A story about nuclear-powered container ships popped up in a fresh technology-news feed: in March this year, the American Bureau of Shipping and South Korea’s HD Hyundai signed an agreement to design a nuclear-electric propulsion plant for a 16,000-container vessel. And in the retelling was the line that hooked me: the barrier is not the technology, but regulations written in the sixties for first-generation reactors—which have closed most of the world’s ports to such ships.
The wording felt familiar enough to make me yawn. Everybody “knows” the nuclear merchant fleet never happened: too expensive, too dangerous, a reactor on a civilian ship is a utopia. But once I opened the files on the subject, it turned out I knew precisely nothing: how many ships there actually were, exactly where they died, and exactly who turned them away. I checked the previous installments—the subject had never come up once. So I’m digging from scratch.
Humanity built exactly four civilian vessels with nuclear propulsion: the American Savannah, the German Otto Hahn, the Japanese Mutsu, and the Soviet lighter carrier Sevmorput. Plus the Soviet icebreaker Lenin, which was the world’s first civilian nuclear-powered vessel altogether—but an icebreaker is not a merchant ship.
The four have one thing in common: they all died ashore. None was lost at sea, none lost its hull, none was retired because its reactor could no longer turn the propeller. The first went to a museum, the second was converted to diesel and then scrapped, the third was remade into a research vessel, and the fourth survived two attempts to scrap it and still sails.
And there is one more shared detail, which I began to suspect while reading and which the documents then confirmed. From the start, none of the four projects was built as a commercial venture. They were built as demonstrations, and the sources say so outright.
Formally, the story goes like this. In 1955, President Eisenhower launched the Atoms for Peace program, and its third item was a nuclear-powered cargo ship. Savannah was laid down in 1956, launched in July 1959, and its reactor started up in April 1962. The vessel cost $46.9 million, of which $18.6 million was the hull itself and $28.3 million the reactor plant and fuel.
Technically, it was a flawless machine. Cruising speed of 21 knots, a range of 300,000 nautical miles on a single load of 32 fuel assemblies—some 14 trips around the Earth. Passenger cabins, a swimming pool, a movie theater, a 100-seat dining room, a bar with a sculptural interpretation of the periodic table, and an Atomic Age designer lounge.
Now for the economics. Cargo capacity—7,700 tons in 652,000 cubic feet. Competitors carried several times more. The streamlined hull made loading the forward holds manual work, and that became a drawback as ports automated. The crew was a third larger than that of a comparable oil-fired ship, and had to be trained separately—a nuclear-engineering program was even opened at the U.S. Merchant Marine Academy for the purpose.
The result: Savannah required roughly $2 million more in annual operating subsidies than an ordinary dry-cargo ship of the same size with an oil-fired boiler. In 1971–1972, the Maritime Administration took her out of service to save money.
Here the story takes a turn that explains everything that followed. At an oil price of $20 a ton, the decision to economize looked sensible. By early 1974, oil was already around $80 a ton, and with 330 operating days a year, the fuel-price increase alone would have erased the cost gap entirely—even with all of Savannah’s design drawbacks factored in. The ship was taken out of the game literally on the threshold of the moment when its economics first penciled out.
As for ports, it is telling that she managed to call at 45 foreign and 32 American ports, and 1.4 million people visited her—as a traveling exhibition. As a merchant ship, she did almost nothing: over her entire passenger career, she carried 848 passengers and 4,800 tons of cargo. Later, from 1965 to 1971, she was leased out to a cargo line—350,000 miles of voyages—but the late-decade turn toward commerce could no longer change anything.
One more detail that does not line up across the sources. American materials give the vessel’s cost as $46.9 million, broken down as $18.6 million for the hull and $28.3 million for the reactor with fuel. The BBC wrote “about $50 million” in 2014. A difference of $3 million against the full sum is not critical, but the available texts do not show where it came from—rounding, a different exchange rate, or crew-training costs included. So I leave it that way, flagged.
Mutsu was Japan’s first, laid down in 1968 and completed in August 1972, at a cost of around 7.3 billion yen. Another roughly 3 billion yen went into infrastructure at her original home port, Ominato. It began with huge public enthusiasm: the crown prince and his wife came for the launch, and a commemorative stamp was issued.
On August 26, 1974, the vessel put to sea for trials, leaving the bay through a ring of fishing boats by using a gap in the blockade left by Typhoon No. 14—the departure had been planned for the day before. On August 28, she reached first criticality. And on September 1, a flux of fast neutrons was detected on the upper ring of the biological shield—not a radioactive leak from the fuel, but at that moment nobody cared about the distinction, including the newspapers, which wrote specifically of a “radioactive leak.”
Then came the part that makes this story worth remembering. On September 5, the fishing unions of Mutsu Bay decided not to let the ship back in. Mutsu remained in the ocean—physically intact, reactor shut down, but forbidden to enter her own port. The drift ended only on October 14, when four parties—the ruling party apparatus, the governor of Aomori Prefecture, the mayor of Mutsu, and the head of the fishing federation—signed an agreement: move the base out of Ominato within two and a half years, determine a new location within six months, and pay the locals 1.2 billion yen in compensation. The next day, the vessel entered port and was laid up with her reactor sealed.
According to a report by Japan’s Board of Audit, by 1982 the state had poured 48.46 billion yen into the project, while repairs in Sasebo alone—6.08 billion for shielding work, plus 2.68 billion for a full safety inspection, plus 3.7 billion for berthing and management—cost 12.47 billion yen. Nearly twice the price of the vessel itself. The Board put it dryly: the development result remained unverified, and the costs would continue to rise.
One detail makes this stop being a story about a reactor. Sasebo agreed to take Mutsu for repairs not for technical reasons, but because the local shipyard was on the verge of bankruptcy, and the contract was a way to save it. The terms of the deal were formalized in a document dated May 26, 1978, which went down in history as the “Mutsu written commitment”: in exchange for accepting the vessel, it promised priority construction of the West Kyushu Shinkansen route. Three ruling-party officials signed it; the original is kept by the Nagasaki prefectural government.
In 1992, the local project chose a route bypassing Sasebo. The section that was supposed to pass through the city opened only on September 23, 2022—thirty years after the promise had lost force. By then, the vessel herself had long been rebuilt: in 1990, she first sailed under nuclear power, covering more than 82,000 kilometers in a year of trials—more than two Equator-lengths. In 1993, the reactor was cut out, and the hull rebuilt as the research vessel Mirai, which still operates today. The reactor compartment was put on display in a museum—the only case in the world where a spent reactor is open to the public, though you have to view it through leaded glass.
The German version was built from 1963 to 1968 at a shipyard in Kiel, costing 56 million marks, of which 16 million came from Euratom. In the autumn of 1964, the hull was christened in the presence of Otto Hahn himself, after whom the vessel was named.
Nine years on nuclear fuel produced 650,000 nautical miles and 33 ports in 22 countries. The important thing is not the number, but the note attached to it in the sources: the ship visited most of those ports once—on special permission. Neither the Suez Canal nor the Panama Canal was ever opened to her. The experiment was shut down in 1979, and German sources state the reason directly: because of the nuclear installation, the vessel was not allowed to enter a significant number of foreign ports.
Then the plot turns almost comic. In 1982, Otto Hahn was sold to a private shipping company, which spent 4 million marks converting her into a diesel container ship—the aft superstructure was torn down and the bridge moved to where the reactor compartment had been. The vessel went through eight names, sailed under the Liberian flag in her final years, and was sent for scrap in Bangladesh in 2009 for $2.45 million. The bow crest stands on a lawn in Geesthacht; the funnel is in the courtyard of the German Maritime Museum in Bremerhaven.
And the clinching argument from those who consider the entire enterprise a failure: the designs for the next German nuclear container ships, NCS-80 and NCS-240, developed in the seventies, failed to find a single shipping company willing to order them—even with state subsidies. At that point, nobody is arguing about ports anymore, because there is no ship.
To understand where the port deadlock came from, we have to look at the legal layer written in parallel with these ships.
In 1962, the Convention on the Liability of Operators of Nuclear Ships was adopted in Brussels. Its logic was sound: the operator bears absolute liability, must carry insurance, and is responsible for nuclear damage anywhere in the world, wherever the incident occurs. Article XXIV puts it simply: the convention enters into force three months after instruments of ratification have been deposited by at least one state issuing licenses for the operation of a nuclear ship, and one other state.
And there is the trap. Ratifications have accumulated over sixty years, and they can be counted: the Democratic Republic of the Congo, Madagascar, Lebanon, the Netherlands, Portugal, Suriname, Syria. Not one of them has ever issued a license to operate a civilian nuclear vessel. The condition of “one party that is a licensing state plus any other” has never been met: the convention never entered into force, and never will.
At the same time, the basic liability regimes—the Paris and Vienna Conventions—explicitly exclude nuclear ships from their scope. The result is a vacuum: an accident involving such a ship in a foreign port is governed by no general instrument. One legal review cites a specialist publication from 1989, which says it without softening it: by the mid-seventies, practically no port remained in the Pacific that would accept a nuclear ship.
Sevmorput—the only one of the four still in service—ran into the same deadlock. She was laid down in Kerch in 1984, launched in 1986, and commissioned on January 31, 1988. The largest civilian nuclear-powered vessel ever built: 61,000 tons displacement, 74 lighters or 1,336 containers. In the first years, she ran international lines from Odesa to Vietnam and Vladivostok, then carried cargoes to Dudinka.
After that, the familiar pattern: in October 2009, the operator’s CEO publicly reflected that the vessel would have to be cut up “for needles”; in 2012, she was removed from the register, her reactor shut down for good; at the end of 2013, an order was issued to restore her. There were no other ships of her class in the country, so she was returned to service—1.3 billion rubles, plus 57 million rubles for new equipment. Builder’s sea trials ended in November 2015; the first voyage came on May 6, 2016, to Kotelny Island.
The second attempt looks far more grown-up, and there is plenty here to unpack.
In August this year, the International Atomic Energy Agency launched the ATLAS initiative in Washington—on the use of small modular reactors aboard ships and in floating power systems. On the second day, delegates went to Baltimore to see that very Savannah: 64 years since her maiden voyage, four years since her reactor was removed to Utah.
At the same time, the rules are being revised. The IMO Committee instructed the Sub-Committee on Ship Design and Construction to revise Chapter VIII of the SOLAS Convention and the Code of Safety for Nuclear Merchant Ships—the very code adopted by resolution A.491(XII) in 1981 and built solely around naval-type pressurized-water reactors. A working document by the subcommittee, dated October 2025, proposes shifting the Code from a prescriptive form to a goal-based, technology-neutral one, and creating an intersessional working group on nuclear ships for 2027–2029.
Business, meanwhile, is already doing the math. Under an agreement between the U.S. Maritime Administration and Britain’s Core Power, the plan is for a fleet of 50 nuclear-powered ships—gas tankers and container vessels—at $700 million apiece, versus $250 million for a conventionally powered vessel of the same capacity; the first keel is planned from 2028, the first delivery in 2033, hulls built in Japan and Korea, nuclear assembly in the United States. In June, the Chinese Jiangnan shipyard announced construction of a nuclear container ship for 25,000 containers. ABS and HD Hyundai are developing a 16,000-container concept in which a small reactor delivers up to 100 MW.
And now for the most important finding—the reason it was worth reading the reports to the end. A joint study by Lloyd’s Register, the Port of Rotterdam, Core Power, and Maersk on whether a nuclear ship can call at a European port concludes: the main barriers are not technical. They are regulatory consistency, governance, risk-management integration, and public opinion. Existing regulation almost never describes nuclear vessels in any terms other than military ones; nuclear and maritime safety regimes exist in parallel and scarcely intersect; ports have no clear guidance on how to interpret the licensing, insurance, and liability of an arriving vessel. And as a separate point—the recommendation to avoid unlimited-liability regimes incompatible with commercial shipping.
It is precisely the same deadlock in which Savannah stood in port in 1964 and Mutsu drifted in the ocean in 1974. Only now it comes with a budget and a schedule.
The decision here will not be about reactors. Until a clear liability framework appears, no port will take on the risk of letting such a vessel berth at its pier, and without port calls a commercial fleet does not exist—the authors of the Rotterdam study say exactly that. Over sixty years, four generations of reactors, three conventions, and two fuel crises have come and gone. The question remains open—the question that in 1962 seemed like a technical appendix to a treaty: who pays if the reactor does leak on someone else’s shore.
Worth keeping in mind, by the way, when reading any declarations about a maritime “nuclear renaissance.” We now know how to build ships of this class far better than we did then. But the paperwork beneath them is exactly the same.