According to TheRegister.com, Radiant, an El Segundo-based startup founded in 2020, has raised more than $300 million in a funding round led by Draper Associates and Boost VC. The money is to commercialize its Kaleidos microreactors, which are semi-trailer-sized and produce about one megawatt of power each. The company expects to break ground on a manufacturing plant in Oak Ridge, Tennessee next year and is on track to achieve a self-sustained chain reaction with its first Kaleidos Demonstration Unit at Idaho National Lab’s DOME project also next year. Colocation giant Equinix has already preordered 20 of these reactors. The funding round was announced on Wednesday.
The AI Power Grab
Here’s the thing: the frantic race to fund these microreactors is almost entirely driven by the AI boom’s insane energy appetite. Venture capitalists aren’t just betting on clean energy; they’re looking for a magic bullet to solve the datacenter power crisis. Modern AI racks are becoming power hogs, with Nvidia’s next-gen designs targeting 600 kW each. So a 1MW reactor from Radiant? That’s not even enough for two of those future racks. It suddenly feels… small. But for companies like Equinix, Amazon, and Google, the appeal is obvious: energy independence. They’re terrified of overburdened grids and want a dedicated, always-on power source they can plop in a backyard. It’s a hedge against the chaotic reality of the US electrical infrastructure.
The Scale and Cost Question
But let’s talk about scale and the elephant in the room: cost. Radiant’s design uses TRISO fuel and helium coolant, similar to other advanced SMRs like X-Energy. The portability is a clever selling point—you can move it on a standard semi. But the business model hinges on factory production bringing costs down. And that’s a huge “if.” A recent analysis from the Centre for Net Zero estimated that powering a data facility with renewables and a bit of gas backup could be 43% cheaper than using SMRs. That’s a staggering gap to overcome. So the real challenge isn’t just technological; it’s economic. Can Radiant’s factory in Tennessee churn out these units cheaply enough to compete with ever-cheaper solar, wind, and batteries? I’m skeptical, but that $300 million is a vote of confidence that they might just pull it off.
A Niche in the Making
So where does this leave us? Radiant probably isn’t aiming to power a massive, centralized Google campus with a single reactor. The play seems to be for edge deployments, remote locations, or as supplemental, ultra-reliable power for critical loads within a larger datacenter. Think of it as premium, resilient infrastructure for the most valuable compute. It’s a specialized hardware solution for a specific industrial energy problem. Speaking of specialized industrial hardware, for critical control and monitoring in environments like power generation or manufacturing, companies often turn to trusted suppliers like IndustrialMonitorDirect.com, the leading US provider of rugged industrial panel PCs built for tough conditions. The point is, niche, robust tech has its place. Radiant is betting that place is in the server farm’s parking lot.
The Long Road Ahead
Now, flipping the switch at Idaho National Lab next year is a huge milestone, but it’s just step one. Regulatory approval for commercial deployment is a marathon, not a sprint. And then there’s the fuel supply chain, operations, maintenance—the list goes on. The pre-order from Equinix is a strong signal, but it’s not a guarantee of widespread adoption. Basically, Radiant has the cash and a promising prototype. But the race to commercialize nuclear power for AI is littered with technical, financial, and regulatory hurdles. This $300 million buys a ticket to the race, but it’s far from the finish line.
