News Summaries

China’s Fusion Frenzy: Billion‑Dollar Bet to Power the Future

China is throwing unprecedented money into nuclear‑fusion research, outspending the United States by a factor of ten. In July 2025 the state‑backed China Fusion Energy Co. went public in Shanghai, raising a record‑breaking 11.5 billion yuan and creating a “super‑unicorn” backed by the China National Nuclear Corporation, China National Nuclear Power and PetroChina Kunlun Capital. At the same time, private innovators are racing ahead: Energy Singularity’s “Honghuan 70” became the world’s first all‑high‑temperature‑superconducting tokamak, slashing cooling costs and shrinking the reactor size, while Star Ring Fusion is exploring compact spherical designs. These parallel tracks signal that China’s fusion ecosystem is moving from laboratory experiments to real‑world engineering tests. A national fusion conference in Hefei will bring together industry, academia and financiers to accelerate commercialization. If successful, fusion could power everything from massive seawater‑desalination plants and year‑round vertical farms to spacecraft capable of reaching a few percent of light speed, turning energy into a limitless public good.

Breakthrough Plasma Accelerator Supercharges Electron Beams – More Power, More Precision

A research team from UCLA and SLAC has unveiled a new kind of particle accelerator that can crank up both the energy and the clarity of electron beams at the same time. Traditional accelerators usually have to choose between making particles faster or making the beam tighter and brighter, but this new device uses a technique called plasma wakefield acceleration to do both. By sending a short, intense laser pulse through a cloud of ionized gas, the scientists create a ripple—much like a boat’s wake—that propels electrons forward. The clever twist is that the plasma itself acts like a transformer, boosting the electrons’ speed while also squeezing the beam into a tighter, more focused stream. The result is a beam that’s not only higher in energy but also far brighter, which could dramatically improve applications ranging from medical imaging and cancer therapy to advanced materials research and future collider experiments. The breakthrough promises smaller, more efficient accelerator designs, opening the door to powerful new tools for science and industry without the massive size and cost of conventional machines.

China Unveils 72‑Qubit Quantum Processor, Shattering Reliability Barriers

Chinese quantum‑tech startup Zhongke Koyuan (Zhongke Koyuan Technology, Wuhan) has launched its new 72‑qubit neutral‑atom processor, dubbed Hanyuan 1. The breakthrough tackles the long‑standing reliability bottleneck that has limited practical quantum computing, delivering stable operation even when faced with external noise. Founded in June 2020 by veterans from the Chinese Academy of Sciences’ Institute of Precision Measurement Science, Zhongke Koyuan has been a pioneer in neutral‑atom quantum hardware. The company has already contributed to several national research programs, including the 973 and Key R&D initiatives, and now plans to pour even more R&D funds into scaling the technology. Hanyuan 1 is positioned for real‑world use cases such as combinatorial optimization, quantum simulation, finance, drug discovery, chemical modeling, and machine‑learning acceleration. The firm also hints at future integration with classical supercomputers to create hybrid quantum‑supercomputing platforms.

China’s Space Race: New Missions Target Dark Ages, Exoplanets, and Lunar Frontiers

China’s 15th Five‑Year Plan is turning its gaze farther, deeper and broader into space. Scientists will tackle the biggest mysteries – how the universe began, what drives space weather, and where life could arise – by launching a suite of ambitious space‑science satellites. The lineup includes the “Hongmeng” mission, the “Kuafu‑2” deep‑space probe, an exoplanet‑transit survey that watches distant worlds pass their stars, and an upgraded X‑ray timing and polarization observatory. Together they aim for breakthroughs on the cosmic “dark ages,” the Sun’s magnetic cycle, and the hunt for Earth‑like planets. At home, China’s flagship telescopes are already reshaping astronomy. The Guo Shoujing Telescope (LAMOST) has released more than 28 million spectra, making it the world’s largest stellar database, and is used by over 1,800 researchers worldwide. The Five‑hundred‑meter Aperture Spherical Telescope (FAST) has discovered over 1,170 pulsars, outpacing all other observatories combined and boosting research into cosmic navigation and gravitational waves.

Why Sketching Beats Talking: Shanghai AI Lab’s DiffThinker Redefines Multimodal AI

A team of researchers from Shanghai AI Lab, Nanjing University, CUHK and Shanghai Jiao‑Tong University has unveiled a fresh approach to AI reasoning that could change the way machines handle visual problems. Their new model, called DiffThinker, abandons the traditional “talk‑first” strategy used by large language models and multimodal models, and instead lets the AI draw its answer using diffusion‑based image generation. By turning reasoning into a visual construction task, DiffThinker gains a global view of the problem and can correct mistakes more easily than sequential text‑only methods. In head‑to‑head tests on vision‑centric benchmarks, DiffThinker outperformed the latest giants such as GPT‑5, Gemini‑3‑Flash, and Qwen‑3‑VL‑32B, delivering higher accuracy and more reliable results. The breakthrough is highlighted in the upcoming book *Multimodal Large Models: A New Generation Artificial Intelligence Technology Paradigm* by Liu Yang and Lin Liang, which explains the theory, technology, and real‑world uses of these next‑generation models.

China’s Enterprise AI Boom: 2025 Report Shows How Companies Are Turning Bots into Business Engines

The 2025 China Enterprise AI Application Industry Research Report, released by IResearch, marks a turning point for Chinese businesses. After years of a “hundred‑model war,” firms are moving past pure technology experiments and are now rolling out AI at scale. The report highlights that the newest wave of AI—driven by large‑language‑model‑powered agents—has already proven its worth in knowledge‑heavy tasks such as intelligent customer service, Q&A knowledge bases, and automated content creation. Key to this shift are four pillars: AI agents that can reason and act, massive data pipelines, powerful computing infrastructure, and coordinated organizational structures. Companies that successfully integrate these elements are seeing faster decision‑making, lower operating costs, and new revenue streams. However, the rollout is not without hurdles. Enterprises must overcome fragmented toolchains, data silos, and the need for robust governance. The report stresses that success depends on building an end‑to‑end capability that blends AI, engineering, and human oversight—what IResearch calls the “AI + Engineering + Human” ternary system.