the damn things. Assuming they don’t blow up, burn down or poison people.

Leave a Reply