In December 2025 China sent its first X‑band synthetic‑aperture‑radar (SAR) satellite, named Dianjian‑1, into a 500‑km sun‑synchronous orbit. Unlike ordinary optical satellites that are blinded by clouds, this tiny 300‑kg craft carries a flat‑panel radar that can see through weather and capture images with better than 0.5‑meter resolution. Using a technique called InSAR, it measures ground and structural movements down to a few millimetres, giving engineers a continuous, all‑weather “CT scan” of critical infrastructure such as dams, pipelines and power plants. The breakthrough isn’t just sharper pictures; it’s the satellite’s ability to revisit the same spot from almost exactly the same angle, within a 200‑metre orbital “pipe”. That precision lets scientists compare images over time and detect the tiniest deformations before they become safety hazards. The system was built from the ground up to meet China’s own engineering needs, avoiding the heavy, generic foreign SAR satellites that lack this high‑precision revisit capability. The first images returned by Dianjian‑1 have already demonstrated its power, confirming that China now has a dedicated “space guardian” watching over its massive energy projects, promising faster risk warnings and cheaper, more reliable monitoring for the nation’s infrastructure.
Read moreA recent experiment compared two AI‑driven research assistants—Co‑Scientist and Robin’s agents—to see how well they can help human scientists. Both systems are built as teams of specialized bots that tackle different steps of the discovery process, from brainstorming hypotheses to checking data. The study found that the agents that dug into existing scientific papers performed better than generic large‑language models on complex tasks. However, the AI still struggled with questions that required deep statistical or bio‑informatics knowledge and leaned heavily on prompts written by humans. Co‑Scientist’s “reflection” and “ranking” bots act like peer reviewers and debate teams, weighing the merits of competing ideas. Robin’s suite is tuned for drug‑repurposing, with one bot picking experiments and another crunching biomedical data. While these multi‑agent systems can speed up literature searches and suggest new experiments, they are not yet ready to replace expert judgment. The researchers stress that AI should be viewed as a collaborative partner that amplifies human insight, not a standalone scientist. The findings highlight both the exciting potential and the current boundaries of AI‑powered discovery tools.
Read moreAt a high‑tech showcase in Shanghai, Matrix Super Intelligence introduced its latest humanoid robot, the MATRIX‑3, and announced plans to roll out up to 100,000 units by 2027. Standing 1.7 m tall and weighing 65 kg, the robot looks and moves like a person, can walk at a natural pace of 3.9 km/h, and works for four hours on a single battery charge. The company says the robot combines four breakthrough technologies: a new AI brain called WAVE that learns from actions rather than just images, a novel linear‑joint design that gives the robot 33 degrees of freedom and strong, precise movement, a 27‑joint dexterous hand capable of micrometer‑level tasks, and a soft, fabric‑like skin that feels safe to touch. Founded in 2024 by former Tesla China design chief Zhang Haixing, the Shanghai‑based team claims they built the third‑generation model in just 25 months—far faster than the five‑to‑seven‑year cycles typical in the industry. During the event, MATRIX‑3 performed a live demo, showing smooth limb motions and interactive gestures. Matrix also opened its own “Super Intelligence Factory,” a fully integrated production line that lets the company control every step from design to mass‑assembly, promising consistent quality and rapid upgrades. With these moves, the firm hopes to push humanoid robots from laboratory curiosities into everyday workplaces, from factories to retail stores.
Read more
Step inside ITER, the world’s biggest fusion experiment, where scientists from dozens of countries are trying to copy the Sun’s power plant on Earth. Inside the massive doughnut‑shaped chamber, hydrogen atoms are heated to temperatures that dwarf even the Sun’s core—about ten times hotter—creating a plasma that can fuse together and release huge amounts of clean energy. Unlike traditional nuclear power, fusion produces no long‑lived radioactive waste and uses fuel that’s abundant in seawater. The video walks viewers through the towering superconducting magnets that squeeze the plasma, the ultra‑precise laser systems that monitor every reaction, and the international teamwork that keeps the project moving forward. It also explains why achieving and sustaining those extreme temperatures is such a scientific milestone and how, if successful, ITER could pave the way for a new era of carbon‑free electricity. With climate change accelerating, the promise of limitless, safe power is more exciting than ever. The footage captures the awe‑inspiring scale of the reactor, the dedication of the engineers, and the hopeful vision of a future where the Sun’s energy is harnessed right here on our planet.
Read moreA research team from the University of Science and Technology of China has unveiled a major breakthrough in photonic quantum computing with their new prototype, Jiuzhang‑4. The machine performed a 3,050‑photon Gaussian boson‑sampling task, shattering previous records and showing that large‑scale, low‑loss photon networks are now possible. The key innovation is a “cubic‑scale” boost in connectivity: instead of adding more hardware to increase computing power, Jiuzhang‑4 re‑uses the same set of photons across different time slots, dramatically expanding the ways photons can interfere with each other. This space‑time hybrid encoding sidesteps the usual problems of photon loss and interference that have limited photonic computers to tiny experiments. Operating at room temperature, the system avoids the costly cryogenic setups required by superconducting qubits, making it a strong candidate for future quantum‑internet nodes and specialized tasks like drug discovery, material design, and complex optimization. Professors Xiao Lei and Xue Peng highlighted that China now leads both the photonic and superconducting routes to quantum advantage, moving the technology from laboratory demos toward practical, scalable applications that could power the next generation of quantum networks.
Read more
Deep Fission, a fledgling company that wants to build underground nuclear reactors to power AI data centers, has announced a fresh push to go public on Nasdaq. The firm is seeking to raise about $157 million by selling shares at $24‑$26 each, which would peg its valuation at roughly $1.66 billion if the deal goes through. The announcement comes after a confusing episode last fall, when Deep Fission claimed it had already gone public through a reverse‑merger with a Delaware shell called Surfside Acquisition. In that deal the private company bought the publicly listed shell, supposedly raising $30 million at $3 per share. However, a quick search of the OTCQB market turned up no record of Deep Fission’s stock, and the company’s own S‑1 filing now says it has never actually traded publicly. The new Nasdaq IPO is the first genuine public offering the company hopes will attract the capital needed to dig and operate its subterranean reactors. Investors are being asked to back an ambitious, high‑risk vision, but the lack of a clear trading history and the lofty valuation raise plenty of eyebrows. The coming weeks will reveal whether Deep Fission can convince the market that its underground nuclear dream is worth the gamble.
Read moreChinese researchers have unveiled a new quantum‑computing prototype called “Jiu Zhang No.4” that could change the game for optical quantum computers. Instead of adding more lasers and detectors—a strategy that quickly runs into loss and interference problems—the team introduced a “space‑time hybrid encoding” technique. The same set of hardware is reused in different time windows, much like a classroom that can host multiple lessons in the same desks. This clever scheduling lets photons interact across both space and time without the need for extra components, dramatically reducing loss while boosting the number of connections between photons. The result is a cubic‑scale jump in connectivity: the machine handled 1,024 input states, produced 8,176 output modes, and achieved interference involving more than 3,000 photons—far beyond previous records. By sidestepping the hardware‑scaling bottleneck, Jiu Zhang No.4 moves optical quantum computing from small‑scale demos toward practical, large‑scale systems that could power specialized tasks such as quantum networking, sampling problems, and even drug‑discovery simulations. Experts say this breakthrough, together with China’s progress in superconducting qubits, positions the country as the only nation to claim quantum‑superiority on two major platforms. The new architecture promises a more affordable path to fault‑tolerant, room‑temperature quantum machines, bringing the quantum era closer to everyday applications.
Read more
A new BloombergNEF forecast says solar power will become the world’s biggest source of electricity by 2035, finally overtaking coal, oil and natural gas. The surge is being driven by two massive trends: the rapid electrification of whole industries and the exploding appetite for artificial‑intelligence computing. Countries are already feeling the shift. Pakistan, for example, installed 25 gigawatts of solar capacity in the past two years after natural‑gas prices spiked following Russia’s invasion of Ukraine. But the story isn’t all sunshine. Data centers that power AI models, cloud services and streaming platforms are set to become the single biggest new electricity customer. BloombergNEF predicts they will add enough demand to justify an extra 1 terawatt of utility‑scale solar, but they will also call for roughly 400 GW of solar, 370 GW of natural‑gas and 110 GW of coal capacity. Because gas and coal plants can run 24/7, the firm expects fossil fuels to supply about 51 % of the additional power needed for data centers through 2050. In short, while the global grid tilts toward clean energy, the AI boom ensures that oil, gas and coal will remain part of the mix for years to come.
Read moreChina has finished wiring the nation with the world’s most advanced communications backbone. By the end of March 2025, the country operated almost 5 million 5G base stations, giving roughly two‑thirds of mobile users 5G service and extending gigabit‑speed fiber to every county and township. The total length of optical cable now tops 74.99 million kilometres – enough to circle the Earth‑Moon distance a hundred times. Looking ahead, the government is already betting on 6G, slated to roll out around 2030. Analysts project a trillion‑yuan (about US$140 billion) ecosystem of products and services by 2035, turning the network from a simple “connection” platform into an “empowerment” engine for AI, autonomous vehicles, smart cities and beyond. In the meantime, telecom‑service giant China Communications Services (00552.HK) reported 2025 revenue of 150 billion yuan, a modest 0.1% rise, with net profit of 3.61 billion yuan and a dividend increase to 0.2241 yuan per share. The firm booked new contracts worth roughly 106 billion yuan in the first half of 2025, driven by AI‑enabled solutions and digital‑transformation projects. The industry is also navigating geopolitical headwinds: China’s Ministry of Commerce launched anti‑dumping investigations into U.S. analog chips, while the China Communications Enterprise Association called for a fair, open market to protect domestic innovators. All signs point to a rapidly expanding, high‑tech communications landscape that could reshape the Chinese economy in the next decade.
Read more