OpenAI raised money recently at a valuation of $100 billion whereas the last time Commonwealth Fusion raised money it did so at a valuation of "$7.2—10.8b (Dealroom.co estimates Dec 2021)". Also, OpenAI is only one of dozens of well-funded organizations in the space.
Source of the latter fact: https://app.dealroom.co/companies/commonwealth_fusion_systems
OpenAI has to face off against giants like Google and Facebook, as well as other startups like Anthropic. There are dozens of other organizations in this space, although most are not as competitive as these.
Commonwealth Fusion has to face off against giants like ITER (funding maybe $22B, maybe $65B, estimates vary) and the China National Nuclear Corporation (building CFETR at ?? cost, while a much smaller experiment in China cost ~$1B), as well as other startups like Helion. The Fusion Industry Association has 37 members, which are all private companies trying to get fusion.
There's probably currently more private investment in AI, and more public investment in fusion. Many of the investments are not publicly available, so a direct comparison between the entire fields is difficult. I choose to focus on two startups with available data that seem to be leading in their respective fields.
I thought about including valuation in the table as well, but decided against it:
I am very excited to see fusion work, simply for the psychological benefit that the future is happening.
With that said, I think you might want to contrast the potential gains if openAI succeeds just a little bit past what they already shipped, vs if Commonwealth demonstrates net energy gain.
Theoretically with large scale learning the model hallucinations and error rate should drop below most human experts, and it will also be better at any task that can be checked. So any math or cs problem below a certain length, the tool will be faster and better than humans.
This means the net value created is the (size of the white collar stem economy - cost to run the model)
If openAI extends to robotics this becomes (size of ~50 percent of human economy - (cost of robots + cost of compute))
At steady state, is the cost lower than (solar + batteries + methane backup generation with a carbon tax) in the year the tech is cost reduced)? Both solar and batteries are still dropping in price so this is a moving target.
If you try to look past hype and just look at fundamentals, factoring in perovskite cells and sodium batteries....I don't know. Gut feel is a fusion reactor has thousands of unique parts and skilled labor is needed to construct one. Quality must be very high or fusion wont be stable and the plant fails. (And is probably not cheap to fix..imagine trying to rip out a damaged magnet from the hundreds of pipes and wires around it..)
My hunch is at scale, optimally producing solar panels and batteries from cheap raw materials with robots will probably beat fusion at any achievable price for fusion. The only use for fusion is if humans run out of cheap land on earth for solar (and all the roofs etc) or spacecraft.
A big chunk of the "hunch" comes from here https://www.lazard.com/research-insights/2023-levelized-cost-of-energyplus/ This shows current solar and batteries is already just a small amount of cost reduction from being strictly dominant, and this is without any improvements like perovskites or sodium.
summary: in order for fusion to create any net value it must beat the future cost of improved solar and batteries at the year fusion equipment is being mass produced. If it doesn't, the value of fusion is negative and it will not be adopted.
The cost to build a tokamak that is projected to reach Q~10 has fallen by more than a factor of 10 in the last 6 years. CFS is building for $2B what ITER is building for maybe $22B, maybe $65B (cost estimates vary).
It's really not clear what the cost of fusion will end up being once it becomes mass produced.
Ok, I think I may have missed a key piece above.
1.8 trillion is currently spent globally to generate electric power.
96.5 trillion is current world GDP.
If AI automation can reduce the cost of 50% of jobs by 50%, then it's value per year is 24 trillion. (much more because AI will enable to the economy to grow)
Obviously if fusion makes electricity cost $0, free, it's value created per year is 1.8 trillion. More realistically, competitive fusion will probably not reduce costs at all - it will simply reduce carbon emissions, which is a cost not priced into that "1.8T" figure. If we say the cost of the carbon emissions are $75 a ton, and 36.8 gigatons are global carbon emissions for electric power generation, then $2.76 trillion is the "externalities" from generating electricity.
So if fusion costs the same as current equipment at scale, then the benefit from fusion is $2.76 trillion.
Also, electric power is usually not the bottleneck resource for economic growth. It's a necessary condition but human labor, IP, economic systems that don't allow mass amounts of theft and inefficiencies - these I think contribute much more.
I think in some ways this says more about the structure of how we as a society finance promising emerging companies than it does about the companies themselves.
If there is equivalent smoothness to the experiment curve for fusion to deep learning, I am not aware of it. It is my impression that fusion experiments are decidedly more qualitative and question-specific, whereas deep learning is all about trying to get a tall enough stack of tricks and a big computer.
OpenAI has being doing various remarkable technical things between 2015 and 2022. The most well-known of them is the 2020 GPT-3 revolution, but there has been a long list of remarkable achievements done by OpenAI before and after that.
So for an actual comparison one should compare what OpenAI has been doing before 2021 (this includes GPT-3) with what CFS has been doing before 2024. Do they have something comparable to the 2021 OpenAI in the sense of technical magnitude of achievements? Yes, fusion might be more difficult, and producing varying brilliant new experiments in fusion is way more difficult, so let's adjust for that, but we would nevertheless like a tech comparison (e.g. with Helion Energy I can point to some of their technical milestones in this sense, and try to ponder the parallels, but I am not all that familiar with CFS and their tech in this sense).
In this sense, I would also love a more detailed argument in favor of the claim that
The leading startup for fusion is Commonwealth Fusions Systems (CFS)
Specifically, I would love to see a better argument for it being ahead of Helion (if it is actually ahead, which would be a surprise and a major update for me).
Helion has raised a similar amount of capital as Commonwealth: $2.2B. Helion also has hundreds of employees: their LinkedIn puts them in the 201-500 employees category. It was founded in 2013, so it is a bit older than CFS or OpenAI.
My general sense is that there's more confidence in the plasma physics community that CFS will succeed than that Helion will succeed.
SPARC is a tokamak, and tokamaks have been extensively studied. SPARC is basically JET with a stronger magnetic field, and JET has been operational since the 1980s and has achieved Q=0.67. It's only a small extrapolation to say that SPARC can get Q>1. Getting to Q~10 involves more extrapolating of the empirical scaling laws and trusting numerical simulations, but these are scaling laws and simulations that much of the plasma physics community has been working on for decades.
Helion uses a different design. This design has been tested much less, and far fewer plasma physicists have worked on it. They also haven't published as much data from their most recent experiment: last time I checked, they had published the temperature they had reached on Trenta, but not the density or confinement time. Maybe the unpublished results are really good, and suggest that the scaling that has worked so far will continue to work for Polaris, but maybe they're not. It's plausible that Polaris will get Q>1 when it is built (planned for 2024), but I'm not as confident about it.
Also, Helion uses D-He3 rather than D-T. This means that they produce far fewer and less energetic neutrons, but it means that their targets for temperature and density / confinement time are an order of magnitude higher. Even if you think D-He3 is better in the long term (and it's not clear that it is), using D-T for initial experiments is easier.
My general sense is that there's more confidence in the plasma physics community that CFS will succeed than that Helion will succeed.
That is, indeed, an important indicator.
Otherwise, tokamaks being an old design works as an argument in the opposite direction for me (more or less along the following lines: tokamak design has been known for ages, and they still have not succeeded with it; perhaps an alternative and less tried design would have better chances, since at the very least it does not have the accumulated history of multi-decade-long delays associated with it).
(I guess, my assumption is that the mainstream plasma community has been failing us for a long time, feeding us more promises than actual progress for decade after decade, and that I would rather bet on something from the "left field" at this point, at least in terms of the chances to achieve commercial viability relatively soon, as opposed to the ability to attract funding or boost headcounts.)
Basically, yes, one thing we are comparing is their (Helion and CFS) respective 2024 and 2025 promises regarding Q>1, but more importantly from my viewpoint, Helion's promise to actually ship electricity to the customers in 2028 does seem overoptimisitic, but perhaps not outrageously so, whereas with tokamaks, what's our forecast for when they have a chance to actually ship electricity to the customers?
Tokamaks have been known for ages. We plausibly have gotten close to the best performance out of them that we could, without either dramatically increasing the size (ITER) or making the magnets significantly stronger. The high temperature superconducting[1] 'tape' that Commonwealth Fusion has pioneered has allowed us to make stronger magnetic fields, and made it feasible to build a fusion power plant using a tokamak the size of JET.
After SPARC, Commonwealth Fusion plans to build ARC, which should actually ship electricity to customers. ARC should have a plasma energy gain of Q~13, and engineering energy gain of about 3, and produce about 250 MWe. They haven't made any public commitments about when they expect ARC to be built and selling electricity to the grid, but there has been some talk about the early 2030s.[2]
The higher temperature is not really what we care about. What we really want is higher magnetic field. These two properties go together, so we talk about 'high temperature superconductors', even if we're planning on running it at the same temperature as before and making use of the higher magnetic fields.
You don't need to have any insider information to make this estimate. Construction of SPARC is taking about 4 years. If we start when SPARC achieves Q>5 (2026?), add one year to do the detailed engineering design for ARC, and then 4 years to construct ARC, and maybe a year of initial experiments to make sure it works as expected, then we're looking at something around 2032. You might be able to trim this timeline a bit and get it closer to 2030, or some of these steps might take longer. Conditional on SPARC succeeding at Q>5 by 2028, it seems pretty likely that ARC will be selling electricity to the grid by 2035.
Specifically, I would love to see a better argument for it being ahead of Helion (if it is actually ahead, which would be a surprise and a major update for me).
I agree with Jeffrey Heninger's response to your comment. Here is a (somewhat polemical) video which illustrates the challenges for Helion's unusual D-He3 approach compared to the standard D-T approach which CFS follows. It illustrates some of Jeffrey's points and makes other claims like Helion's current operational poc reactor Trenta being far from adequate for scaling to a productive reactor when considering safety and regulatory demands (though I haven't looked into whether CFS might be affected by this just the same).
Artificial intelligence is not the only exciting emerging technology. Another one that I am personally familiar with is fusion.
It seems interesting to compare companies working on emerging technologies to get a feel for how surprising developments in these companies are. Some startups working on developing AI have grown rapidly, but how surprising is this growth?
The leading startup for AI is OpenAI. The leading startup for fusion is Commonwealth Fusions Systems (CFS). Here is some basic information for these two companies:[1]
OpenAI was founded 2-3 years before Commonwealth Fusion Systems.[2][3] Other than that, their trajectories have been surprisingly similar.
OpenAI currently has 770 employees.[4] Commonwealth Fusion Systems currently has “>600 employees and >100 contractors.”[5]
OpenAI has received about 5 times as much funding as CFS. Most of this money came from a single investment by Microsoft in 2023.[6] If we look at OpenAI from 2-3 years ago to control for the different founding dates, then OpenAI had raised less money than Commonwealth Fusion Systems. It would not surprise me if CFS gets significantly more investment in the next several years as well.
The $10B investment in OpenAI occurred after the launch of ChatGPT, 7 years after the company was founded. Commonwealth Fusion Systems’ first major project is SPARC, a tokamak experiment which should be completed in 2025, also 7 years after the company was founded. The goal of SPARC is to demonstrate Q>1, or net energy gain from the plasma,[7] as quickly as possible and then work up to demonstrating Q~10.[8] Fusion projects are notorious for falling behind schedule,[9] but CFS has maintained its schedule so far and construction is well underway. It is not certain if SPARC will succeed in demonstrating Q>1 in 2025 or 2026, but if they do, they will likely see a significant increase in hype - and funding.
OpenAI was definitely not making a profit as recently as 2022.[10] Their revenue has increased dramatically since then,[11] so they might be making a profit now. Commonwealth Fusion Systems is definitely not making a profit now.
I’m not entirely sure what conclusions should be drawn from this comparison. It seems easy to overestimate the relative importance of an industry that is dominant locally (either in your city or in your social network) and to be unaware of developments that are more distant from you. OpenAI and CFS are obviously not the only actors in AI and fusion, respectively, and I have not done a systematic comparison between the two fields. The technical details of these two fields are very different and so place different demands on their workforce and capital. Nevertheless, this comparison provides some evidence that investors’ estimates of the value of AI and fusion are not wildly different.
See text below for sources.
Introducing OpenAI. (December 11, 2015) https://openai.com/blog/introducing-openai.
MIT launches multimillion-dollar collaboration to develop fusion energy. (March 9, 2018) https://www.nature.com/articles/d41586-018-02966-3.
Savov, Vance, & Ludlow. OpenAI Staff Near Total Mutiny With Threat to Join Microsoft. Bloomberg. (November 20, 2023) https://news.bloomberglaw.com/us-law-week/openai-staff-threaten-to-go-to-microsoft-if-board-doesnt-quit.
Dan Brunner. The high magnetic field path to fusion energy. 65th Annual Meeting of the APS Division of Plasma Physics NO05:1. (November 1, 2023)
Bass. Microsoft Invests $10 Billion in ChatGPT Maker OpenAI. Bloomberg. (January 23, 2023) https://www.bloomberg.com/news/articles/2023-01-23/microsoft-makes-multibillion-dollar-investment-in-openai.
Q is the ratio of energy injected into the plasma to the energy produced by fusion. To accommodate inefficiencies in the rest of a fusion power plant, you would typically want to operate at Q>5.
Rodriguez-Fernandez et al. Overview of the SPARC physics basis towards the exploration of burning-plasma regimes in high-field, compact tokamaks. Nuclear Fusion 62.4. (March 1, 2022) https://iopscience.iop.org/article/10.1088/1741-4326/ac1654.
Fusion projects are also often large international collaborations. Missing schedule and budget targets could be because of the institutional structure of many fusion experiments, rather than because of something inherent to fusion technology itself.
Erin Woo & Amir Efrati. OpenAI’s Losses Doubled to $540 Million as It Developed ChatGPT. The Information. (May 4, 2023) https://www.theinformation.com/articles/openais-losses-doubled-to-540-million-as-it-developed-chatgpt.
Amir Efrati. OpenAI’s Revenue Crossed $1.3 Billion Annualized Rate, CEO Tells Staff. The Information. (October 12, 2023) https://www.theinformation.com/articles/openais-revenue-crossed-1-3-billion-annualized-rate-ceo-tells-staff.