November 21, 2026
Heard in AI

Marathon Fusion says it made gold from mercury, beating CERN's record

Marathon Fusion says it turned mercury into stable gold at ten times CERN's record, but the company post that Peter Diamandis read on Moonshots did not say which CERN figure it beat. On the same episode, Alexander Wissner-Gross said he hopes transmutation ends gold's scarcity premium.

A briefing reports one development at a point in time. We may correct or clarify it later; a new development gets a new briefing. How our formats work

Based on Moonshots with Peter Diamandis, episode published October 23, 2026 (recorded October 23, 2026)

For thousands of years, alchemists tried to turn ordinary metals into gold. Peter Diamandis, the founder of XPRIZE and host of Moonshots with Peter Diamandis, opened the episode by saying a San Francisco startup claims to have done it, turning mercury into "real, stable gold." Diamandis called it "the age of abundance." Alexander Wissner-Gross, a computer scientist and founder of Reified, called it "the dawn of femtotech."

The episode was recorded and published on October 23, 2026. The panel spent less time on the physics than on a question about money: what cheap gold would do to people who treat gold as the safest place to keep their wealth.

What the company claims

The startup is Marathon Fusion. Diamandis read from a post that he said the company had put up "this morning or last night": "last month, we demonstrated high rate alchemy turning mercury into real stable gold, beating CERN's record by a factor of 10."

Diamandis explained the business idea this way: line a fusion reactor's wall with mercury, and the plant produces clean energy and gold at the same time. Fusion is the process that powers the sun. Light atomic nuclei are forced together, and the reaction releases energy and fast-moving neutrons. According to Diamandis, a 2025 paper by Marathon estimated about two tons of gold per year from a gigawatt-scale plant, which could double the plant's revenue.

The hosts then played a company video. In Marathon's own account, it has spent four years building fusion technology in downtown San Francisco and plans to make its first reactor available to partners in 2027 for producing medical isotopes and testing materials. The video walked through the design. Lithium is melted and evaporated onto a solid plate, and that layer is loaded with hydrogen isotopes to serve as a target. Deuterium, a heavy form of hydrogen, is then ionized and accelerated through an electric field into the target, which produces fusion. Marathon's description of this beam-driven design says the reactions occur in a thin surface layer of the target and release helium and high-energy neutrons.

The video made several further claims. It said the company expects to double its fusion power density every month over the next year. It said Lawrence Livermore National Laboratory ran a similar reactor for more than 25,000 hours in the 1980s. It said demand for radiopharmaceuticals, medicines built from radioactive isotopes, "has gone exponential." And it said recent experiments had produced rare medical isotopes and tritium and had "validated" a pathway to producing electricity and gold together. These are the company's own claims.

"We've got fusion reactors in downtown San Francisco," Diamandis said. "You're walking by them. You don't even know."

How neutrons make gold

Wissner-Gross explained the physics. Marathon's process bombards mercury-198, one particular form of mercury, with neutrons. The incoming neutrons interact with the mercury nuclei, and some of those nuclei are transmuted into gold. (Mercury is element 80 and gold element 79, so the two sit side by side on the periodic table.)

He tied this to a question Dave Blundin, founder of Link Ventures, had asked earlier in the episode: how is gold actually made? Most of the gold on Earth, Wissner-Gross said, is currently thought to come from neutron stars and collisions between them. Neutron interactions are "not terribly high yield," he said, but they are a real way to produce gold.

That is where the "femtotech" label comes from. A femtometer is a millionth of a nanometer, about the size of an atomic nucleus. Wissner-Gross defined femtotech as "technology at the length scale of the atomic nucleus." He said the show talks too little about nanotechnology, and even less about this.

How small CERN's record is

The benchmark Marathon claims to have beaten comes from a very different machine. In May 2025, CERN reported that its ALICE experiment had measured lead turning into gold inside the Large Hadron Collider. When lead nuclei pass very close to one another without colliding, their electromagnetic fields can knock out three protons, which turns lead (element 82) into gold (element 79). ALICE measured a peak rate of about 89,000 gold nuclei per second. Over the collider's entire second run, from 2015 to 2018, the four main experiments together produced about 86 billion gold nuclei, a total of roughly 29 picograms, or 29 trillionths of a gram. None of it survived: the nuclei were moving so fast that they broke apart when they hit parts of the accelerator.

The post Diamandis read did not say which CERN figure Marathon beat. If it was the peak rate, ten times that is still under a million gold nuclei per second, a stream far too small to weigh in ordinary terms. The two tons per year Diamandis cited describe a hypothetical gigawatt-scale power plant, not the demonstration the company announced.

The argument about gold

For Wissner-Gross, the appeal was what transmutation could do to gold's reputation. He said he sometimes speaks at family-office events, where wealthy families manage their own investments, and "there's always a gold bug in the audience somewhere." He described them as people pushing gold on the premise that the world is ending, urging listeners to move all their assets into gold. He added that he was not giving investment advice, then said: "they're always wrong."

He accepted that gold is useful in manufacturing. His target was the belief that gold is permanently scarce and therefore a great long-term store of value. He pointed to diamonds, saying De Beers and the diamond industry were driven to the edge of bankruptcy by synthetic stones. He said he would love to see Marathon, another company or a whole transmutation industry "drive all these gold bugs off the cliff" and make gold "too cheap to meter."

The same argument had come up earlier in the episode in another form. Diamandis asked Wissner-Gross why he was so against crypto and Bitcoin. Wissner-Gross said he is not against crypto as such, but he considers most crypto activity, and Bitcoin as an asset class, an unproductive use of capital. Diamandis asked whether gold was any better. Wissner-Gross said gold can be a productive asset when it is used to make useful things, but much of it is held "because it's shiny or because it has some historic cultural relevance." He pointed to Blundin as "exhibit A," and Blundin protested that he was going to melt "these" down. Wissner-Gross said he does not see that as a productive use of capital, which he called scarce "for the moment at least, mid-singularity," his term for the current stretch of fast, AI-driven change.

The discussion also heard a defense of crypto. It conceded that much of crypto is unproductive and full of "dodgy" projects, but held that blockchain had solved the Byzantine generals problem, the long-standing puzzle of how parties who cannot fully trust one another can agree on one shared record. On that view, authentication can now be decentralized, and decentralization is needed for abundance. Wissner-Gross's reply: "it's so magical that it's unproductive." Near the end of the gold segment he put the two side by side: "It is just like Bitcoin, I would argue. It is largely a waste of capital."

The aluminum precedent

Diamandis offered history as evidence that a precious metal can lose its premium. He told the story of a dinner at Napoleon's at which every guest ate with gold utensils except the king of Siam, who was served with aluminum ones. He added that the Washington Monument is capped with aluminum because it was "the rarest element at that time." Today aluminum is the foil wrapped around lunch sandwiches. "We have so much of it," Wissner-Gross said.

The National Park Service's history of the monument confirms the aluminum capstone but corrects the reason. Aluminum is abundant in Earth's crust; it was expensive because it was hard to extract as a metal. When the aluminum apex was set on the monument on December 6, 1884, the metal cost about $1.10 an ounce, roughly the price of silver. William Frishmuth cast the piece, and it was shown at Tiffany's before installation. In 1886 Charles Martin Hall developed an electrolytic method of extracting aluminum, and Paul Héroult arrived at the same process independently soon afterward. In aluminum's case, the metal was plentiful all along; what changed was the cost of getting it out of the ground.

"Whatever you think is scarce, give it a second," Diamandis said. "Technology transforms scarcity into abundance over and over again everywhere."

Not time to sell yet

The panel also added a warning. Before anyone "freaks out and sells all their gold," the discussion noted, the cost of the process means it will first help make energy abundant through fusion, well before it really affects the gold supply. Blundin observed that gold doubling the profitability of a power plant would be a lot. What the panel treated as finished was the popular story that gold is the scarcest thing and so deserves a premium. That part, the discussion concluded, is "cooked."

Blundin closed the segment looking at the gold itself. He called it "epically incredible" that the only way gold has come to exist in the universe is the collision of two neutron stars, something he said has been true for 4 billion years. And now, he said, gold is being made.

Share this article

Go to the original

Sources & further reading

  1. 01
  2. 02
  3. 03

From the conversation

Podcast episodes

Moonshots with Peter Diamandis

SpaceX’s AI Business Up 830%, OpenAI Agents Lie, and Startup Turns Mercury Into Gold | EP #305

Episode published (recorded )This article draws on 0:51–1:07, 8:09–9:13, 9:24–10:09, 1:54:31–1:59:05, 1:59:22–2:00:39, 2:00:45–2:00:58 and 2:00:59–2:01:25 (approximate times)

Article history

Updates to this article