Here’s the conundrum: if fusion wants to fulfill the promise of providing clean power to the world on a relevant and timely scale, it must a) demonstrate a working prototype ASAP, and b) swiftly proceed with widespread deployment. The latter necessitates exponential growth until more than 100 fusion power plants are operational worldwide every year. However, since fusion power plants are large, complex nuclear devices, the construction time, i.e., the time between final investment decision and full power operation, will be lengthy (~10 years). Exponential growth could necessitate the start of Generation 2 (Gen2) construction while Gen1 is not yet operational. However, to realize exponential growth, the volume of Gen2 must be an order of magnitude larger than that of Gen1. Consequently, either Gen2 is very similar to Gen1 – resulting in limited learning – or unproven concept improvements must be implemented on a large scale, which entails significant technical and economic risks. If Gen1 is one tokamak, would Gen2 consist of ten stellarators? And would Gen3 involve one hundred ICF devices? The combination of exponential growth and an inevitably long intergeneration time impedes effective learning and innovation. The remedy is to concurrently develop many concepts for Gen1 and then down-select during exponential growth. This is costly, so the question is how many concepts should be developed concurrently. To answer that question, a simple probabilistic calculation is set up, which considers the expected value of future revenues of fusion power plants, as well as the fact that future revenues are worth less today the farther they are away (discounting). The (analytical) calculation indicates that such a ‘value-led’ strategy requires investment in accelerating the development of fusion power and the parallel testing of all reasonable concepts. This appears to be precisely what we observe today, with the sharp increase in fusion funding - both private and public – and the proliferation of companies working on different concepts.
[mehr]
