Trump Administration Eyes Offshore Nuclear to Power AI Boom

THIRTY YEAR BUILD OUT

By Irina Slav - Jul 23, 2026

• The Trump administration is exploring offshore nuclear power as part of its plan to expand U.S. nuclear capacity.
• 
  
• Floating nuclear projects are gaining momentum globally, with Russia already operating one plant while U.S., South Korean and Chinese firms develop commercial offshore reactor technologies.
• 
  
• Despite growing interest, offshore nuclear faces regulatory, national security and environmental hurdles before large-scale deployment becomes a reality.
• 
  

The Trump administration is studying the deployment of offshore nuclear power generation, with the Marine Minerals Administration and the Nuclear Regulatory Commission agreeing this week to coordinate on any future developments. The announcement comes amid surging electricity demand driven by Big Tech.

“Submerged reactor systems have been safely deployed in naval applications for decades, demonstrating their potential as a reliable source of energy in demanding marine environments. While no commercial deployment on the Outer Continental Shelf is planned or approved at this time, it could greatly strengthen America's energy security in the future,” the acting director of the Marine Mineral Administration, Matt Giacona, said.

The new partnership is part of a broader push to boost the share of nuclear power in the U.S. energy mix, with Reuters recalling in a report that President Trump wants to increase nuclear capacity in the country fourfold by 2050. Indeed, nuclear, which previously fell out of favor because of the high upfront costs and a perception of high risk, has recently been regaining that favor because of its zero-emission principle of operation and the fact that it supplies baseload power.

That last fact has really tipped the scales in favor of nuclear as Big Tech emerges as the main driver of global electricity demand growth because of its focus on artificial intelligence development and application. The industry—just like every other, and households—needs a reliable power supply, and that means either baseload from hydrocarbons or nuclear, or wind and solar plus batteries, with battery technology still not advanced enough to secure the 24/7 uninterrupted supply necessary for data center operators.

Nuclear is making a return, but floating nuclear generation specifically is not really a popular technology globally, yet. There is only one offshore nuclear power station in operation, and that is Russia’s Akademik Lomonosov, which was built to power the Far Eastern Chukotka Peninsula and Yakutia regions. The station began commercial operation in 2020 and was recently reported to have generated its first billion kilowatt-hours.

The facility works with reactors similar to those used in Russian nuclear-powered icebreakers, and Rosatom, which built the station, plans to build four more and export the technology, offering plant capacity of 100 Mwe and a service life of 60 years.

China is also working on floating nuclear power generation. One company recently unveiled plans to build a nuclear-powered offshore logistics hub that would produce zero emissions, including through using nuclear fuel for powering ships.

It seems interest in offshore nuclear is growing, with two recent news reports about companies working on the technology. One of these is a U.S. startup by the name of Bluecore Energy, which is developing small modular reactors that can be deployed on floating barges and supply electricity to ports and other infrastructure, Baird Maritime reported this week.

In another report from this week, Samsung Heavy Industries was named as a signatory of a partnership deal with another U.S. company, Sargent & Lundy, for the development of small modular reactors for floating offshore power generators. The deal would see the Korean heavyweight develop a small modular reactor standard platform, following its development of a conceptual design for such a facility. In short, floating nuclear, which was mocked by some when the Akademik Lomonosov was launching, has now become an interesting avenue of energy exploration.

Yet not all in the U.S. federal government are fans of offshore nuclear. The Interior Department, for instance, has national security concerns related to the technology, with Reuters citing the agency as noting radar interference as one specific problem. Environmentalists, separately, believe offshore nuclear power generation is “a terrible idea” that would harm marine life.

By Irina Slav for Oilprice.com

CHEAPER, FASTER, EASIER 

China’s AI Offensive Gains Ground in Central Asia

By Eurasianet - Jul 22, 2026

• China is expanding its AI influence in Central Asia through the newly established World Artificial Intelligence Cooperation Organization and bilateral technology partnerships.
• 
  
• Kazakhstan emerged as China's key regional partner, signing more than $15 billion in agreements spanning AI, infrastructure, energy, finance, and electric vehicles.
• 
  
• Other Central Asian states are also deepening cooperation with Beijing through projects in AI, manufacturing, education, healthcare, and agriculture.
• 
  

Central Asian states are tilting towards Beijing amid intensifying competition between the United States and China for predominance in artificial intelligence development. Top officials from Kazakhstan, Kyrgyzstan, Tajikistan and Uzbekistan gathered recently in Shanghai to attend the World Artificial Intelligence Conference, where Chinese leader Xi Jinping presented an ambitious vision for Beijing’s leadership in the global AI race. The American and Chinese models for AI development differ significantly. Ahead of the conference on July 16, officials from 29 countries, including the four participating Central Asian states, signed an agreement to establish the World Artificial Intelligence Cooperation Organization (WAICO), headquartered in Shanghai and apparently designed to help Beijing promote its global AI ambitions. 

Kazakh President Kassym-Jomart Tokayev, who notably addressed the AI gathering right after Xi’s opening speech, praised China’s leadership on AI, saying that “more than half of the world’s AI researchers are from the PRC, Chinese innovators hold the largest number of AI-related patents, and the country’s dynamic innovation ecosystem is advancing at a breakthrough pace across many areas of cutting-edge technology.” To prove his point, Tokayev offered to host WAICO’s first annual meeting in Astana, as well as set up WAICO’s Central Asian office in the Kazakh capital. In addition, following Tokayev’s meetings with Xi and Chinese executives, more than 70 deals worth over $15 billion were signed, many of them dealing with AI, digital infrastructure and other high-tech initiatives. 

Tajikistan’s Ministry of Industry and New Technologies also announced bilateral agreements with the PRC in Shanghai, while Uzbekistan’s digital development minister met with major Chinese entities to discuss joint AI projects. Those talks did not yield any specific deals, however. Meanwhile, Turkmenistan did not send a delegation to Shanghai, but the country’s Ministry of Communications signed a MoU with China’s National Development and Reform Commission, a powerful super-ministry, covering AI cooperation.

Kazakhstan

The $15 billion portfolio of deals signed in Shanghai also included agreements on transport, manufacturing, energy and critical minerals. While few specifics have been disclosed, some of the major deals included:

• An investment agreement with China’s Guoyou Materials Group for the construction of a multifunctional terminal in Kazakhstan’s Caspian port of Kuryk, a key hub of the Middle Corridor.
• An MoU with China Communications Construction Company for the construction of a 600 MW hydroelectric facility in the Almaty region.
• A cooperation agreement signed by China’s Acre Coking & Refractory Engineering Consulting Corporation, Kazakhstan’s largest steel producer, Qarmet, and the Development Bank of Kazakhstan to construct two new coke battery complexes and a coke gas purification system in the Karaganda region. 
• A financial agreement between the Development Bank of Kazakhstan and China Central Depository & Clearing Co. to facilitate greater participation by Chinese investors in Kazakh financial instruments and strengthen links between the two countries’ financial systems.
• An MoU signed by Samruk-Kazyna, Freedom Holding Corp., the Astana city administration, and China’s Geely Auto Group to develop EV-charging infrastructure and introduce AI solutions to Kazakhstan’s automotive industry. 
• A licensing agreement between Kazakhstan’s Astana Group and the PRC’s auto giant Chery Holding Group to manufacture Chinese EVs Omoda and Jaecoo at the Astana Motors Manufacturing plant.
• An MoU between Kazakhstan National University (KazNU) and the Suzhou University to establish the former’s campus in China. 

Separately, Kazakhstan’s prime minister instructed the Ministry of Ecology and other agencies to finalize negotiations with China Tianying Inc within two weeks and submit for approval an investment agreement for the construction of new waste management facilities in at least three major Kazakh cities.

Kyrgyzstan

Kyrgyzstan’s state-owned agricultural company Kyrgyz Agroholding signed an MoU with the China National Electric Engineering Corporation to build a fertilizer plant in the Central Asian nation with a capacity of up to 600,000 tons per year. The same Kyrgyz company also made a deal with Chinese biotechnology company Shijiazhuang Shineon to build a veterinary vaccine plant in Kyrgyzstan that will annually produce 25-30 million doses that meet international GMP standards. 

Turkmenistan

Turkmenistan and China signed a package of agreements spanning energy, artificial intelligence, science and technology, education, healthcare, culture, transport, and logistics. Key deals included the establishment of a joint Turkmen-Chinese Energy Education Center, a new framework for AI cooperation, a 2026-2028 scientific and innovation partnership program, and a five-year bilateral cooperation plan covering 2026-2030.

Turkmen civil servants are taking part in China’s two-week specialized seminar, learning about operational mechanisms of the World Trade Organization and integration into the global economy, reports Orient. And Turkmen teachers and scholars underwent training in China, where they learned about modern teaching methodologies, the HSK testing system, and the use of digital technologies in education.

By Eurasianet

The AI Paradox Reshaping Clean Energy Research

By Haley Zaremba - Jul 23, 2026

• AI is straining global power grids and energy security even as it becomes indispensable for managing them.
• 
  
• Researchers in China used AI to identify a new molecule that can revive dead EV batteries for thousands of extra recharge cycles.
• 
  
• A University of Toronto team used AI to find six new metal alloys for jet engines and nuclear reactors in just a few weeks, a process that once took years.
• 
  

Artificial intelligence is a double-edged sword for the energy sector. The rapid buildout of data centers to support widespread integration of large language models in virtually every economic sector imaginable, from our energy grids to your electric toothbrush – yes, really – is pushing energy demand growth projections to unprecedented levels, threatening to far outpace energy capacity additions and imperil energy security on a global scale. On the other hand, artificial intelligence holds enormous promise for improving energy efficiency in a wide range of systems and may hold the key to unlocking next-gen clean energy methods and technologies that could be integral to enabling feasible decarbonization pathways.

In the clean energy sector, artificial intelligence is being used to improve forecasting models for more sophisticated and accurate predictions of energy supply and demand, leading to greater grid stability at a time when our electricity grids have never been more stressed. Researchers are also increasingly using large language models to conduct "needle in a haystack" type inquiries to find the best methods and materials for certain use cases, accomplishing in months what would take a whole team of scientists years to achieve through trial and error methodologies.

“Finding new materials, catalysts or processes that can produce stuff more efficiently is the sort of ‘’needle in a haystack’ problem that AI is ideally suited to,” the Financial Times reported last year in an article musing about “How AI might save more energy than it soaks up.”

In China, for example, scientists used large language models to identify a new molecule, lithium trifluoromethanesulfinate (LiSO2CF3), capable of replenishing lithium ions in dead electric vehicle batteries for thousands of recharge cycles, significantly extending their lifespan. “We had no idea what kinds of molecules could do that job or what their chemical structures would be, so we used machine learning to help us,” Chihao Zhao, part of the research team at Fudan University, was quoted by Scientific American last year.

In nuclear fusion research, artificial intelligence is coming in handy for rapidly modelling how different materials will stand up to the extreme temperatures involved in creating and maintaining plasma, an integral component of the fusion process. The scientists at the Ames National Laboratory in Ames, Iowa who have created the modelling tool, called DuctGPT, say that the use of AI has slashed a monthslong process down to just hours of computation. “Now when you ask it, ‘I want to design a material for fusion that has all x, y, z properties that are critical for use in fusion reactors. Tell me the combination of elements which satisfy the criteria,’ it will give you those combinations of elements with properties,” Ames Lab Scientist Prashant Singh told Interesting Engineering.

Just this month, researchers at University of Toronto Engineering have announced that they used large language models to successfully develop six new metal alloys that could transform the functionality and durability of  extreme environments including jet engines and nuclear power plants. The AI-powered system identified all six alloys in just a few weeks, an incredible feat compared to traditional scientific methods.

“There’s enormous demand for materials that can stand up to huge swings of temperature and pressure, such as what you would find inside a jet engine or in the steam generators inside nuclear power plants, anywhere conventional steel just can’t survive,” Yu Zou, who led the project, told Interesting Engineering.

These breakthroughs, and the speed that they are being achieved in, could be transformative for the deployment of next-gen clean energy. By slashing research timelines and finding materials that improve the performance and durability of clean energy infrastructure, experimental technologies that would have otherwise been prohibitively expensive to design, test, and develop can now be made scalable and commercially viable. And the timing could not be better, as we draw closer to major global decarbonization deadlines

By Haley Zaremba for Oilprice.com

Energy IPOs Are Booming Thanks to AI’s Insatiable Thirst for Power

By Haley Zaremba - Jul 21, 2026

• Energy companies have raised $12.6 billion in IPOs so far in 2026, already nearly triple 2025's full-year total of $4.3 billion.
• 
  
• Investors are chasing next-gen bets like nuclear fusion and space-based solar as AI's power needs balloon, with Meta signing a 1-gigawatt space solar deal with Overview Energy.
• 
  
• Solar PV led every new source of U.S. power capacity for 28 straight months through the end of 2025, making up 72.6 percent of all electricity additions, per FERC.
• 
  

The artificial intelligence boom has created an era of unprecedented uncertainty in the global economy. Large language models are evolving as quickly as they are being integrated, creating a major headache for anyone trying to project their future impact or even their current energy footprint. But while we don’t know exactly how much energy will be needed to power the tech sector in coming years, we do know that it will be a whole lot, and the market is already reacting accordingly.

“Energy companies are raising money at IPO at their fastest pace this century, taking advantage of investors’ hunt for new ways to bet on the boom in power-intensive AI data centres,” states a recent report from the Financial Times. In the first half of this year, the money raised in initial public offerings for energy startups was the highest since 1999, when the first dot-com boom spurred a similar gold rush. And the rate of growth is staggering: in 2025, energy companies raised a total of $4.3 billion for the entire year. The total for 2026 is already at $12.6 billion, and we still have another half year to go.

“Investors started by buying AI-linked names like Nvidia. Then they said, ‘hold on, every chip needs energy to power it’,” RBC clean energy analyst Chris Dendrinos was quoted by the Financial Times. “That’s put a huge tailwind behind these companies.”

This spending spree includes a wide range of energy companies, including unproven and next-gen energy technologies that are enjoying a windfall of funding that they may not otherwise have achieved. The tech sector is investing heavily in pie-in-the-sky energy research like nuclear fusion, enhanced geothermal energy, and space-based solar power. Earlier this year Meta, the company behind Facebook and Instagram, signed a deal with startup Overview Energy to develop as much as 1 gigawatt of space-based solar power.

The market is particularly bullish about nuclear fusion, which has finally broken through on Wall Street after years of struggling to go private. Historically, nuclear fusion research is so expensive and seen as so experimental that it’s been funded by governments and huge public projects. But now fusion startups are popping up in astonishing numbers and seeing successful IPOs.

While it may seem risky to put too much attention into unproven technologies, an all-of-the-above approach will be absolutely essential to keep up with the energy monster that AI is creating. “There’s no way to get there without a breakthrough,” Sam Altman, co-founder and CEO of ChatGPT firm OpenAI, said at the 2024 World Economic Forum in Davos, Switzerland. “It motivates us to go invest more in fusion,” he went on to specify.

The new energy gold rush is also keeping clean energy companies highly competitive at a time when the policy environment has never been more antagonistic. Solar photovoltaics was the single largest source of new power generation capacity in the United States for 28 consecutive months through the end of 2025, ultimately accounting for 72.6 percent of all electricity additions according to data from the Federal Energy Regulatory Commission (FERC).

Miguel Stilwell d’Andrade, chief executive officer of Portuguese electric utilities company EDP, recently told Semafor that we are currently “living in what arguably is one of the best periods to invest in renewables in the US over the last 20 years.” As a result, solar power is booming under the Trump administration, despite its best efforts to dampen domestic renewable growth.

The invisible hand of the market is simply too strong, and energy security is the imperative of the day. “For years, clean energy has been sold as a moral imperative. Now it is simply an economic and geopolitical necessity,” Forbes reported earlier this year. “It’s not about emissions. It’s about resilience and price stability.”

By Haley Zaremba for Oilprice.com

MIT Framework Maps Grid Weak Spots Before Climate Disasters Hit

By Haley Zaremba - Jul 22, 2026

• MIT's new framework pairs climate projections with county-level grid infrastructure data to pinpoint exactly where power shortfalls are most likely to hit.
• 
  
• Test runs on New England and Texas grids found blackout risk could jump fivefold by 2050 if climate projections aren't built into system planning.
• 
  
• Building that climate resilience into decarbonized grids costs almost nothing extra, researchers found, a sharp contrast to pricier fixes like seawalls.

Researchers at MIT have developed a new framework to map out geographical hotspots and pain points for our beleaguered electrical grids. As extreme weather events and soaring summer heat waves become more commonplace, and energy demands on the electric grid skyrocket thanks to data center hyperscalers, understanding the minutiae of climactic patterns as well as how they map onto regional trends in energy supply and demand will be critical to ensure grid stability and security going forward.

This summer’s blistering heat waves have underlined the great importance of preparing for increased occurrence and intensity of extreme climate events as the world grows warmer. While the worst is still to come, our grids are already overburdened and underprepared for near-term energy and climate projections, much less under the harsher conditions expected in coming decades. The United States Department of Energy admits that “the grid we have today does not have the attributes necessary to meet the demands of the 21st century and beyond” and says that it is “working with public and private partners to develop the concepts, tools, and technologies needed to measure, analyze, predict, protect, and control the grid of the future.”

Our grids need to expand at an unprecedented scale in order to keep up with future energy demands while also responding to the changing nature of our energy landscape as electricity from wind and solar becomes an increasingly central pillar of the world’s energy flows. And not only do our grids need to grow larger, they need to grow smarter. The International Energy Agency writes that “the acceleration of renewable energy deployment calls for modernising distribution grids and establishing new transmission corridors to connect renewable resources – such as solar PV projects in the desert and offshore wind turbines out at sea – that are far from demand centres like cities and industrial areas.”

This is where a mapping tool becomes absolutely critical. Building a better, more reliable energy grid isn’t just about how, it’s about where those grids are built. As MIT news sums up, “For energy systems that power a reliable grid, the future is all about location.”

The MIT framework, described in a paper published last week in the esteemed scientific journal Nature Energy, uses climate projections with county-level power infrastructure data to show that “long-term adequacy challenges in decarbonized grids arise from the interplay between meteorological conditions and system design, driven by prolonged renewable generation shortfalls tied to fine-scale infrastructure siting choices.” In other words, better data and better planning through the use of tools such as this framework will be critical to achieving decarbonization outcomes and adapting to climate change without sacrificing energy security.

To test out their framework, the scientists used data from New England and Texas grids. The simulation showed that if regional climate projections are not taken into account when designing energy systems, these regions could face a fivefold increase in energy shortfalls by 2050, with increased risk of blackouts. But when climate change factors are taken into account, resilience can be achieved with very low levels of additional expenditure.

“As we mitigate climate change with renewables, we can also adapt to climate change by using future weather projections in our power system planning, and the extra costs of that adaptation are, at least in this study, not much,” senior author Michael Howland, MIT’s Jeffrey Cheah Career Development Professor, was recently quoted in an MIT News report. “It’s different from other climate adaptation studies, where building a big seawall or other mitigation efforts are really expensive. In this case, if we’re smart when we design our power system decarbonization plans, it could cost almost nothing extra to simultaneously adapt to climate change.”

By Haley Zaremba for Oilprice.com

Commodities Beat Tech This Decade, and Wall Street Still Won't Buy In

By City A.M - Jul 23, 2026

• Commodity indices are up 200 percent since October 2020, beating the Nasdaq's 145 percent gain over the same stretch, yet energy and materials still make up less than 6 percent of the S&P 500.
• 
  
• The Magnificent Seven will spend nearly $800bn this year on the AI buildout, close to half of it on raw materials and energy, without funding the resource supply that buildout depends on.
• 
  
• Western oil majors, Jeff Currie's "Munificent Seven," hand back 14 to 15 cents of free cash flow per dollar of market value versus roughly 2 cents for Big Tech, yet trade below pre-Iran-war levels.
• 
  

The best performing asset class of this decade is also the most under owned. Since October 2020, when we first called for a decades-long super-cycle in commodities, the broad commodity indices like the S&P GSCI are up 200 per cent; gold 140 per cent. This year commodities are up 37 per cent with petroleum up 81 per cent. The bottlenecks rotated – gold, copper, silver, coffee, cocoa, oil and most recently diesel. But not the trend. In contrast, crypto indices are up 157 per cent, the Nasdaq 145 per cent and the S&P 500 117 per cent. By investor returns, hard assets have beaten technology.

You would never know it by what investors hold. Energy and basic materials represent less than six per cent of the S&P 500, less than a third of its long run weight. During the first part of this decade institutional investors used sustainability as a reason to dissolve real-asset sleeves, yet poured money into green energy that required vast amounts of raw materials like copper. Six years on, superior returns have produced little reallocation. Performance is supposed to attract capital, but it hasn't. That is the physical capital paradox.

What makes this even more remarkable is what the capital chased: the artificial intelligence buildout, the largest commodity short in history. The Magnificent Seven will spend nearly $800bn this year, of which close to half is for raw materials and energy: copper for power transmission, critical minerals for hardware, fuel and electricity for datacentres. The energy footprint of the five biggest buyers of AI compute is nearly 4m barrels of oil equivalent a day, more than most major industrialised nations. Investors have funded one of history's largest resource demand shocks while refusing to fund the resources themselves.

And that demand shock is not peaking, it's compounding. The conventional view is that the age of mechanisation came first and the digital age followed. We would argue the sequencing goes the other way. What the 20th century mechanised was repetitive and structured, because the binding constraint was not horsepower, it was cognition. Think the assembly line, the tractor, the loom. A century into the machine age, 80 per cent of physical work is still done by human hands, and artificial intelligence removes that constraint. Now the real age of mechanisation can be realised, and every robot is copper, rare earths, batteries and joules.

Magnificent Seven vs Munificent Seven

Compounding the paradox, investors could hardly be better paid to correct the mistake. In contrast to the Magnificent Seven stands the Munificent Seven: ExxonMobil, Chevron, ConocoPhillips, Shell, TotalEnergies, BP and Equinor, the western majors supplying the AI buildout's energy. The munificence is literal. These companies hand back 14 to 15 cents of free cash flow per dollar of market value, against roughly 2 cents for their magnificent counterparts. One group is priced for a future it may not fully capture while the other is paying handsomely in the present and priced as if the present were about to end. Today diesel and gasoline margins are at record levels as the war in Ukraine ravages stretched supply infrastructure, yet the Munificent Seven trade lower than before the US-Iran war. Rarely has the market offered such generous terms to hold what it needs most, and rarely been so completely declined.

The refusal has its reasons, none of them about price. A generation of allocators carries the scars of the 2010s, when the capital destruction from funding energy and metals projects was nothing short of epic. Today, however, passive vehicles allocate by size, not by price, mechanically buying whatever is largest and trending. The paradox persists: the marginal buyer is no longer looking at the price signal.

This tells you how it ends. Another year of best performing returns is unlikely to shift allocations. What will reprice under owned physical assets is a crisis, not the returns. It is the moment the physical world fails to deliver, which is knocking on the door. In the 1970s energy allocations surged because fuel ran short, not because energy returns rose. In the 2000s investors ploughed into metals because China emptied warehouses. Investors followed scarcity you could see: the queue, the empty shelf or the outage.

The system is now poised to create that crisis. Record margins are not being met with new investment. Investors are funding record demand growth while starving the supply side, with more mechanisation to come. At the same time the insurance policies are being exhausted: spare capacity, inventories and strategic reserves. The next disruption, which is already underway, will be experienced, not talked away.

The paradox will likely end abruptly, through a physical crisis that makes it impossible to ignore. The best returns in the market are transparent, but the market is unlikely to claim them until it is forced to. When that day comes the capital will arrive in abundance and at a higher cost. It was visible the entire time, it just wasn't owned.

By City AM