Nuclear-powered merchant ships have been called "almost ready for commercial use" for more than half a century, only to be stopped each time by the twin barriers of ports and insurance. The memorandum of cooperation (MOC) signed between the U.S. Maritime Administration (MARAD) and Britain's Core Power lays out two numbers: 50 U.S.-flagged nuclear merchant ships, with construction to begin in 2028. The NS Savannah, launched in 1959, was not retired because of a reactor accident—it was undone by a crew size 1.3 times larger than conventional ships, by ports refusing to accept it, and by economics that depended on subsidies. In 67 years, only four nuclear cargo ships have ever been built. Of the three that entered commercial service, the Savannah and the Otto Hahn were retired after five and roughly ten years respectively, while only the Soviet Union's Sevmorput kept running for more than 30 years until it aged out. Whether this new plan differs from the past hinges on whether the regulatory and insurance frameworks that didn't exist back then can be built alongside the ships themselves—and part of the funding for that effort is coming from Japanese trading houses and shipbuilders.
Cutting short the research phase, aiming to break ground in 2028
The signing took place on August 24, 2026, at the U.S. Department of Transportation headquarters in Washington. Core Power, the U.S.'s partner, is a British maritime nuclear company founded in 2018. It has been jointly researching regulations and safety with classification society Lloyd's Register and shipping giant Maersk for container ships carrying Generation IV reactors. In June 2026, it also began a feasibility study for a floating power plant using BWXT's 195MWe-class mPower reactor.
The MOC itself only refers to a "future U.S.-flagged nuclear merchant fleet," without specifying the number of ships or vessel types. What has been reported as Core Power's own vision is a mix of LNG tankers and container ships totaling 50 vessels, with a construction start target of 2028.
Mikal Bøe, who leads Core Power, said: "We're not asking for another study. We're aiming to begin construction starting in 2028." The output of the past decade of work on maritime nuclear power has been feasibility studies and regulatory research reports. Bøe's remark amounts to a declaration that this phase is now over.
MARAD Administrator Stephen Carmel struck a different tone. While calling nuclear propulsion a serious commercial opportunity, he said it must be approached as a complete system—one that brings together safety, security, regulatory approval, and viability as an investment. At the same signing ceremony, one side spoke of a target year for breaking ground, while the other spoke of the maturity of the overall system.
Core Power's own materials also outline a division of labor: conventional hull construction would draw on the shipbuilding capacity of allied nations such as Japan and South Korea, while dedicated facilities in the U.S. would handle reactor installation, initial fuel loading, testing, refueling, and fuel removal. It's a design that separates where shipbuilding capacity is concentrated from where nuclear fuel is handled. However, both parties have explicitly stated that the MOC is neither a procurement contract nor a budget commitment. The figure of 50 ships remains, for now, a target rather than an order.
On cost, Maritime Executive has reported that each nuclear merchant ship would run about $700 million, versus roughly $250 million for a conventional ship—2.8 times the price. This per-unit figure is not an official estimate from MARAD or Core Power; it comes from a single trade publication, and no other outlet has yet corroborated it. Still, it conveys a sense of the scale involved. If a shipowner's capital investment on the same route nearly triples, recovering that through fuel savings alone becomes difficult.
Why three ships were grounded by ports, and only one sailed until old age
In 67 years, only these four nuclear cargo ships have been built and launched: the United States' NS Savannah, West Germany's Otto Hahn, Japan's Mutsu, and the Soviet Union's Sevmorput. Of these, three—the Savannah, the Otto Hahn, and the Sevmorput—reached commercial operation, while the Mutsu never advanced beyond the demonstration stage. The Savannah and Otto Hahn were retired after five and roughly ten years of operation respectively, citing economics and port entry restrictions. The Soviet (later Russian) Sevmorput was refused entry at several of its own domestic ports and at Vancouver immediately after entering service, but it went on to operate mainly on domestic routes along the Northern Sea Route. According to its operator, after more than 30 years in service, it is expected to be replaced by a conventionally powered vessel sometime after 2024 due to its age.
| Year | Event |
|---|---|
| 1959 | NS Savannah launched (U.S., construction cost $46.9 million) |
| 1962 | NS Savannah begins demonstration voyages |
| 1965 | NS Savannah receives commercial operating license |
| 1968 | Otto Hahn enters service (West Germany) |
| 1969 | Mutsu launched (Japan) |
| 1970 | NS Savannah ends commercial cargo transport |
| 1971 | NS Savannah defueled and laid up |
| 1974 | Mutsu suffers radiation leak during power-up testing |
| 1979 | Otto Hahn retired |
| 1988 | Sevmorput enters service (Soviet Union) |
| 2018 | Core Power founded (UK) |
| 2024 | Sevmorput's replacement announced due to age |
| 2026 | MARAD and Core Power sign MOC |
| 2028 | Target year for construction start |
Laying out this timeline reveals that the reasons for retirement were not concentrated on reactor failures. What stopped the Savannah, the Otto Hahn, and the Mutsu were conditions on land, and only the Sevmorput ran until it aged out. Note that the dates for each ship from 1959 through the 1990s are based on records from specialized media and reference sources.
The NS Savannah was launched in July 1959 and, after repeated demonstration voyages from 1962 to 1965, received a commercial operating license from the U.S. Atomic Energy Commission (AEC). Its construction cost was $46.9 million. The reactor performed as designed, and technical reliability remained high throughout its operational life. What undermined its profitability was the operating conditions: it required a crew 1.3 times larger than a conventional ship, and personnel costs remained persistently elevated.
Then ports became the wall. U.S. port authorities and labor unions were reluctant to accept the ship, while Japan, Australia, and New Zealand refused it entry. The route could not be made viable without subsidies, and cargo transport ended in 1970, followed by defueling and lay-up in 1971. Ironically, a cumulative 1.5 million visitors toured the ship at its various ports of call. As an exhibit, it drew crowds; as a cargo ship, it never reached profitability.
The economics of a nuclear ship come down to whether fuel savings outweigh the added costs of crew, insurance, and port procedures. In the Savannah's case, that gap never closed. Moreover, if even one port refuses entry, the entire design of that route collapses. The economic viability of a nuclear ship is determined less by a reactor's thermal efficiency than by how many ports are willing to let it in.
Japan's nuclear ship Mutsu was launched in June 1969. On September 1, 1974, during power-up testing, radiation leaked from the vessel. The cause was a defect in the shielding structure, not a malfunction of the reactor itself. Even so, public backlash did not subside, and disputes over accepting the ship at its home port continued. The Mutsu was decommissioned—its reactor removed—without ever entering commercial service, and it was later converted into the ocean research vessel Mirai.
Only the Sevmorput followed a different path. Immediately after entering service, it too was refused entry at several of its own domestic ports and at Vancouver, hitting the same wall as the other three ships at least once. Even so, because it had an alternative route in the form of domestic transport along the Northern Sea Route, it continued operating, and in the 1990s it also served the Vietnam route. It kept operating even after a major refit in 2016, and multiple voyages have been confirmed between 2022 and 2023. From 2024 onward, it is expected to be replaced by a conventionally powered vessel due to its age, though the exact retirement date has not been confirmed.
The question of crewing also returns, in a different form, in today's plans. Certification requirements for crew members handling reactors are listed by MARAD and Core Power as an item still to be worked out. As the Savannah's track record shows, the corresponding personnel and training costs get added onto a conventional ship's baseline. The reported $700 million construction cost per vessel is purely an initial capital figure; the gap in operating costs accumulates separately, on top of that.
What all four ships have in common is not that their reactors broke down. It's that the institutions meant to receive the ships were never put in place before the ships themselves. Decisions by port authorities, acceptance by labor unions, and the allocation of liability in the event of an accident—all of these were negotiated only after the ships were already completed. The Sevmorput hit the same wall immediately after entering service, but because it had an alternative domestic route along the Northern Sea Route, it had the room to keep running until it aged out.
188 ships in the fleet, 50 more, and a nominal $35 billion
According to MARAD's official dashboard, the U.S.-flagged oceangoing merchant fleet (self-propelled vessels of 1,000 gross tons or more) numbered 188 ships as of March 2026. Applying the target of 50 new ships to this base—50 divided by 188—comes to about 27%. That is a plan to newly build, using nuclear power, a fleet exceeding a quarter of the current one. However, the 188 figure represents ships currently in operation, while the 50 figure represents ships yet to be built; these are not figures from the same point in time, and the comparison should be read only as a sense of scale.
Looking at the dollar figures as well: simply multiplying the reported $700 million per ship by 50 yields a nominal $35 billion. Converted at the Bank of Japan's published exchange rate on August 24, 2026 (roughly ¥159 to the dollar), that comes to about ¥5.565 trillion (all yen conversions below use the same rate). This is a rough figure that includes no discounts, economies of scale, or inflation adjustments, and it does not include the cost of building land-based infrastructure such as dedicated shipbuilding facilities or fuel-cycle plants. Given that LNG tankers and container ships have different specifications, the assumption that all 50 ships would carry the same per-unit price is also a simplification.
Even so, this rough estimate indicates the order of magnitude of funding required. No source has yet been identified for financing on the scale of $35 billion. Until that specific plan is disclosed, the 2028 construction-start target remains just a number on a schedule.
The same reports also touch on timing. Maritime Executive puts the delivery target for the first vessel at 2033, implying a process spanning five years from construction start to delivery. How quickly the remaining 49 ships can be lined up will depend both on securing shipyard capacity and on establishing mass-production capability for reactors.
Where that $35 billion flows is a matter of direct interest to Japan's shipbuilding industry. If the division of labor under which hulls are built at Japanese and South Korean shipyards materializes, part of this sum would land at East Asian shipyards. The breakdown has not been disclosed, and how much would be allocated to hulls versus reactors has not been decided either.
China's 200MW thorium reactor and the premise of a 95% share
The reason the U.S. is relying on Japan and South Korea for hulls lies in the distribution of shipbuilding capacity. According to UNCTAD's Review of Maritime Transport 2025, of global ship construction output in 2024 (completed deliveries by gross tonnage), China accounted for 54.57%, South Korea for 28.02%, and Japan for 12.56%—a combined 95.15% for the three countries. China's presence looms large in new orders as well: reports indicate that in 2025, Chinese shipyards captured more than 60% of new orders by compensated gross tonnage. U.S. shipyards represent only a tiny share of this landscape, and lack the capacity to supply 50 ships domestically on their own.
China, too, is reaching toward nuclear-powered merchant ships. In December 2023, Jiangnan Shipyard, a unit of China State Shipbuilding Corporation (CSSC), unveiled the design for a 24,000 TEU-class container ship called "KUN-24AP," powered by a thorium-based molten salt reactor, at Marintec Shanghai, and obtained Approval in Principle (AiP) from classification society DNV. Reports from 2025 cite technical details of this concept—a 200MW thermal (200MWth) thorium molten salt reactor and a 14,000 TEU-class vessel—with the stated specifications varying somewhat depending on when the announcement was made.
DNV's approval in principle is a milestone indicating that a design is likely to fit within the regulatory framework; it does not signify a construction contract or completed detailed design. From a rough sense of principal dimensions, machinery layout, and weight distribution, a project moves through detailed design before steel is ordered and construction finally begins. China's plan, too, currently remains at the drawing-board stage.
South Korea is moving in the same direction. HD Hyundai has begun jointly developing a 16,000 TEU-class nuclear-powered container ship with the American classification society ABS. The very shipbuilding nations to which the U.S. plans to entrust its hulls are also striving to take the lead on the reactors that will go into them. For shipyards in Japan and South Korea, nuclear merchant ships represent not only potential orders but also targets of their own technological development.
What separates the two sides is not how far the drawings have progressed. When the moment comes to move from design to steel, it is the 95% bloc that already possesses the shipyards, skilled workers, and supply networks. The fact that America's plan is designed to outsource hulls to East Asia is a choice made in recognition of this asymmetry.
Japanese shipbuilders and trading houses on Core Power's shareholder list
Japanese companies are deeply embedded in Core Power's capital structure. Investors including Mitsui, Mitsubishi, and Sumitomo are reported to have collectively invested about $200 million (roughly ¥31.8 billion). This includes a capital increase round of about $80 million (roughly ¥12.7 billion) in May 2023, underwritten by 13 Japanese companies including Imabari Shipbuilding and Onomichi Dockyard, which reportedly made Japanese companies the majority shareholders of Core Power. This capital raise took place more than three years before the current MOC and was not linked to any agreement with the U.S. government.
The shareholder structure and the proposed division of labor point in the same direction. The arrangement in which the U.S. takes on reactor installation and fuel loading/removal while East Asian shipyards handle the hulls essentially mirrors the flow of capital in the work itself. For Japanese shipbuilders, this also represents a pathway by which the plans of a company they hold shares in translate directly into work at their own shipyards.
Here lies a historical turn of events. The very shipbuilding companies from a country that could not accept the Mutsu in 1974 now count themselves among the major shareholders in America's nuclear merchant ship program. However, what stalled with the Mutsu was not shipbuilding capability—it was the ports and public opinion on the receiving end. Even if Japanese shipyards were to build nuclear merchant ships bound for the U.S., whether those ships could enter Japanese ports would be an entirely separate negotiation.
The timing of the investment is also worth noting. The capital increase by 13 Japanese companies, including Imabari Shipbuilding and Onomichi Dockyard, took place in May 2023—more than three years before U.S. cooperation took concrete shape. Japan had already taken its seat as a shareholder before the U.S. committed in earnest to nuclear-powered merchant ships.
The source of this capital also bears on the plan's staying power. Before federal funding moves, the cost of advancing the plan falls on the private sector. The presence of trading houses and shipbuilders among Core Power's shareholders means that the builders and the buyers sit within the same circle of capital. Given that the Savannah could never turn a profit without subsidies, having private capital take the lead this time counts as a genuine difference from the past.
What will move the 2045 forecast is insurance and ports, not reactors
Classification society DNV projects that nuclear propulsion will reach commercial scale around 2045. Core Power's stated construction-start target is 2028—a gap of 17 years between the two. This gap stems less from differing estimates of reactor development timelines than from the sheer volume of procedures that must be settled outside the reactor itself before a ship can actually sail.
For a nuclear ship to enter service on a given route, flag-state approval alone is not enough. Authorities in each port of call must agree to accept the ship, and liability in the event of an accident must be settled through treaties and insurance contracts—otherwise, a shipowner cannot commit to operating it. In the Savannah's era, this framework did not exist, and each port call required individual negotiation. What differs in the current plan is the design philosophy of running this institutional groundwork in parallel with reactor development itself. That is what Carmel meant by calling it a "complete system."
Opposing arguments have also become more concrete. In 2024, the Bulletin of the Atomic Scientists cited, as reasons for skepticism toward nuclear merchant ships, the risk of becoming a target for piracy or terrorism, the immaturity of insurance systems, and the need for rigorous regulatory review. Two of these—insurance and regulatory review—overlap directly with the items that the Core Power camp says it will address in parallel with technical development. The remaining issue—how to protect nuclear fuel aboard a ship at sea—belongs to the security dimension Carmel raised, and the answer to it rests not with shipyards or regulators, but with each nation's maritime security apparatus.
There are three benchmarks to watch. Whether construction begins on U.S. facilities to handle reactor integration and refueling. Whether major port-of-call nations put acceptance procedures in place. And whether liability insurance that shipowners can actually secure underwriters for becomes a viable product. All three should show signs of movement before 2028; if they don't, the construction-start date will slip further back.
When Bøe named the year 2028, he wasn't talking about building a fourth experimental ship. Once these three conditions are met, nuclear merchant ships will move beyond being prestige projects for nation-states and become an option shipowners choose on the basis of profitability. What has remained unsolved for 67 years is not the reactor. It is these three things.