AI sparks only dim odds of nuclear chain reaction
The author is a Reuters Breakingviews columnist. The opinions expressed are his own.
By Robert Cyran
NEW YORK, Oct 7 (Reuters Breakingviews) -Artificial intelligence really has gone nuclear. The hulking energy demands of machine learning prompted the United States to sign off on reopening a mothballed atomic energy plant at the infamous Three Mile Island, to follow another one due to be restarted in Michigan next year. These market and government signals have created buzz about a broader comeback for the controversial power source, but there are big economic and project management challenges to overcome before any new reactors will be built.
The U.S. Department of Energy estimates that the country will need 20% more electricity over the next 10 years. It’s a far cry from the 1960s, when consumption doubled, but it would nevertheless represent a dramatic increase from the stagnancy of the previous two decades. Remote facilities used to analyze and store data, and their expanding size, are just one reason for the growth. A giant factory building boom is another. Powering nearly anything that uses fossil fuels, and the surge in AI augur even greater demand.
Generating the necessary supply, and cleanly, will be difficult. The Three Mile Island reactor, for example, will produce more than 800 megawatts of power. Sam Altman, the co-founder of ChatGPT maker OpenAI, talks about building a series of mega-data centers that would consume 5 gigawatts each, equal to six of the units managed by Constellation Energy CEG.O.
To put it all in context, the United States has about 1,200 GW of utility-scale electricity generation, according to the U.S. Energy Information Administration. New power for data centers needs to be available around the clock, and most technology developers prefer it to be green. The feat becomes harder as demand swells.
If one 5 GW data center were ever built, it would require around 10 million solar panels to operate at typical efficiency. The amount of land required, in densely populated areas, alone would be staggering. But it’s the 24/7 component that is becoming even more problematic.
Wind and solar depend on Mother Nature, and while batteries can store this erratic power, costly long-term mismatches between supply and demand will increasingly occur if customers keep clamoring for more. The alternatives are to install polluting gas turbines or to erect five atom-splitting plants producing 1 GW each. These simplistic calculations are one reason Nvidia NVDA.O boss Jensen Huang said recently that it’s impossible to win the AI race without nuclear power.
If he is right, a lot will have to change. Nuclear plants are a huge source of carbon-free energy and far safer than coal, but fears over the long-lasting waste from power plants, and Chernobyl- and Fukushima-like accidents leave the industry less popular than wind or solar power. Thirty-five years of Homer Simpson cartoonishly working as a safety inspector at a nuclear facility probably hasn’t helped the industry’s image either. Worse, investors fret about cost overruns and lengthy delays. Electricity derived from nuclear hasn’t grown since 2006, according to the Energy Institute trade group.
Given the risks involved and the lack of demand for big new sources of power, few utilities were willing to allocate the necessary capital. Those that dared suffered. The Vogtle reactors in Georgia were the country’s most recent. They took some 15 years to build and cost about twice their originally projected $14 billion.
By contrast, capital expenditures for solar and wind projects run at a fraction of the price. They are also added to the grid faster with fewer surprises. The levelized cost of energy, essentially the expense of building and running a power plant over time, is estimated by Lazard analysts at $190 a megawatt hour for new nuclear. It’s less than a third as much for the two most popular renewables. This disparity, and the continued reduction in green energy prices, make solar the default option whenever possible.
Firing up dormant nuclear plants is competitive. Although reactors are costly to build, they are relatively cheap to run. Lazard estimates it at $32 per MWh, on a par with new solar initiatives and half the price of existing coal plants.
Constellation puts the tab for restarting the Three Mile Island facility at $1.6 billion. Safety concerns are minimal, considering it ran for 40 years after the partial meltdown in 1979 that closed its sister reactor. Government tax credits might be worth around $100 million a year. Moreover, Microsoft MSFT.O will buy the output at a rate to be between $110 and $115 a MWh, according to Jefferies analysts, who estimate this would generate a leveraged rate of return for Constellation of more than 30%.
There are only a few other existing, unused U.S. plants. After that, more nuclear means new construction. Existing sites would at least curb some costs, thanks to the available infrastructure and probably faster public approvals. After that, it would require starting from scratch.
For that to happen, utilities would want greater assurances that long-term power demand exists. After all, three dozen approved U.S. nuclear plants were cancelled in the 1980s when growth was slower than anticipated. Builders also would be seeking to borrow billions of dollars at reasonable interest rates. These constraints would depend on companies signing long-term contracts; governments assuming some of the financing risks; infrastructure funds and banks jumping on the bandwagon; and securing greater public support.
There are some encouraging signs: Microsoft’s willingness to commit to buying power, a $1.5 billion Department of Energy loan for the Michigan reactor and an endorsement from 14 of the biggest financial institutions, including Goldman Sachs and Brookfield, of the idea to triple the world’s nuclear power by 2050. It’s more progress in a month than the industry has experienced in a decade, but it also remains decidedly insufficient.
A bigger push would come from standardization. Hiring thousands of workers and teaching them specialized skills for a giant one-off customized project would be a recipe for more frustrated builders and financiers. Agreeing on a model reactor and committing to multiple plants would dramatically reduce construction costs. The miniature versions touted by some entrepreneurs have yet to be proven, meaning that convergence around bigger, established designs is a more likely outcome. Government intervention to force consensus would be ideal, but also is probably politically untenable in the United States.
If everything somehow came together, the Energy Department said last month that the cost of nuclear power might fall as low as $60 per MWh. At that rate, it would be more expensive than solar, but cheaper than natural gas. The amount of coordination required, however, may be as hard to manage as fission itself.
Follow @rob_cyran on X
New US nuclear reactors have all but vanished https://reut.rs/4eTSLnf
Natural gas and renewables dominate US grid https://reut.rs/4eyBccG
Editing by Jeffrey Goldfarb and Pranav Kiran
Related Assets
Latest News
Disclaimer: The XM Group entities provide execution-only service and access to our Online Trading Facility, permitting a person to view and/or use the content available on or via the website, is not intended to change or expand on this, nor does it change or expand on this. Such access and use are always subject to: (i) Terms and Conditions; (ii) Risk Warnings; and (iii) Full Disclaimer. Such content is therefore provided as no more than general information. Particularly, please be aware that the contents of our Online Trading Facility are neither a solicitation, nor an offer to enter any transactions on the financial markets. Trading on any financial market involves a significant level of risk to your capital.
All material published on our Online Trading Facility is intended for educational/informational purposes only, and does not contain – nor should it be considered as containing – financial, investment tax or trading advice and recommendations; or a record of our trading prices; or an offer of, or solicitation for, a transaction in any financial instruments; or unsolicited financial promotions to you.
Any third-party content, as well as content prepared by XM, such as: opinions, news, research, analyses, prices and other information or links to third-party sites contained on this website are provided on an “as-is” basis, as general market commentary, and do not constitute investment advice. To the extent that any content is construed as investment research, you must note and accept that the content was not intended to and has not been prepared in accordance with legal requirements designed to promote the independence of investment research and as such, it would be considered as marketing communication under the relevant laws and regulations. Please ensure that you have read and understood our Notification on Non-Independent Investment. Research and Risk Warning concerning the foregoing information, which can be accessed here.
