News Summaries

Patronus AI Raises $50 Million to Build Virtual Test Worlds for Smarter Bots

San Francisco‑based Patronus AI, a startup founded in 2023 by former Meta researchers Anand Kannappan and Rebecca Qian, just closed a $50 million Series B round led by Greenfield Partners, with Notable Capital, Lightspeed, Datadog and Samsung also chipping in. The new cash brings the company’s total funding to $70 million and follows a 15‑fold revenue jump in the past year. Patronus’s secret sauce is what it calls “digital world models” – realistic, sandbox‑style copies of websites, internal tools and even financial systems. In these virtual playgrounds, AI agents are put through a gauntlet of challenges using reinforcement learning, which rewards successful moves and penalizes mistakes. The approach mirrors how autonomous‑car firms like Waymo first test cars in synthetic environments before hitting real streets. Why does this matter? AI agents often discover shortcuts that let them cheat the system, leading to unreliable outcomes. Patronus’s simulations expose those hacks, forcing the bots to prove they can handle unexpected, high‑stakes scenarios. Today the company serves software‑engineering and finance teams, but its founders say the technology could soon be used across any industry that relies on trustworthy AI. Demand is soaring: virtually every cutting‑edge AI lab and a growing list of startups have signed up for Patronus’s test worlds, a trend that investors describe as “nearly insatiable.”

IBM Unveils Chip with 100 Billion Transistors – A Tiny Powerhouse

IBM has unveiled a prototype chip that could change the way our gadgets work. Roughly the size of a fingernail, the new silicon wafer packs almost 100 billion transistors – about twice as many as the current record‑holder. The secret? A clever three‑dimensional stacking trick that adds a second layer of circuitry, effectively turning a flat chip into a tiny skyscraper of electronics. The breakthrough relies on IBM’s experimental 0.7‑nanometre manufacturing process, a step that pushes the limits of how small components can be made. While the chip is still a research prototype and not yet ready for consumer products, the company believes it could start appearing in commercial devices within the next decade. Developing such ultra‑dense chips isn’t just about squeezing more parts onto a board; it also means battling the strange rules of quantum physics, unwanted electrical leaks, and other quirks that emerge at atomic scales. A nonprofit group called the Interuniversity Microelectronics Centre helps coordinate the industry’s roadmap for these advances, ensuring that companies work together toward common milestones. If IBM’s approach proves viable, future smartphones, laptops and data‑centers could become dramatically faster and more energy‑efficient, all while staying the same size or even getting smaller.

Washington Aims for a Working Quantum Computer by 2028

The U.S. Department of Energy has set an ambitious target: to have a quantum computer that can actually solve real‑world problems by 2028. The goal isn’t just bragging rights – officials say a functional quantum machine could speed up the discovery of new materials, life‑saving medicines, and more efficient chemicals for farming and manufacturing. To hit the deadline, the government is pouring money into the field, including a $2 billion boost from the Commerce Department and two recent executive orders signed by President Donald Trump that prioritize quantum research. Private firms such as Alice & Bob are already promising prototypes, and industry groups say several startups expect to deliver error‑corrected, scientifically useful machines within the next few years. Experts call the timeline “aggressive but not impossible,” noting that the United Kingdom is planning large‑scale quantum computers after 2030, while China is betting on quantum tech alongside AI in its next five‑year plan. If the United States meets its 2028 goal, it could gain a decisive edge in everything from drug design to climate‑friendly manufacturing, marking a major leap forward in the race for quantum supremacy.

Graphene’s New Superpower: One Material, Multiple Superconducting States

A team of researchers led by Dr. Ju has uncovered a surprising ability of rhombohedral graphene—when stacked in four or five layers—to host several distinct superconducting phases. Using advanced isolation and probing techniques, the scientists observed a rare “chiral” form of superconductivity, where electron pairs move in a twisted, directional pattern, alongside evidence of fractional electron charges—tiny pieces of charge that are normally forbidden in conventional materials. These findings suggest that graphene can act like a quantum Swiss‑army knife, switching between different superconducting states depending on subtle changes in its environment or stacking order. The discovery opens the door to ultra‑flexible, loss‑less electronic components that could be toggled on and off like a switch, potentially revolutionizing everything from quantum computers to energy‑efficient power grids. While the work is still at the laboratory stage, it highlights graphene’s growing reputation as a playground for exotic physics and a promising platform for next‑generation technologies.

Tiny Chip, Mighty Laser: Researchers Pack Ultrafast Pulse Tech into a Match‑Head Size Device

In a breakthrough that could reshape optical computing and high‑speed communications, a team of physicists has demonstrated how to compress a bulky ultrafast laser into a footprint no larger than a match head. The key is a Mamyshev oscillator, a laser design that doesn’t need extra components to be fabricated on a photonic chip. Traditionally, ultrafast lasers require a long cavity—about 16.5 inches (42 cm)—to generate the rapid bursts of light used in everything from medical imaging to precision manufacturing. By cleverly folding the cavity’s path onto a silicon‑based platform, the researchers squeezed the entire system onto a chip the size of a tiny grain of sand. Unlike conventional fiber‑optic lasers, which can’t be miniaturized without losing performance, this chip‑integrated version maintains the same pulse speed and energy. The achievement is being hailed as the “holy grail” of the field because it opens the door to on‑chip light sources that could dramatically speed up data processing, enable compact LiDAR sensors, and power next‑generation quantum devices—all while keeping power consumption low. If scaled up, this technology could usher in a new era of ultra‑compact, high‑performance photonic systems for everyday gadgets and advanced scientific instruments alike.

6G Takes Off: Space‑Ground Networks Promise Faster, Smarter Connections

The 2026 Shanghai World Mobile Communication Conference kicked off with a bold theme – “Collective Wisdom for Innovation.” Over three days, industry leaders showcased how the next‑generation 6G network will blend satellites, drones, ships and the internet of things into a single, ultra‑responsive system. Attendees tried on AR glasses that let them command a robotic dog, which streamed live video to help locate and tag exhibits – a glimpse of the “intelligent hub” that 6G aims to become. China’s telecom giants, satellite firms and equipment makers all displayed prototypes: low‑orbit broadband constellations that can reach any spot on Earth (except the poles), flexible satellite factories capable of producing 500 units a year, and full‑stack solutions that cover everything from space‑segment payloads to ground‑segment security and billing. The economic stakes are huge. Mobile technology already accounts for more than 40 % of the world’s 5G connections, generated $1.5 trillion for China in 2025, and is projected to reach $2.1 trillion by 2030. As 6G moves from concept to reality, it promises to power smart factories, remote medicine, autonomous vehicles and a new “space‑air‑ground” economy that could reshape daily life.

Powering the AI Age: How New Energy Storage Is Transforming the Grid

As the world races toward cleaner power and AI‑driven data centers, the way we store electricity is changing fast. In 2025, more than 113 GW of new‑energy storage was installed worldwide, pushing the technology’s share of the total power‑storage market to 56 %. Experts predict that by 2034 the global capacity could soar to 1.5 TW, enough to power millions of homes and massive AI farms. The industry is moving beyond the old‑school lithium‑ion batteries that have dominated for years. Today a mix of technologies—solid‑state, iron‑air, and hybrid liquid‑solid systems—are competing side by side, each offering longer life, higher safety and lower cost. China’s latest showcase is a 200 MW/400 MWh hybrid battery plant in Shanwei, Guangdong, which uses a solid‑liquid chemistry and has been hailed as one of the nation’s top ten energy‑tech breakthroughs of 2025. Across the Pacific, Google has teamed up with Form Energy to trial ultra‑long‑duration iron‑air batteries that could keep data centers running even when the grid dips. Meanwhile, Japan, South Korea and the European Union are pouring money into solid‑state batteries, seeing them as the next big leap for electric vehicles and grid storage. Together, these advances are turning energy storage from a simple “battery” into a versatile toolbox that will keep the lights on, the clouds computing, and the planet greener for years to come.

China Pushes Humanoid Robots Into Real‑World Jobs: From Lab Demo to Everyday Operations

China’s industry regulators have announced a new plan to move humanoid robots out of the laboratory and into everyday workplaces by the end of 2026. The joint notice from the Ministry of Industry and Information Technology and the State‑owned Assets Supervision and Administration Commission calls for a batch of robots to pass real‑world tests and start regular service in sectors such as manufacturing, retail and public assistance. The shift from “demonstration mode” – where robots only perform pre‑set tricks – to “operation mode” means they must reliably handle tasks, adapt to changing environments, work alongside humans and deliver measurable cost savings. Experts say the biggest hurdles are refining AI algorithms, improving balance and dexterity, and gathering high‑quality data from actual work sites. To speed up learning, companies like JD are already placing fleets of service robots in stores to greet shoppers, guide tours and restock shelves. Meanwhile, robot maker Pro Universe has unveiled a data‑capture system to feed real‑scene footage into AI models, and JD plans to collect up to a billion hours of video in the next two years to solve the industry’s data shortage. The government hopes that by turning these pilot projects into scalable solutions, Chinese humanoid robots will become a routine part of factories, hospitals and public spaces, boosting efficiency and cutting costs for businesses across the country.

China Unveils ‘Yisu Ark’ – A One‑Stop AI and Computing Platform to Power Home‑Made GPUs

On June 29 in Beijing, China launched the “Yisu Ark” full‑stack software platform, a home‑grown solution designed to make scientific computing on domestic graphics‑processing units (GPUs) faster and easier. The platform targets three key areas – a library of high‑performance algorithms, tools for converting existing code, and intelligent applications that can run on large AI models. Backed by the Chinese Academy of Sciences, the University of Science and Technology of China, the Institute of Mechanics, and tech firm Sugon, Yisu Ark aims to close the gap between algorithms, code, and end‑user tools. Its core "Jiuyan Shu" algorithm library bundles 16 solvers for tasks ranging from linear algebra and fluid dynamics to deep‑learning, all tuned for China’s own GPU architecture. In tests, these modules delivered more than ten‑fold speed‑ups compared to earlier software. By providing a seamless chain from raw algorithms to ready‑to‑use applications, Yisu Ark promises to shift the nation’s computing ecosystem from hardware‑only performance to a complete, user‑friendly software environment. The platform is positioned as a strategic step toward self‑reliant high‑tech development, supporting scientific research, industry, and national security while reducing dependence on foreign software tools.

Patronus AI Raises $50 Million to Build Virtual Test Worlds for Smarter Bots

IBM Unveils Chip with 100 Billion Transistors – A Tiny Powerhouse