News Summaries

China’s Space Station Delivers Record‑Breaking Science: From Space‑Born Mice to Ultra‑Hot Materials

In 2025 the Chinese space station became a bustling laboratory, sending up 867.5 kg of experiments and bringing back nearly 84 kg of samples packed with more than 150 TB of data. Researchers from dozens of disciplines tackled big questions, filing over 50 patents along the way. A highlight was the first mouse experiment in orbit, which proved a complete life‑support loop—screening on Earth, live transport up, feeding in microgravity, and safe return—paving the way for systematic studies of how mammals react to space. In microgravity physics, scientists unraveled how high‑temperature alloys solidify, clarified phase formation in iron‑based magnetostrictive alloys, and discovered why super‑cooled liquids switch from “strong” to “brittle.” They also pushed laser‑heating tech to 3,100 °C without a container, a breakthrough that could guide the creation of ultra‑high‑temperature materials both in space and on the ground. New sensor technologies were tested in orbit, confirming their performance in the harsh space environment. Altogether, 31 on‑orbit projects were completed, establishing a versatile suite of facilities that support life science, material research, and cutting‑edge space applications, cementing the station’s role as a world‑leading hub for space‑based innovation.

Anthropic Teams Up with Allianz to Bring Responsible AI to the Insurance World

AI research lab Anthropic is expanding its foothold in big‑business AI, and its newest partner is German insurance heavyweight Allianz. The deal, announced on Friday, will embed Anthropic’s large‑language models—especially the Claude Code coding assistant—into Allianz’s daily workflows, giving every employee access to AI‑driven code generation and custom AI agents that can handle multi‑step tasks while keeping a human in the loop. A third component of the partnership adds a transparent logging system that records every AI interaction, helping Allianz meet regulatory requirements and maintain clear audit trails. The move follows a string of high‑profile contracts Anthropic signed in October, including a rollout of Claude to Deloitte’s half‑million staff and an integration of its models into IBM’s product suite. According to a December survey by Menlo Ventures, Anthropic now commands about 40% of the enterprise AI market and over half of the AI‑coding segment, up from 32% earlier in the year. While financial terms weren’t disclosed, the collaboration signals Anthropic’s aggressive push to become the go‑to AI provider for large, regulated industries like insurance, where responsible, auditable AI is becoming a competitive necessity.

China’s Tianyue Advanced Unveils Giant 12‑inch Silicon Carbide Wafers at National Manufacturing Expo

At the grand “Building a Strong Nation Through Manufacturing” exhibition in Beijing’s National Museum, Tianyue Advanced stole the spotlight by displaying its full line of 12‑inch silicon‑carbide (SiC) wafers. The show, part of the 14th Five‑Year Plan showcase, highlights China’s push to dominate next‑generation semiconductor materials that power electric vehicles, AI chips and augmented‑reality glasses. After a decade of home‑grown research, Tianyue moved from pioneering 2‑inch and 8‑inch SiC wafers to now mass‑producing 12‑inch discs – a size that offers more than double the usable area of its predecessor. The larger wafers cut chip‑making costs, boost yields and improve heat‑dissipation, making high‑performance devices cheaper and more reliable. The company’s breakthrough includes stable crystal growth at 2,300 °C and ultra‑low defect rates, putting it on par with, and in some cases ahead of, overseas rivals. Industry insiders say the new wafers could enable AI processors like NVIDIA’s upcoming Rubin GPU to pack more power into smaller packages, and could help AR glasses shift from niche gadgets to everyday wear by producing many more optical components per wafer. Tianyue’s achievement is being billed as a milestone for China’s semiconductor ambitions, turning a strategic national priority into a tangible, market‑ready technology.

Knitted Satellite Set to Scan Earth with a Radar‑Weave Antenna

A British satellite called CarbSAR is about to lift off, and it’s dressed in something you’d expect to see in a knitting workshop, not a space launch pad. Engineers at Oxford Space Systems have taken a standard industrial knitting machine – the kind used to make jumpers – and adapted it to stitch together ultra‑thin wires of tungsten coated in gold. The resulting fabric forms a lightweight, mesh‑like radar dish that will unfold once the satellite reaches orbit. CarbSAR, built in partnership with Surrey Satellite Technology Limited, is a small, low‑cost spacecraft designed to capture high‑resolution radar images of Earth’s surface. Unlike traditional metal dishes, the knitted antenna is flexible, lightweight, and can be packed into a compact satellite bus, saving both weight and launch costs. The mission, scheduled for launch on Sunday, will test whether this novel approach can deliver the sharp, all‑weather imaging needed for everything from climate monitoring to disaster response. If successful, the UK Ministry of Defence plans to roll out a whole fleet of similar “knitted” satellites later this decade, creating a new generation of affordable, high‑performance Earth‑observation tools. The project also raises questions about the environmental impact of satellite debris, but for now the focus is on proving that a piece of high‑tech knitwear can work wonders from space.

Google’s Gemini AI Set to Power the Next Wave of Humanoid Robots

Robots that walk, lift, and even talk like people are no longer the stuff of science‑fiction. Thanks to cheaper motors, longer‑lasting batteries and smarter sensors, dozens of startups—from Agility Robotics to Tesla’s Optimus—are racing to build human‑shaped machines. Across the Pacific, about 200 Chinese firms are in the same race, according to industry group CMRA. What’s changing the game now is artificial intelligence. Companies such as OpenAI and Tesla are betting that true human‑like intelligence will only emerge when AI can understand and manipulate the physical world. That’s why many AI labs are turning their attention to robotics, hoping to give machines the ability to learn new tasks on the fly. Enter Google DeepMind. In November, the lab hired the former chief technology officer of Boston Dynamics, the legendary robot maker. Rather than building its own robot, DeepMind’s CEO Demis Hassabis envisions Gemini—Google’s latest AI model as a universal “operating system” for robots. Think of Gemini as Android for phones: a flexible brain that any robot maker can plug into, letting a wide variety of humanoid platforms become instantly smarter. If the vision succeeds, we could soon see Gemini‑powered helpers in factories, homes, and beyond, turning the dream of everyday robot assistants into reality.

Scientists Spot Quantum “Memory” in Bilayer Graphene Particles

A research team from the Weizmann Institute has uncovered fresh evidence that a tiny particle system can retain a record of its earlier quantum configurations. Led by Dr. Jehyun Kim and Himanshu Dev, the group used bilayer graphene – essentially two atom‑thin sheets of carbon stacked in a honeycomb pattern – as a playground for exotic quasiparticles called non‑Abelian anyons. These anyons are prized for their potential to store and process information in ways that are inherently resistant to errors, a key goal for future quantum computers. By carefully tuning the electric and magnetic environment of the graphene layers, the scientists were able to guide the anyons along precise paths and then measure how the system’s quantum state evolved. Their measurements revealed that the particles “remembered” aspects of their prior states, a behavior that had only been hinted at in theory until now. This memory effect could be harnessed to create more stable quantum bits, or qubits, that retain information even when disturbed. The discovery not only validates long‑standing predictions about anyon dynamics but also opens a practical route toward building fault‑tolerant quantum devices using readily fabricated graphene structures. The team’s findings are expected to accelerate experimental work in topological quantum computing and inspire new designs for robust quantum hardware.