Google DeepMindAnnounceand US nuclear fusion startupsCommonwealth Fusion Systems (CFS) will use AI technology to optimize the operating efficiency of the nuclear fusion reactor "Sparc", which is expected to significantly accelerate the commercialization of clean energy.
Received investment from NVIDIA, Google, etc.
Commonwealth Fusion Systems, a nuclear fusion startup from Massachusetts, recently announced the completion of a new round of financing totaling US$8.63 million. Investors include NVIDIA and Breakthrough Energy Ventures, a fund backed by Bill Gates, the founder of Google and Microsoft. The company also attracted support from global heavyweights such as Morgan Stanley, Stanley Druckenmiller, and a consortium led by Japan's Mitsui & Co. and Mitsubishi Corporation.
Challenges and Breakthroughs in Nuclear Fusion Technology
Nuclear fusion is considered one of the most promising clean energy sources of the future, offering the advantage of not producing long-lasting radioactive waste. However, to achieve nuclear fusion power generation, ionized gas "plasma" must be heated to temperatures exceeding 1 million degrees Celsius and maintained in a stable state within the reactor.
The "prototype reactor" currently under development at CFS is a compact device using tokamak technology. It utilizes superconducting magnets to confine and compress plasma, thereby releasing enormous amounts of energy. The goal is to achieve the first-ever "net nuclear fusion energy gain" by 2027, a breakthrough achievement where the energy output exceeds the energy input.
The key role of AI technology
In this collaboration, DeepMind will provide CFS with a number of AI technologies, building on its previous experience using deep reinforcement learning to control tokamak plasmas:
• Open source plasma simulator TORAX:The ability to perform fast and accurate nuclear fusion plasma simulations allows CFS to perform millions of virtual experiments, testing and adjusting operating parameters before SPARC actually operates.
• AI agent system AlphaEvolve:By exploring countless potential operating scenarios in a simulated environment, quickly identify the most efficient path to achieve net energy gain.
• Instant control of AI navigation system:A dedicated "AI navigator" was developed to dynamically control the plasma when SPARC is operating at full power, effectively dispersing the huge amount of heat concentrated in a small area and protecting the solid materials closest to the plasma.
Future Outlook and Industry Impact
The ultimate vision of this collaboration is to establish AI as the foundation for intelligent, adaptive systems in future nuclear fusion power plants. DeepMind emphasizes that future "AI navigators" will be able to learn more complex adaptive strategies than those designed by engineers, balancing multiple constraints and objectives and rapidly adjusting optimal control algorithms.
Notably, Google has also invested in CFS to support the commercialization of its technology. This collaboration not only represents an innovative application of AI technology in the energy sector, but also has the potential to revolutionize global sustainable energy development. As the 2027 target date approaches, the results of this collaboration will serve as a crucial indicator of whether nuclear fusion technology can be commercialized as planned.
