Match made in heaven?
Maybe.
Tech bros tend to have confidence, one could argue maybe too much confidence, in high tech fixes to sticky problems. They’ve been having an “oh shit” moment as power demands from AI appear to be going thru the roof.
Above, Diane Olick outlines one of the major initiatives in to small nuke tech, from Amazon.
Andrew Ross Sorkin, asks the right questions below. Will NIMBYism be a problem?
Maybe not, because plants can be built at existing sites.
Possibly more on point, will there be first-of-a-kind stumbles and costs?
I can only wish them lots of luck.
Google will back the construction of seven small nuclear-power reactors in the U.S., a first-of-its-kind deal that aims to help feed the tech company’s growing appetite for electricity to power AI and jump-start a U.S. nuclear revival.
Under the deal’s terms, Google committed to buying power generated by seven reactors to be built by nuclear-energy startup Kairos Power. The agreement targets adding 500 megawatts of nuclear power starting at the end of the decade, the companies said Monday.
The arrangement is the first that would underpin the commercial construction in the U.S. of small modular nuclear reactors. Many say the technology is the future of the domestic nuclear-power industry, potentially enabling faster and less costly construction by building smaller reactors instead of behemoth bespoke plants.
“The end goal here is 24/7, carbon-free energy,” said Michael Terrell, senior director for energy and climate at Alphabet’s Google. “We feel like in order to meet goals around round-the-clock clean energy, you’re going to need to have technologies that complement wind and solar and lithium-ion storage.”
The nuclear-power industry’s fortunes are increasingly getting hitched to Big Tech. Power demand is rising in parts of the U.S. for the first time in years, much of it driven by the need to build more data centers for AI. That has sent the tech industry on the hunt for massive amounts of energy.
Last month, Constellation Energy and Microsoft struck a deal to restart the undamaged reactor at Pennsylvania’s Three Mile Island, the site of the country’s worst nuclear-power accident. Earlier this year, Amazon purchased a data center at another Pennsylvania nuclear plant.
The 500 megawatts of generation that would be built by Kairos for Google is about enough to power a midsize city—or one AI data-center campus.
The agreement answers questions that have bedeviled smaller-reactor designs: What customer would pay the higher price for a first-of-a-kind project? And who would order enough to get an assembly line started? The concept, which remains to be proven, is that building the same thing over and over in a factory would drive down costs.
Kairos plans to deliver the reactors between around 2030 and 2035. Financial terms weren’t disclosed, but the companies entered into a power-purchase agreement, similar to those used between corporate buyers and wind- and solar-energy developers.
The project site—or whether there could be reactors at multiple locations—hasn’t been determined, the companies said.
Google would have data centers somewhere in the region near the Kairos reactors, but it hasn’t been determined whether they would receive power directly from the nuclear plants or from the grid. Google could count the addition of nuclear power toward meeting its carbon-reduction commitments.
Instead of water, which is used in traditional reactors, the Kairos design uses molten fluoride salt as a coolant. The units for Google will include a single 50-megawatt reactor, with three subsequent power plants that would each have two 75-megawatt reactors, Kairos said. That compares with about 1,000 megawatts at reactors at conventional nuclear-power plants.
Kairos will have to navigate complex approvals through the U.S. Nuclear Regulatory Commission, but already has clearance to build a demonstration reactor in Tennessee, which could start operating in 2027.
Kairos has a manufacturing development facility in Albuquerque, N.M., where it is building test units. They don’t have nuclear-fuel components but are something of a practice run at building and operating full-size plants to test systems, components and the supply chain.
Mike Laufer, chief executive and co-founder at Kairos, said the demonstration project and the Albuquerque plant are helping the company avoid spiraling costs, a pitfall of the conventional nuclear industry.
Red flag here is that the proposed plants have never been successfully deployed, and all the cost projections are based on everything going right not just with the final product, but with an extremely complex manufacturing process that has yet to even identify what all the inevitable wrinkles and hiccups are.
Tech Giants Bring Big Bucks for Small Nukes
♫♪ I like big bucks and I cannot lie!
Computing looks like the prime example of an industrial process that does not rely on constant power. When you have a workload distributed over a myriad of processors and you have to shut some of them down because you don’t have enough power, all the threads and processes just run slower, with mission critical ones having priority access. This should make nuclear plants serving AI centers suitable for being diverted in case of blackouts on the rest of the grid.
So if wind and solar are so cheap and fast to install while nuclear is so expensive and takes so much time to build, why are the tech companies so interested in nuclear? Certainly, nuclear’s high capacity factor allows for fuller utilization of computing resources, but I think it’s more than that. Tech companies spending their own money can see through the hype and bogus LCOE stats put out by Lazard, not to mention the hysteria and regulatory monkeywrenching of nuclear.
As so often happens, nook boosters justify ongoing nook boosting, & nook operation, plans, & desires by claiming, explicitly or not, humans always behave rationally. All we have to do to see that that’s false is to look at their own behavior. And their ignoring of all the rest of human behavior is of course flabbergasting at the same time it’s utterly predictable.