News Summaries

Quantum Threat? Peter Shor Says He’s Not Freaked Out About Cracking the Internet

Peter Shor, the mathematician behind the famous quantum‑computing algorithm that could render today’s encryption obsolete, says the danger is real but not a cause for panic. He notes that quantum computers are still “toys” – fragile, error‑prone machines that can’t yet run his algorithm at scale. Yet the rapid progress in both hardware size and error‑correction techniques is turning those toys into serious tools faster than many expected. Shor stresses that quantum machines won’t magically outperform classical computers at everything; they excel only at specific problems like factoring large numbers, which underpins most online security. He doubts they’ll ever predict stock markets or replace everyday computing. Governments and tech giants are already reacting. Google aims to finish its shift to post‑quantum cryptography by 2029, and a recent U.S. executive order mandates that high‑impact federal systems adopt quantum‑resistant safeguards by 2031. While the clock is ticking, Shor remains confident that the scientific community will stay ahead, developing new safeguards before quantum computers become powerful enough to break the internet’s current locks.

China’s Open‑Source AI Giants Capture One‑Third of the Global Market

China’s home‑grown open‑source large language models (LLMs) now represent roughly 30 % of the worldwide market, a milestone that signals the country’s growing influence in artificial‑intelligence research and commercial deployment. The figure was highlighted at the 2023 China Open‑Source Conference, where industry leaders gathered under the theme “Open Source United, Joint Effort for Win‑Win.” Attendees discussed how collaborative development, shared codebases, and community‑driven innovation are accelerating breakthroughs in robotics, semiconductor design, and intelligent‑agent architectures. At the same event, analysts from Omdia projected China’s semiconductor market to exceed $812 billion by 2026, driven by a 262.9 % surge in storage‑chip sales. Parallel advances in robot body design and embodied intelligence—showcased by firms like GRG Banking—are pushing humanoid robots from labs into public spaces such as shopping malls in Changsha. The rapid expansion of open‑source LLMs is also reshaping talent pipelines: developers are increasingly leveraging community tools for rapid prototyping, while enterprises are betting on these models to cut costs and speed up product rollout. As China’s AI ecosystem matures, the blend of massive data resources, government support, and a vibrant developer community is set to keep its share of the global AI market on an upward trajectory.

China’s Green Power Push Powers Up AI – How Open‑Source Models Are Winning the Global Race

China is turning its wind‑blown plains and sun‑lit deserts into a massive, low‑cost power grid that feeds the country’s newest AI super‑computers. By locating data centers in Inner Mongolia, Xinjiang and other western regions where electricity is cheap and clean, China solves the “energy anxiety” that has slowed AI growth elsewhere and makes the most of its abundant renewable resources. At the same time, Chinese firms are rolling out open‑source large‑language models that match the performance of top U.S. systems but cost a fraction of the price to run. Platforms such as DeepSeek and Zhipu have slashed token‑processing fees, prompting a surge in usage: Chinese models handled 23.5 trillion tokens in a single week, outpacing U.S. models for ten weeks straight. Their open‑source nature lets startups, midsize companies and developers tap cutting‑edge AI without huge R&D budgets. The government is pushing “AI+” across the economy, embedding intelligent models in factories, logistics, retail and even household robots. More than 80 % of key industries now use AI, boosting productivity and driving a core AI market worth over 1.2 trillion yuan. Looking ahead, China is eyeing space‑based computing and quantum chips to keep its AI advantage soaring.

Guangdong Rolls Out Massive 5G‑A, 6G and 10G Optical Network Push to Power Future Tech

China’s top leaders are fast‑tracking the next wave of digital infrastructure. In a sweeping policy brief, the State Council called for an aggressive rollout of the enhanced 5G‑A network, while also urging research into a future 6G system and satellite‑based internet services. In Guangdong, the provincial government unveiled a ten‑year action plan that blends cutting‑edge research with real‑world deployment. The plan targets ultra‑fast optical terahertz links, full‑optical communications, and integrated space‑sea‑land networks, while backing the development of new digital baseband chips, RF front‑end chips, 6G modules and next‑generation network gear. Meanwhile, neighboring Guangxi is using the same momentum to upgrade its urban backbone. Its five‑year urban renewal blueprint calls for massive upgrades to underground pipelines, flood‑control systems, and water‑treatment facilities, all built to civil‑defense standards. The region will also pilot “factory‑network integration” projects and performance‑based payment models to ensure long‑term sustainability. Together, these initiatives aim to create a seamless, high‑capacity communications fabric across South China, laying the groundwork for everything from smart factories and autonomous vehicles to resilient emergency services. The coordinated push for 5G‑A, 6G research, and a 10G optical backbone signals China’s ambition to dominate the next generation of global connectivity.

Embodied AI Breakthroughs 2025‑2026: From Seeing the World to Acting in It

The 2025‑2026 research wave in embodied intelligence has delivered a series of game‑changing advances that bring machines one step closer to truly human‑like perception and action. Scientists have unveiled new sensor‑fusion algorithms that let robots combine visual, tactile, and auditory data in real time, creating a richer, more reliable picture of their surroundings. Coupled with these perception upgrades are novel motor‑control frameworks that translate that sensory input into fluid, adaptive movements—think robots that can pick up delicate objects, navigate cluttered homes, or even assist in emergency rescues without pre‑programmed paths. Key highlights include: • A unified learning model that simultaneously trains vision and motion modules, cutting training time by half. • Energy‑efficient hardware chips that process sensory streams on‑device, reducing latency and dependence on cloud servers. • Open‑source benchmarks that let researchers compare embodied agents across tasks like object manipulation, locomotion, and social interaction. These breakthroughs promise smarter service robots, more responsive prosthetics, and autonomous systems that can learn on the fly. The community is encouraged to join the DAMO Developer Matrix platform, share findings, and collaborate on the next frontier of embodied AI.