Computing
Researchers at the Hong Kong University of Science and Technology have built ReAAP, a reconfigurable processor that pairs a smart software compiler with adaptable hardware to run deep neural networks up to 5.7x faster than a leading GPU and 3.3x faster than an ARM CPU.
Northwestern University researchers jet-printed flexible artificial neurons that fire like real brain cells – realistic enough to trigger live mouse neurons – pointing toward neuromorphic chips that could one day cut the punishing energy cost of AI.
The US Department of Defense has tapped Argonne spin-out Parallel Works to link its supercomputing centres to AWS, Azure, Google Cloud and Oracle through one secure platform – speeding up the AI and simulation workloads now reshaping defence research.
Chinese researchers say a nine-qubit quantum system beat a 10,000-node classical network at weather prediction for under 1% of the cost, a hint that compact quantum machines could undercut today’s giant AI data centres.
Researchers in Texas have created a graphene leaf tattoo that reads a plant’s moisture in real time and can act as an artificial synapse, raising the prospect of forests and fields that monitor fire and drought themselves.
Lloyds Bank has used IBM quantum hardware to successfully flag a money mule hidden in transaction data, an early sign quantum computing could sharpen banks’ fraud defences.
Elon Musk has unveiled AI1, SpaceX’s first orbital data-centre satellite — a 70-metre-wingspan craft carrying a Nvidia-class compute payload that SpaceX wants to mass-produce and fly off Earth’s power grid, days ahead of its IPO.
Multiverse Computing ran a pretrained Llama model through quantum circuit blocks on IBM’s 156-qubit machine and cut its perplexity while adding almost no parameters — the first end-to-end “quantum enhancement” of a production LLM.
Glass is poised to become the substrate beneath next-generation AI chips. Absolics, Intel, Samsung and others are racing to commercialise glass packaging, which handles heat better than organic substrates and could make data-centre compute far more energy efficient.
Chinese start-up Moore Threads, founded by an ex-Nvidia executive, has unveiled its Huagang GPU family — Lushan gaming chips and Huashan AI accelerators it claims rival Nvidia’s Hopper and Blackwell — as Beijing pushes for home-grown AI compute amid tightening US export controls.
Sandia National Laboratories is testing NextSilicon’s reconfigurable accelerator, a chip that rewires itself to the software it runs, promising big power savings over Nvidia GPUs without costly code rewrites.
WHY THIS MATTERS IN BRIEF Small traders need fast low latency access to NYSE for trading and this speed up…