Question for Virginia Policymakers: “So, if new data centers are unlikely to be powered by SMRs, fusion or space solar for the next few decades, what will they be powered by?”

Over the past few weeks in Virginia, we’ve seen the release of a major new JLARC report, a breathless press release from Gov. Glenn Youngkin (with enthusiastic quotes by numerous leading Virginia Republicans and Democrats), and a superb blog post by energy expert Ivy Main (“The data center energy crisis is now official“) on the interrelated issues of data centers – and, specifically, the rapid growth of data centers and the massive amounts of energy they are projected to consume – and how to power those data centers in an economical, environmental,  sustainable and technologically feasible way. A few highlights from the JLARC report include:

• “Northern Virginia is the largest data center market in the world”
• “During the presentation, commission members were visibly shocked as they learned that a 1,000-megawatt data center campus (like the one in my district) requires more electrical power than the entire output of the Lake Anna nuclear plant. If we don’t act now, our energy infrastructure, environment, and local neighborhoods will continue to bear the brunt of this unchecked growth.” – Del. Josh Thomas
• “Building enough infrastructure to meet growing data center demand will be difficult” (that’s an understatement!)
• Data center energy demand growth could lead to increases in transmission charges for residential customers, financial risks to utilities…
• According to JLARC, “small modular nuclear” technology isn’t expected to start coming online until at least 2035.

And from Ivy Main’s blog post:

• “on the threat to Virginia’s energy supply, JLARC is blunt: Building enough infrastructure to provide electricity for even just half the data centers projected for development across the state will be difficult, requiring far more generating facilities than are under development today.”
• “As for the current policy of allowing completely unconstrained data center growth – indeed, subsidizing it as we do now with tax exemptions to the tune of nearly a billion dollars per year – JLARC notes we are headed for a tripling of the state’s electricity usage over just the next decade and a half.  Meeting that much demand, says the report, would be ‘very difficult to achieve,’ even if the state jettisoned the carbon emission limits imposed by the Virginia Clean Economy Act (VCEA). For those of you unfamiliar with the vocabulary of bureaucrats, ‘very difficult to achieve’ is a term of art that translates roughly as, ‘This is nuts.'”
• “Legislators tend to be optimists, and they are already betting on bright, shiny objects like SMRs, fusion, and anything else not close enough for its costs and drawbacks to be fully evident.”

The bottom line is that the growth of data centers, as well as the growth of artificial intelligence (AI) is driving *massive* demands for electricity, but to date, nobody’s really figured out where all that power is going to come from. (according to this article in the Virginia Mercury, Dominion Energy’s massive, 176-turbine offshore wind project off of Hampton Roads “will produce enough electricity to meet the needs of about 600,000 homes, or just a bit more than the energy that will be required to power a single data-center mega project that is currently being developed in Hanover County.” Wow, that is just wild. The ENTIRE offshore wind development in Virginia would only be enough to power a SINGLE massive data center? Crazy. And that raises the question, how are we EVER going to power these massive power-sucking projects? Hmmmm…

Hence, it was eyebrow-raising (to put it VERY mildly) to see Gov. Glenn Youngkin’s December 17 press release – with supportive quotes by numerous leading Virginia Democrats and Republicans, that “Commonwealth Fusion Systems, the largest private [nuclear] fusion company, will make a multi-billion dollar investment to build the world’s first grid-scale commercial fusion power plant at the James River Industrial Center in Chesterfield County, Virginia, on a site owned by Dominion Energy.” Honestly, when I first saw the press release, I was wondering if it was some sort April Fool’s-style joke, or maybe just confusing nuclear FUSION with nuclear FISSION, or…something?

But nope, it’s real – Youngkin et al actually seem to think that nuclear FUSION power is right around the corner, and that – as Youngkin breathlessly, embarrassingly put it – “This is an historic moment for Virginia and the world at large.” Except that it almost certainly is NOT either of those things. In fact, according to Scientific American, “fusion plants might be feeding power into the grid by around 2050 and then could become steadily more important to the energy economy in the second half of the century, especially post-2060.” And, according to the U.S. Department of Energy, the optimistic/aggressive case is that commercially viable nuclear fusion power MIGHT – if all goes well – start coming online in the late 2030s or 2040s. So much for nuclear fusion powering Virginia data centers for a long while to come (decades? and even then, will it be economically competitive with solar, wind, etc?).

But then why are so many Virginia politicians so enthusiastic about the prospects for nuclear fusion and/or small modular nuclear reactors (another technology that appears to be many years away from technological and economic viability)? And why are so many media outlets – pretty much all of them – taking Youngkin et al’s optimism about nuclear fusion power (and SMRs) at face value? At least two possible answers spring to mind: 1) most politicians and journalists are not, to put it politely, experts when it comes to energy in general, and nuclear power technology specifically; 2) people are desperate for a feasible, environmentaly and economically sustainable technological solution to the incredible challenge of fueling the growth of data centers and AI, particularly here in Virginia, given that “Virginia hosts the largest data center market in the world and is home to more than 35% (~150) of all known hyperscale data centers worldwide” (plus, this is growing rapidly).

The problem, of course, is that – as the saying goes –  “wishing doesn’t make it so.” So sure, it would be GREAT if we had a viable, economically competitive and environmentally sustainable solution to the challenge of powering data centers and AI. But if we’re counting on nuclear fusion and/or SMRs to actually provide that solution in the short-term, it’s not looking great at the moment. For a lot more on this topic, see this brilliant post by  Michael Liebreich, one of the world’s leading clean energy experts. According to Liebreich:

• “The most powerful tech titans in the world have been humbled by the realization that their plans for world domination could be stymied by something as prosaic as electricity – and have embarked on a land grab for whatever sources of dispatchable power they can, triggering something of a gold rush.”
• It’s recently “dawned on everyone that the rate-limiting factor in the growth of AI was not going to be compute power, it was going to be electrical power.”
• “Training a generative AI model requires power. A lot of power. It also requires that power to be concentrated in one location…”
• “For the US, I expect data-center capacity will somewhat more than double by 2030, adding around 30GW, and the rest of the world will add no more than 15GW.”
• “The challenge lies in the nature of the additional demand: it will be highly localized, and it must be available 24/7. In addition, it must be clean.”
• “When you combine a need for large concentrated new demand for dispatchable power with extreme pressure on emissions, it is perhaps inevitable that the hyperscalers would look to nuclear power.”
• “NuScale, once the US’s closest SMR company to commercialization, started by promising $58/MWh, but had to cancel its first projects when this was revised up to $89/MWh. But that was after taking into account $30/MWh from the Inflation Reduction Act and a $1.4 billion direct subsidy, so the real figure, at least five years before commissioning and in today’s money, is $140/MWh.” 
• “If it looks like nuclear fission might not be the answer for your multi-billion-dollar AI data center, there’s always fusion… I will be the first to celebrate if he’s right. But I find it hard not to be reminded of PG&E’s 2009 PPA for 200MW of power from space-based solar company Solaren by 2016.”
• “So, if new data centers are unlikely to be powered by SMRs, fusion or space solar for the next few decades, what will they be powered by?”
• “In the end, the tech titans will find out that the best way to power AI data centers is in the traditional way, by building the same generating technologies as are proving most cost effective for other users, connecting them to a robust and resilient grid, and working with local communities. More, in other words, of what the best of them are already doing.”

Anyway, there you have it: according to Michael Liebreich, who knows more about this stuff than pretty much anyone (let alone any Virginia politician, lol), there’s no magic bullet – whether SMRs, nuclear fusion, dilithium crystals/warp drives (haha) or whatever else – to provide sufficient energy to power data centers and AI, let alone to power them economically and in an environmentally sustainable manner. So sure, let’s keep investing into R&D on nuclear fusion, SMRs, etc., but let’s NOT expect that these will come to fruition in a scalable, economically competitive manner, anytime soon (more likely, we’re talking the 2040s or 2050s, and even then, who knows if fusion or SMRs will be cost competitive with solar, wind, geothermal, etc.).