The energy sector’s decarbonization is critical to addressing the challenge of climate change. The need for carbon-free energy sources is further intensifying due to the rise of new technologies like artificial intelligence (AI), which consumes a lot of energy.
Besides clean electricity, there’s also growing energy demand for carbon-free heat generation, water desalination, and direct carbon capture from the atmosphere.
While renewable energy sources like solar, wind, and geothermal are gaining traction, there’s an even more powerful option that offers high reliability, high capacity, and consistency while having a relatively low carbon footprint.
That powerful carbon-free energy source is fusion power plants, which aim to supply a significant fraction of the world’s future clean energy demands.
Fusion Power is the Way to Accelerate Decarbonization
Fusion is among the most environmentally friendly energy sources because there are no harmful atmospheric emissions. This means fusion doesn’t contribute to greenhouse gas emissions (GHG) and, by extension, global warming.
At the same time, estimates suggest that the price of fusion energy will be competitive even relatively early in its technological development. Studies extrapolate capacity costs for future fusion power plants in the range of 1–10$/W, resulting in cost estimates for fusion energy to range between 20–100 $/MWh.
The fusion process joins two light elements to form a heavier nucleus while releasing massive energy. The reactions take place in a state of matter called plasma, which is a hot, charged gas made of positive ions and free-moving electrons with unique properties.
This is unlike fission, in which a heavy element is split into fragments. For fuel, fusion utilizes sources like hydrogen and lithium, which are abundant and accessible.
Now, a fusion reactor is a device that uses nuclear fusion to generate electricity. The ITER Tokamak is the world’s largest and most powerful fusion reactor.
ITER is an international nuclear fusion research and engineering project among China, the European Union, India, Japan, Korea, Russia, and the United States that aims to create energy through a fusion process similar to that of our Sun and stars.
Replicating this nuclear fusion at an industrial scale on Earth can provide virtually limitless clean, safe, and affordable energy. Hence, constant efforts have been put into recreating it and harnessing the energy from it.
Click here to learn what makes nuclear fusion the ultimate clean energy solution.
Introducing a Commercially Viable Fusion Power
It has been many decades since scientists first understood the theory of nuclear fusion. Since then, large-scale projects have been working to close the gap toward a fusion power plant using magnetic confinement approaches.
This includes the STEP spherical tokamak program in the UK, the SPARC tokamak in the US, and CFETR in China, as well as the tokamak ITER.
While tokamaks have been viewed as the most promising path to a fusion power plant, they face challenges due to their susceptibility to plasma disruptions. This poses a significant challenge for commercially viable operations. As a result, research is underway to solve this issue through prediction and mitigation methods. However, there have been no experimental demonstrations showing the feasibility of a power plant until now.
A research paper1 from German startup Proxima Fusion demonstrates the first quasi-isodynamic (QI) stellarator with low fast particle losses. The fusion energy reactor design is believed to offer the quickest route to commercially viable fusion power.
Named Stellaris, it is “designed to operate in continuous mode and be intrinsically stable,” said Proxima’s co-founder and CEO Francesco Sciortino. “No other fusion power plant design has yet been demonstrated to be capable of that.”
While stellarators are similar to the tokamak in terms of comparable energy confinement properties, their design avoids problematic current-driven plasma disruptions. These stellarators with no net toroidal current in the plasma have the ability to run in a steady state. This means they intrinsically reduce thermal and mechanical component fatigue and eliminate the need for an expensive plasma current drive system.
So, stellarators have several advantages over the tokamak, such as being more stable and needing less power to operate. The absence of a Greenwald-like density limit of stellarators, meanwhile, means they can operate at significantly higher densities, which is beneficial due to the favorable scaling of fusion power with fuel density.
Now, the QI stellarator is particularly promising for reactor applications because of its inherently reduced self-induced currents. Among the investigated stellarator options, the modular, low-shear, QI stellarator has the highest technological readiness level, which is explored in the world’s largest stellarator, the Wendelstein 7-X. It is a product of over €1.3 billion in funding from the German Federal Government and the European Union.
W7-X went into operation a decade ago but only reached its full design specifications just a couple of years ago. W7-X experiments have now achieved the reduction of neoclassical transport to a level at which turbulence is the major driver for heat transport in the plasma.
This means optimized stellarators can now be designed to accomplish neoclassical transport levels similar to those of tokamaks. As a result, optimized stellarators are fundamentally limited by the very same physical constraints as tokamaks while maintaining the crucial stability of stellarators.
However, while the Wendelstein 7-X is being developed for research and is located at the Max Planck Institute for Plasma Physics (IPP) in Germany, Stellaris could power the grid one day.
“IPP is a pioneer of stellarator optimization. In recent years, we have been able to design stellarators whose physics properties are predicted to grant unprecedented performance. This still leaves many technological and engineering challenges, problems that have been courageously addressed by Proxima Fusion in collaboration with IPP in this first-of-its-kind study,” said Prof. Dr Per Helander, head of the Stellarator Theory Division at the Max Planck IPP. “This is important and necessary work on the path toward a fusion power plant, which we hope to accelerate through this collaboration.”
The design will come to life with Proxima’s first demonstrator, which is expected to be finished in six years. The demo stellarator Alpha will be the first-ever fusion device to showcase net energy production in a steady state.
This will lay the foundation for the company’s first 1GW fusion reactor, which it hopes to power up sometime in the 2030s.
Leveraging AI to Address the Biggest Challenge
Spun out from the Max Planck Institute for Plasma Physics, Munich-based Proxima raised €20 million in funding last summer to build the first generation of fusion power plants based on QI stellarators with high-temperature superconductors and turn fusion into a viable business.
The preempted and oversubscribed seed round was led by Swiss VC firm Redalpine with participation from German government-backed DeepTech & Climate Fonds, the Bavarian government-backed Bayern Kapital, and the Max Planck Foundation.
Its existing investors, including the Tomorrow Fund, Wilbe, High-Tech Gründerfonds, UVC Partners, and Plural, also doubled down on their pre-seed investments.
Proxima Fusion’s focus, as we detailed, is QI stellarators, which it says hold promise for a safe, carbon-free, and effectively limitless source of energy. A stellarator is a ring of magnets in the shape of a doughnut. These magnets are precisely positioned and can contain plasma.
At the time, the company noted that stellar optimization results have disrupted the field of nuclear fusion, enabling it to tackle the challenges with an approach that focuses on engineering and simulation and leverages advanced computing.
Advances in computational power are what’s allowing Proxima Fusion to address the biggest challenge with stellarators — the complexity of designing and building them. It was the very complexity that made scientists choose tokamak in the 1960s.
However, AI supercomputers are now helping the Proxima team, which includes engineers from Google, SpaceX, Tesla, MIT, and McLaren Formula-1, build its fusion reactor design.
AI allows the company to rapidly iterate the best fusion reactor designs based on key parameters such as cost, efficiency, and material availability. This eliminates the need to build multiple prototypes, enabling Proxima to start building a functioning demonstrator right away. According to Sciortino:
“The understanding of complex geometry and its consequences is everything in stellarators. AI is helping Proxima to uncover patterns that lead to simpler, faster, and cheaper designs.”
Stellaris, here, is designed to generate more power per unit volume than any other stellarator. For this, high-temperature superconducting (HTS) magnets are used to create stronger magnetic fields. This not only allows for size reduction but also enables more efficient, cost-effective, and faster-to-build reactors.
The company also uses only existing materials, which means the reactors can be built with available supply chains.
“For the first time, we are showing that fusion power plants based on QI-HTS stellarators are possible. The Stellaris design covers an unparalleled breadth of physics and engineering analyses in one coherent design. To make fusion energy a reality, we now need to proceed to a full engineering design and continue developing enabling technologies.”
– Dr. Jorrit Lion, co-founder and chief scientist of Proxima Fusion
With all these results, Ian Hogarth, a partner at Plural, said Proxima has proved what it set out to do with Stellaris positioning:
“QI-HTS stellarators as the leading technology in the global race to commercial fusion.”
The company is planning its Stellarator Model Coil (SMC) demo magnet in 2027, which will test and validate the technology behind the coils used in the reactor, proving the feasibility of using HTS materials in such a complex magnetic coil system. This will fully de-risk HTS technology before building a full-scale stellarator device like “Alpha,” which is expected to achieve net fusion energy production.
“Given increasing global energy demands and the escalating need for European energy security, unlocking limitless, clean energy through fusion has never been more urgent, and Proxima is committed to leading Europe into a fusion-powered future.”
– Sciortino
Click here to learn how AI is helping improve nuclear fusion.
Big Nuclear Push Among Big Tech
Interestingly, not just small private companies but even big public organizations are making advances in the field. While they are investing in fusion power plants, their focus is primarily on small modular reactors (SMRs), which are not fusion power plants but a type of nuclear fission reactor. These SMRs are not only fast to put online as construction takes less time, but they are also more efficient than large-scale nuclear reactors driven by AI mania.
Being regarded as a major driver of economic growth in the future, both the private sector and the governments have invested billions of dollars in the research of fusion power plants as they remain in the exploration stage and are yet to go public.
Nuclear fusion has actually been under research for about 75 years, and despite that, it has not generated net energy in a meaningful way. According to Cathie Wood’s Ark Invest’s Big Ideas for 2025 report, while private fusion companies are promising breakthroughs in the next couple of years, “commercialization could take another ~15 years.”
1. Microsoft (MSFT -2.14%)
Last year in Sept., Microsoft secured a deal with Constellation Energy (CEG -7.29%) to reopen the Three Mile Island nuclear plant to help the tech giant meet its soaring energy demand. The facility was closed down in 2019 but will now open in 2028 and will be operational until at least 2054.
Constellation Energy Corporation (CEG -7.29%)
“The decision here is the most powerful symbol of the rebirth of nuclear power as a clean and reliable energy source,” said Joe Dominguez, CEO of Constellation Energy Corp. (CEG), a $75.7 bln market cap company whose shares are currently trading at $241.65, up 7.65% YTD.
The largest producer of carbon-free energy in the US, Constellation operates a fleet of nuclear plants. The restart of the Three Mile Island will provide over 800 MW of power, all of which will be purchased by Microsoft under the 20-year power supply deal.
In addition to this, Microsoft has a contract with Helion to buy electricity from its first fusion power plant once it starts generating power in 2028.
Microsoft Corporation (MSFT -2.14%)
With a market cap of $2.95 trillion, MSFT shares are currently trading at $398.60, down 5.81%. The company pays a dividend yield of 0.84%. For Q4 2024, the company reported revenue of $69.63 billion, a growth of 12.3% year-on-year, while earnings Per Share (EPS) stood at $3.23.
Microsoft leads in commercializing AI, with its artificial intelligence business reaching $13 billion in annualized revenue run-rate this year. This has been thanks to its close partnership with ChatGPT maker OpenAI. The company also released Copilot-branded AI assistants last year.
2. Google (GOOGL -1.92%)
Google has partnered with Kairos Power for as many as seven SMRs with a total capacity of 500 megawatts. This will help the seven-year-old startup bring its first commercial reactor online by the end of this quarter and additional reactors by 2035. Kairos has also secured $300 million in funding from the US Department of Energy for its Hermes project, a nuclear reactor demonstration plant.
Calling this agreement “landmark,” Michael Terrell, Google’s senior director of energy and climate, said:
“We feel nuclear can play an important role in helping us to meet our demand, and helping us to meet our demand cleanly and round the clock.”
Besides supporting clean growth, Terrel believes nuclear energy will help them “deliver on the progress of AI.” Moreover, if these projects scale globally, it can also help Google “deliver enormous benefits to communities and power grids around the world,” he added.
In addition to this, Google has also invested in TAE Technologies, a company that operated in stealth mode for years and has raised over a billion dollars to develop fusion power. Google has been in partnership with TAE since 2014.
Alphabet Inc. (GOOGL -1.92%)
With a market cap of $2.08 trillion, GOOGL shares are currently trading at $171.65, down 10.05%. The company pays a dividend yield of 0.47%. For 4Q24, its parent company, Alphabet, reported a revenue of $96.47 billion and an EPS of $2.15.
The company reported its revenue coming in at $11.96 billion, driven by the AI boom, though it still trails behind Microsoft and Amazon (AMZN -3.42%). It is projecting $75 billion in capital expenditures this year, driven by the buildout of data centers and infrastructure for AI.
Conclusion
Interest in nuclear energy, both fusion and fission, is surging as the world seeks a clean, reliable power source to meet growing energy demands, sustain AI initiatives, and achieve ambitious net-zero carbon goals.
When it comes to fusion energy, historically, tokamaks are used as magnetic confinement devices in fusion reactors. But while they keep plasma hot, stellarators are now getting attention due to their ability to keep plasma stable. With AI addressing the biggest challenge of stellarators—complexity—they could become the preferred option for a future fusion energy plant.
Here, Proxima Fusion is leading the charge toward a commercially viable solution by leveraging AI-driven design, high-temperature superconductors, and stellarator technology. With its first demonstrator on track for completion in six years and plans for a 1GW fusion reactor in the 2030s, Proxima is setting new standards in the race to unlock a limitless, carbon-free power source.
As global energy demands surge and the urgency to decarbonize intensifies, fusion is fast becoming an imminent reality!
Click here to learn about the first commercial nuclear fusion project announced.
Study Reference:
1. Lion, J., Anglès, J.-C., Bonauer, L., et al. (2025). Stellaris: A high-field quasi-isodynamic stellarator for a prototypical fusion power plant. Fusion Engineering and Design, In Press, Corrected Proof, 114868. https://doi.org/10.1016/j.fusengdes.2025.114868