In late August, amid the blinding New Mexico heat, politicians and investors wielding ceremonial spades broke ground for a new facility that aims to trap the Sun in a bottle. Pacific Fusion, a start-up founded and backed by elite scientists and tech entrepreneurs, is building a nuclear fusion test site in Albuquerque that it hopes will be a crucial step towards generating commercial electricity from the same reaction that powers the Sun. Eric Schmidt, the former Google chief executive and an investor in Pacific Fusion, said on stage that if fusion eventually produced the majority of America’s electricity, “it solves all the problems” created by the energy demands of AI.

That promise of abundant, cheap and clean energy — and of freeing humanity from its reliance on burning fossil fuels — is why US technology groups including Google and Nvidia, billionaires such as Sam Altman, Jeff Bezos and Bill Gates, energy majors and the Trump family’s media group are pouring billions of dollars into private fusion companies. Helion Energy, backed by Altman, aims to supply electricity to Microsoft by 2028. In the first eight months of this year, fusion start-ups attracted a record $3.8bn in private funding, according to Pitch Book, overtaking last year’s total of $3.3bn, which was itself an all-time high.

But while there is broad agreement that fusion science has advanced substantially, experts and start-ups disagree over how soon companies can deliver meaningful power to the grid and whether providing that electricity will make commercial sense. No fusion experiment has yet produced more electricity than the facility consumes, let alone demonstrated a power plant that can operate reliably. Alain Bécoulet, chief scientist of the publicly funded international fusion research project ITER, says connecting a fusion power plant to the electricity grid “will never happen” within the next five to 10 years.

Insiders warn that as competition for funding intensifies, companies are making overly optimistic claims about the cheapness of the electricity they hope to produce. A senior fusion executive who does not wish to be named says it is difficult to estimate costs for a technology that has never operated commercially. Some price pledges amount to companies “just making things up”, the executive says, in order to attract enough investment to continue their research — “the most Silicon Valley-type bullshit you could possibly do”.

The dispute is becoming more pressing as fusion companies move from laboratory experiments towards prototype plants requiring billions of dollars in capital. Developers are increasingly turning to governments, institutional investors and public markets, raising the risk that exaggerated claims could misdirect capital and undermine support not only for long-term fusion research but for other sources of low-carbon energy. Some investors, experts and even some executives fear that overpromising could ultimately damage confidence in the wider field.

One expert describes the risk as a “Theranos moment” for fusion, comparing the enthusiasm around the sector’s most ambitious targets with the excitement that surrounded the Silicon Valley blood-testing start-up before its technology claims unravelled.