Physics / mechanism
A molten salt reactor is a fission generator, so it sits inside the same physical envelope as other fission plant: energy released by neutron-induced fission of heavy nuclei, removed by a coolant loop and converted to electricity. The single available source is a general primer on nuclear fission rather than a reactor-specific technical document ref, and it does not describe salt chemistry, fuel dissolution, operating temperature, neutron spectrum, or breeding behaviour. Those parameters are therefore not evidenced on this page.
What the source does establish is the framing that matters commercially for any fission concept, including molten salt designs: levelised cost of electricity is treated as the weak point and dispatchable baseload output as the strength ref. The author’s stated conclusion is that nuclear is a partial rather than a full answer to data centre power demand, with scepticism concentrated on timeline and cost rather than on physics ref.
Any molten salt reactor claim should therefore be assessed against those two axes first, since a novel coolant and fuel form changes the licensing path and the first-of-a-kind capital cost, both of which feed directly into LCOE and time to first power.
Competitive landscape
The sources do not support a comparison between molten salt reactors and adjacent fission architectures such as light water small modular reactors, high-temperature gas-cooled reactors or sodium-cooled fast reactors. At the level the source does support, molten salt reactors compete inside the broad category of nuclear fission for hyperscaler power procurement, against which the source’s own verdict is scepticism on cost and delivery timeline ref. Demand-side interest is documented: Google and Microsoft are described as actively pursuing nuclear supply, with an Amazon arrangement blocked at the time of writing and Meta also seeking nuclear power ref.
Evidence base
- A November 2024 investor primer on nuclear fission concludes that nuclear is “probably not the answer” to data centre power demand, but is a partial answer ref.
- The same source summarises its findings as “LCOE bad, baseload good”, identifying cost per MWh as the principal weakness and firm output as the principal strength ref.
- Scepticism in that source is directed at timeline and cost rather than at the underlying fission physics ref.
- As of November 2024, Google and Microsoft were pursuing nuclear power agreements, an Amazon arrangement had been blocked, and Meta was also seeking nuclear supply ref.
Frontier (open questions)
- What operating temperature, salt composition and fuel form does a given molten salt design use, and what corrosion rate has been measured in a loop at those conditions over what duration?
- What is the projected LCOE of a molten salt reactor at nth-of-a-kind scale, and how does it compare with the fission cost figures that the primer judges unfavourable ref?
- What licensing pathway applies to a liquid-fuelled reactor in the target jurisdiction, and what is the demonstrated time from application to construction permit?
- Has any molten salt reactor delivered electricity to a grid or to a data centre load, and at what capacity factor?
Synthesised 2026-08-31 from 1 KB sources by the resynth pipeline; citations are KB source slugs.