Marathon Fusion Says It Made Gold From Mercury at 10x CERN

Marathon Fusion, a San Francisco start-up, said on X on 21 October 2026 that it had turned mercury into gold using fusion neutrons. “Last month, we demonstrated high-rate alchemy: turning mercury into real, stable gold, beating CERN’s record by a factor of 10,” the company wrote, in a post that had about 2.2 million views by the next day.

These are the company’s own claims. No independent verification of the gold result had been published when we read the coverage, and RuntimeWire reports that the X thread “offered no experimental data”. The same post says Marathon is now building a fusion reactor in San Francisco.

Business Pill · GRID COST ALLOCATION EXPLAINED

A short Business Engineer explainer on who pays for grid upgrades. It teaches the general idea only and says nothing about any company or person in this piece.

What Marathon Says It Did

The thread makes three claims. On gold, a follow-up post says: “Mercury transmuted into stable gold, using fusion neutrons, at 10x the record set by CERN in 2025.” On isotopes, the company says that “Last week, we enriched isotopes in the world’s most powerful plasma centrifuge, beating the Soviet record that stood for 40 years.”

The third is a plan: “Now, we’re building a fusion reactor in San Francisco to achieve record power density.” The company’s website lists the reactor’s targets as fusion power of 1kW and a power density of 20 kW/m 2.

Transmutation means changing one element into another by altering the nucleus, which chemistry cannot do. Marathon’s website describes the route: a fusion neutron hits mercury-198, which emits two neutrons and briefly becomes a radioactive isotope before decaying into stable gold-197.

What Was Posted on X

Marathon Fusion’s own post on 21 October, with a video. The company says that last month it turned mercury into stable gold at ten times CERN’s record, that last week it enriched isotopes in what it calls the world’s most powerful plasma centrifuge, beating a 40-year-old Soviet record, and that it is now building a fusion reactor in San Francisco. These are the company’s claims; this piece quotes the post and the thread’s follow-up.

How Strong Is the Evidence

For scale, BusinessToday reports that CERN’s ALICE experiment “estimated that it generated only about 29 picograms of gold during four years of operation.” The posts claim a rate, not a quantity, and do not say how much gold Marathon made. As we read it, ten times CERN’s rate would still be a laboratory amount.

BusinessToday adds that the company shared energy-spectrum plots identifying mercury-197 and gold-197 signatures, and notes that “the plots alone do not establish the total quantity of gold produced, its purity, the efficiency of the process or whether it can be scaled commercially.”

RuntimeWire describes a separate result, Marathon’s lithium-isotope enrichment, and says that “lithium-isotope enrichment result has stronger support, including outside laboratory certification”. It does not say whether that is the plasma-centrifuge result in the thread.

Where the Idea Came From

BusinessToday notes that Marathon’s 2025 research paper described a proposed production method supported by nuclear simulations, and that the new claim of an experimental result is a separate development. Marathon’s website says that its research has shown mercury-198 can be used as a neutron multiplier “while generating thousands of kilograms of stable gold per gigawatt of fusion power per year”.

The same page says gold production “can approximately double the revenue generated by fusion power plants”, putting a gigawatt-scale plant at $223M a year from electricity alone and $482M a year with gold. Those are the company’s projections for a future plant.

On funding, Pulse 2.0 reported a $5.9 million round led by 1517 Fund and Anglo American. The website lists the US Department of Energy among its partners.

The Structural Read

As we read it, the gold claim and the isotope claim carry different weight: RuntimeWire says the thread “offered no experimental data”, and describes a separate lithium-isotope result with outside laboratory certification.

As we read it, gold is the headline but isotopes are the nearer business: BusinessToday reports that Marathon says it has started producing radioactive copper isotopes by irradiating zinc with fusion-generated neutrons.

As we read it, the reactor plan is about neutrons, not grid power: Marathon’s website lists a fusion power target of 1kW for the San Francisco machine.

Three Implications

FOR INVESTORS IN FUSION As we read it, the claim to watch is data: the posts give a rate relative to CERN, and BusinessToday notes the plots alone do not establish quantity or purity.

FOR THE GOLD MARKET BusinessToday reports that the 2025 paper’s estimate of around two tonnes of gold a year per gigawatt of thermal power is a modelled projection, not a demonstrated output.

FOR AI POWER BUYERS As we read it, a 1kW fusion target does not change the power picture behind Nvidia’s statement that over 200 gigawatts of projects are waiting in U.S. interconnection queues.

The Business Engineer Lens

This story maps onto the Business Engineer framework The State of AI Data Centers.

The framework argues: “The structural bottleneck isn’t computing, it’s electrons.”

As we read it, fusion is one of the long-dated answers to that bottleneck, and Marathon’s first reactor targets neutrons for isotopes rather than electricity. The framework was published in December 2025 and does not discuss Marathon Fusion.

What Is Not Established

In the material we read, Marathon had not published the amount of gold produced or an independent measurement; the energy-spectrum plots BusinessToday describes do not establish quantity or purity. The documents we read do not name a reactor site address beyond San Francisco, a construction timeline or permits. None of the sources names an AI company or data-centre operator as a customer or investor.

Connecting the Coverage

Today, in our piece on Nvidia’s cooling claims, we added Nvidia’s statement that “over 200 gigawatts of projects are waiting in U.S. interconnection queues.” Our backlash map today counted local pushback against data centres in 12 places across four countries in three days.

As we read it, Marathon’s reactor is not a power source for that demand: its stated target is 1kW of fusion power, aimed at producing neutrons for isotopes. What’s new is a fusion start-up saying it has moved its gold idea from simulation to a laboratory result.

The Bottom Line

Marathon Fusion says it turned mercury into stable gold at ten times CERN’s 2025 rate and enriched isotopes in a record plasma centrifuge. In the material we read, the gold claim has no published quantity or independent check, and RuntimeWire says the thread offered no experimental data; it says a separate lithium-isotope result has outside certification. As we read it, the test of the claim is the data, not the post.