Could Lab-Made Gold Crash Prices? The Real Test Is Cheap, Scalable Supply
Marathon Fusion's mercury-to-gold claim raises a real long-term supply question. But a claimed experiment, simulated plant output and economically saleable bullion are three different milestones.
Listen to this update
Ready when you are.
Quick takeaway
Making gold is possible nuclear physics; making cheap gold at market-moving scale is a different challenge. Euronews reports that Marathon Fusion claims to have converted mercury into stable gold in a September experiment announced October 21. Independently verified output and cost are not established in the reviewed materials. Its industrial production figures remain simulations. The valuation question is whether repeatable, affordable and acceptable new supply could eventually weaken gold's scarcity premium—not whether a laboratory announcement proves an imminent diamond-style price collapse.
What is claimed—and what is still missing
Euronews' October 23 report describes Marathon's announcement of a mercury-to-stable-gold experiment and a claimed tenfold improvement over a CERN benchmark. This is a claim of a physical demonstration, not merely the company's earlier computer-modeling proposal. It remains a company claim rather than an independently verified commercial result.
The reviewed public material does not establish the recovered quantity, reproducibility, independently assayed product composition or a complete production cost. Those are decisive gaps when the question shifts from whether atoms can be transformed to whether the process could supply the bullion market.
The hero image is an editorial illustration, not a photograph of Marathon's experiment.
Gold is an element; diamonds are a form of carbon
GIA describes laboratory diamonds made through high-pressure, high-temperature growth or chemical vapor deposition. These processes arrange carbon into diamond crystals; they do not turn one element into another. Laboratory and natural stones have essentially the same underlying material properties, while growth features can allow specialist identification.
Marathon's proposed route changes atomic nuclei: energetic neutrons convert mercury-198 into mercury-197, which can decay by electron capture to stable gold-197. A successfully isolated, pure stable product would be real gold, not a gold-coloured imitation.
Material equivalence does not imply economic equivalence. Diamond manufacturing shows why an affordable substitute can disrupt scarcity pricing, but its cost curve and price history cannot simply be assigned to nuclear gold production. Nor does authentic gold establish that every sample is immediately pure, nonradioactive or impossible to trace.
A record comparison is not a production statement
CERN's ALICE report describes about 29 picograms of gold nuclei created during the LHC's 2015–2018 Run 2. That is 29 trillionths of a gram, using a lead-ion mechanism; the energetic nuclei subsequently fragmented after striking accelerator equipment.
Marathon proposes a different route and stable target isotope. Without a clearly specified, comparable measurement, the claimed tenfold improvement cannot establish its recovered mass, purity or cost. Multiplying CERN's cumulative 29-picogram total by ten would not be a valid estimate of Marathon's product.
The useful next evidence is an independently documented production rate and recovered yield—not the size of a multiplier applied to a different experiment.
Tonnes per year are still a model, not a factory result
Marathon's technical paper uses neutronics simulations to estimate roughly two tonnes of gold per gigawatt of thermal fusion power per year. Thermal power and electrical output are different denominators. The often-cited five tonnes per gigawatt of electricity assumes approximately 2.5 gigawatts of thermal power; it is not a demonstrated plant yield.
Commercial output would require reliable plant operation, enriched feedstock, suitable materials, recovery systems and safe handling. The design must also support the fusion fuel cycle. A promising calculated reaction rate does not establish an operating plant's availability or an all-in cost per saleable ounce.
Stable gold does not mean an instantly saleable product stream
The paper's Section 4.1.2 models cooling periods of roughly seven to eighteen years under different assumptions and criteria because other radioactive isotopes can accompany stable gold-197. Feedstock purification can reduce the delay. This is not a universal eighteen-year requirement or independently established regulatory approval.
For valuation, the gap between production and sale matters. Purification, storage, financing and eventual market acceptance belong in the cost calculation alongside the reactor. Metal produced today but monetized much later is not economically equivalent to immediately deliverable bullion.
How much new supply would matter?
Euronews illustrates one scenario using 100 plants, each with one gigawatt of electrical output and five tonnes of annual gold production. That arithmetic produces 500 tonnes per year. It is a hypothetical fleet, not an announced deployment schedule or an output forecast.
Against the World Gold Council's published 2025 estimates, 500 tonnes would equal about 13.6% of mine production of 3,671.6 tonnes, or about 10% of total annual supply of 5,002.3 tonnes, including recycling and net producer hedging. These are different supply denominators, not interchangeable measures.
The WGC also estimates 222,600 tonnes of above-ground gold at end-Q2 2026. The hypothetical annual 500 tonnes is about 0.22% of that stock. A stock comparison does not make additional supply irrelevant: not all existing gold is offered for sale, and prices can respond to changing marginal flows and expectations. None of these percentages translates mechanically into a percentage price decline.
The real threat is cheaper scalable supply—not the word laboratory
Our valuation interpretation is conditional. If producers eventually demonstrate large, repeatable volumes of acceptable gold at a competitive full cost, the prospect of a more elastic supply base could weaken the scarcity premium. Expectations could adjust before all of that metal physically reaches the market.
The opposite mistake is to assume commercial success is inevitable because transmutation is physically possible. Production economics must survive engineering, utilization, purification, delayed cash receipts and the gold price itself. If a project's economics rely on gold sales, a lower gold price would also reduce that revenue—a feedback, not a one-way path to unlimited cheap bullion.
This announcement therefore warrants monitoring as a long-run supply risk, not treating it as proof of an imminent glut or dismissing it as fake science. It establishes neither a diamond-style crash forecast nor a guaranteed floor under gold prices.
Independent verification of recovered mass, isotope composition and reproducibility.
Sustained production at meaningful scale, not just a record-rate comparison.
A transparent all-in cost including feedstock, recovery, storage and financing.
A credible path to product acceptance and sale, with radiological requirements addressed.
An industrial deployment plan whose gold output remains explicitly a forecast until demonstrated.
