The claim comes from Marathon Fusion, a San Francisco fusion startup. On October 21, 2026 it posted on X that it had turned mercury into stable gold using fusion neutrons, at what it says is ten times the rate of CERN's 2025 result. The X thread is the company's own account.
The nuclear route is plausible. A 14 MeV neutron from deuterium-tritium fusion knocks a neutron out of mercury-198, making mercury-197, which decays to stable gold-197. The company laid this out in a 2025 arXiv preprint, which is a simulation and hasn't been peer reviewed.
The evidence is thin. The thread shows one gamma-ray spectrum with the mercury-197 and gold-197 signatures. It doesn't say how much gold was made, its purity, the efficiency, or whether it can scale. Nobody has independently verified it. For scale, CERN's ALICE team counted about 29 picograms of gold from the LHC's 2015 to 2018 run, so even ten times that is a few hundred picograms.
On gold prices, the amounts are far too small to matter, and the company's paper notes that freshly made gold would be radioactive and need years of storage before sale. Coverage by BusinessToday gives the same picture: the physics is established, the scale is unproven.
🕒 As of October 2026
