Conventional nuclear plants are built individually, on site, over a decade, and routinely cost several times the original estimate. Small modular reactors propose fixing that by manufacturing units in a factory and assembling them on site.
Why the approach could work
Because factory production has a learning curve and site construction does not. Every unit built in a controlled environment is cheaper than the last. Every plant built on a unique site starts over.
Smaller units also reduce the capital at risk. A project that costs a fraction of a conventional plant and takes a fraction of the time is financeable by parties that cannot contemplate a decade-long commitment of billions.
And smaller reactors can use passive safety – shutting down and cooling without power or operator action, through physics rather than systems.
Why it has not happened
Regulatory frameworks were written for large individual plants. Approving a design once and building many copies requires a different process, which regulators have been constructing while applicants wait.
The economics depend entirely on volume. The first unit is expensive; the hundredth is cheap. Getting from one to a hundred requires orders that nobody places until the price falls, which requires the volume.
Several announced projects have been cancelled after costs rose during design.
Where deployment is real
A small number of units operating or under construction, mostly in countries with state-backed programmes. Interest from data centre operators seeking continuous carbon-free power is the newest and most serious demand signal.
The honest assessment
A plausible solution to a real problem, not yet demonstrated at the scale its economics require. The next few years determine whether the learning curve materialises.