San Francisco startup Marathon Fusion says it uses nuclear fusion to turn mercury into stable gold
San Francisco nuclear fusion startup Marathon Fusion said in a statement posted on X on October 21 that it had demonstrated turning mercury into stable gold, at a production rate more than 10 times the CERN-related record. The company said high-energy neutrons produced through nuclear fusion converted mercury-198 into mercury-197, which decays into stable gold-197.
The company did not disclose specific output or costs. Its 2025 preprint paper, which was not peer reviewed, showed in modeling that about 2 tons of gold per year could be produced per gigawatt of thermal power. The report also noted that the newly produced gold contains radioactive isotopes and would need to be stored for several years, estimated at about 7 to 18 years, before it can meet safety standards. CERN previously produced about 29 picograms of gold during Large Hadron Collider operations.
Historical experiments have produced tiny amounts of gold through nuclear transmutation, including Glenn Seaborg's work in the 1980s. Marathon Fusion was founded by Adam Rutkowski, a former SpaceX engineer, and Kyle Schiller, and its reactor power scale is currently far smaller than a commercial gigawatt-scale concept.
Original sources:
Title: Gold from mercury? US startup's nuclear fusion experiment raises a billion-dollar question Publisher: ET Now Publication date: 2026-10-23 URL: https://www.etnownews.com/companies/mercury-to-gold-how-us-startup-marathon-fusion-plans-to-make-lab-grown-gold-article-156391114 Source type: media report Verification status: full text actually read in this review, verification date 2026-10-26 Supported content: supports the company's statement, process description, the claim of more than 10 times the CERN record, and that output and commercial viability were not disclosed
Title: Lab-grown gold: why gold is not the next diamond Publisher: Fractional Publication date: 2026-10-22 URL: https://www.fractional.company/insights/lab-grown-gold Source type: analysis article Verification status: full text actually read in this review, verification date 2026-10-26 Supported content: supports the X statement, no experimental data disclosed, preprint modeling (about 2 tons per gigawatt), radioactive storage requirements, and comparison with mine output
ABAB AI Insight
Marathon Fusion's statement focuses on using neutrons from deuterium-tritium nuclear fusion to achieve nuclear transmutation from mercury-198 to gold-197. This is different from the mechanism of lab-grown diamonds, which involves rearranging carbon atoms; gold, as a single element, can only be produced by changing the atomic nucleus. The company says the rate exceeds the CERN-related record, but it has not provided independently verifiable production data.
Its preprint modeling assumes about 2 tons of gold per year at gigawatt-scale thermal power, while global mine output is about several thousand tons per year. Newly produced gold needs long-term storage to reduce radioactivity. Historically, there have already been trace gold-production experiments through nuclear transmutation (such as Seaborg's and CERN's picogram-level output); this is a company-level demonstration claim rather than the first scientific record.
Observation points include whether reproducible experimental data or peer-reviewed results are released, the actual output and radioactive decay time, and whether other institutions independently replicate it.
ABAB News · Cognitive Law
The company says it turns mercury into stable gold using nuclear fusion neutrons at a rate 10 times the relevant CERN record.
Preprint modeling indicates annual output of about 2 tons of gold per gigawatt of thermal power, but output and costs were not disclosed.
The newly produced gold contains radioactive isotopes and must be stored for several years before it can meet safe-use standards.