
China’s 2nm AI Chip Race Shakes Global Tech With 40% Efficiency Edge
China’s race to build advanced artificial intelligence chip is quietly gaining pace, with a relatively unknown startup now drawing attention.

China’s race to build advanced artificial intelligence chip is quietly gaining pace, with a relatively unknown startup now drawing attention.

In a major step forward for sensing technology, researchers at the Massachusetts Institute of Technology (MIT) have developed a new

A team of Chinese scientists has developed a new crystal that could transform navigation in extreme environments, from deep oceans

Chinese scientists from Sichuan University and Shenzhen University unveiled the world’s first full-chain framework for directly producing hydrogen from seawater,

A team of researchers has turned to one of nature’s most delicate creatures, the butterfly, to design a material that

Inside Cornell’s NanoScale Science and Technology Facility, a student in cleanroom gear carefully operates a 3D lithography system, where even

In a major step for chemistry and drug development, researchers have designed a new way to edit molecules without rebuilding

China’s Northwestern Polytechnical University has released the world’s first open-source flight control software for bamboo-frame drones. The system cuts control delay from 20 milliseconds down to 8–10 milliseconds by tuning algorithms to handle bamboo’s natural vibrations. Free and modular, it lowers costs for eco-friendly UAVs in forest monitoring and education. The team says future work will improve weather resistance.

UK’s Loughborough University physicists have built a brain-inspired chip that makes some AI tasks up to 2,000 times more energy efficient than software. The memristor device uses random nanopores to process time-dependent data directly in hardware, skipping energy-heavy computing steps. Tests on chaos prediction and image recognition succeeded. The team says the approach could help solve AI’s rising energy demands. Further work is needed for real-world noise.

Harvard’s Stephanie Gil and a multi-university team have created “cy-trust,” a framework that helps robot fleets and self-driving cars decide which data to trust. Each agent assigns a 0-to-1 trust score to incoming information using onboard sensors and signal processing. Low-trust data gets ignored, preventing hacked or greedy agents from causing crashes. Tested in lab experiments against fake-identity attacks. Traditional security isn’t enough for physical systems.
We uphold the highest standards of journalistic integrity and ethical reporting, ensuring that truth and transparency prevail in all our content.
contact@modernmechanics24.com