The post points to a January article claiming Meta will back deployment of eight TerraPower Natrium reactors, each rated at 345 megawatts, with thermal storage that could raise site output during peaks. Natrium is not a conventional light-water reactor. It is a sodium-cooled fast reactor paired with molten-salt heat storage, and the first Wyoming unit is still a demonstration plant scheduled for the early 2030s. That framing drove most of the reaction. People read the announcement less as "eight plants are coming" and more as "Meta is putting money behind a nuclear startup and wants future access to power for AI." The missing numbers mattered. With no disclosed dollar amount or contract structure, several readers saw it as impossible to tell whether this is meaningful project finance, a soft offtake commitment, or mostly a headline.
The strongest consensus was that the timeline is the story, and it looks implausible. Commenters pointed out that even countries with recent nuclear construction muscle like South Korea and China still take close to a decade for new reactors. TerraPower has not yet brought a single Natrium unit online. That makes an eight-reactor fleet by the early 2030s read like aspiration, not execution. Regulatory approval did not reassure skeptics much either. The company has a construction permit for the Wyoming project, but people noted that finishing the plant, proving it was built to spec, and clearing the operating process still take years. Others pushed back on the idea that this is just ordinary plant inspection, because
first-of-a-kind nuclear projects tend to get stuck in exactly those late stages.
The technical discussion was more grounded than the hype. A few readers corrected confusion in the article and comments about the design. Natrium uses liquid sodium as coolant and
molten salt for energy storage, not a molten-salt
thorium reactor. That distinction matters because sodium fast reactors carry a long history of difficult materials and maintenance problems. People cited Superphénix and Monju as reminders that hot sodium systems are reactive, corrosive, and hard to operate reliably over time. Even commenters rooting for advanced nuclear said the safety case is not the main concern here. The bigger issue is whether a first-of-a-kind sodium design can be built and maintained cheaply enough to become a repeatable commercial product.
Meta itself was barely the mystery to most readers. The answer for why it can spend this aggressively was simple: its ad business throws off cash, and AI spending is not limited to selling a chatbot. Commenters pointed to internal use, recommendation systems, ad targeting, ad generation, and the broader push to secure power for data centers. The mood was cynical about both AI and corporate PR, but still mildly positive on one narrow point: if AI
capex ends up underwriting new firm low-carbon power, that is a better miss than another metaverse-scale money burn.