Silicon Photonics Chip Race Heats Up: China Joins Global Foundry Battle

Silicon Photonics Chip Race Heats Up: China Joins Global Foundry Battle

The race to mass‑produce silicon photonics chips – tiny devices that turn electrical signals into light – is gathering steam worldwide. While industry giants such as TSMC, Samsung, Intel, GlobalFoundries, Tower and UMC are pouring billions into new factories and advanced packaging, Chinese players like SMIC, Huali Semiconductor and Yuexin are quietly stepping into the foundry arena, hoping to catch up in the so‑called post‑Moore era. AI, cloud computing and big‑data workloads are driving an explosion in demand for ultra‑fast optical links. Forecasts suggest that 100 GbE and faster Ethernet chips will jump from 36.6 million units in 2024 to over 80 million by 2029, with silicon photonics chips leading the surge – rising from 9.6 million to 45.5 million in the same period. TSMC is already testing a breakthrough CPO (co-packaged optics) technology on its 3 nm process, aiming to ship samples in 2025 and enable 1.6 Tbps optical transmission later that year. It is teaming up with firms like Broadcom, Ayar Labs and Lightmatter while filing a flood of patents. Samsung is matching the push, establishing a dedicated R&D hub in Singapore and recruiting top talent worldwide to secure a slice of the market. China’s massive domestic demand, strong policy backing and growing expertise could give it a foothold, but the industry still faces hurdles in equipment, materials and talent. The outcome of this global foundry war will shape the future of AI chips and high‑speed data centers.

Read more
Silicon Photonics Chip Race Heats Up: AI Fuels a Global Foundry War

Silicon Photonics Chip Race Heats Up: AI Fuels a Global Foundry War

The race to mass‑produce silicon‑photonic chips is turning into a full‑blown foundry war, driven by the exploding demand for AI, cloud and high‑speed data centers. Chinese players such as SMIC, HHGrace and CanSemi are quietly testing the waters, hoping to tap a massive domestic market and government backing to catch up in the post‑Moore era. At the same time, global giants are throwing serious weight behind the technology. TSMC plans to ship its first 1.6‑terabit optical modules by late 2025, using a new CPO (co‑packaged optics) process built on its 3 nm node and partnered with Broadcom, Ayar Labs and other Silicon Valley innovators. Samsung has launched a dedicated silicon‑photonic R&D hub in Singapore, recruiting former TSMC and Intel talent to accelerate its own roadmap. Tower, GlobalFoundries, Intel, UMC and STMicroelectronics are also expanding capacity or acquiring assets to secure a slice of the market. Forecasts from LightCounting show 100 GbE and faster Ethernet optical chips climbing from 36.6 million units in 2024 to 80.5 million by 2029, with silicon‑photonic parts growing fastest—from 9.6 million to 45.5 million units in the same period. While challenges remain in equipment, materials and skilled staff, the AI‑driven surge is giving the industry irreversible momentum, and the outcome of this battle will shape the future of AI‑focused chips.

Read more

China’s Home Robots Get Smarter: New Tech Brings Millimeter Precision to Everyday Tasks

Chinese robots are moving from flashy stunts on TV stages to real‑world helpers in kitchens and living rooms. A research team led by Professor Dong Hao at Peking University has given service robots a kind of “spatial imagination.” Before picking up a flower or stacking a cup, the robot first runs a virtual simulation, maps out exact angles and positions, and then checks its own movements step‑by‑step, correcting any drift. In lab tests the system completed seven delicate chores with a 79 % success rate, and in actual homes it achieved 68 % success across tasks such as arranging bouquets, closing jars and stacking cups – a big leap over earlier methods that often failed at millimeter‑level alignment. At the same time, heavy‑duty robots are becoming more autonomous and reliable. A joint project by Midea’s Lancheng Laboratory, KUKA (Guangdong) and Guangdong Jiya built a new generation of 350‑1200 kg robots that meet international performance standards, boast a mean‑time‑between‑failures of 100,000 hours, and incorporate home‑grown servo motors, reducers and smart‑control software. These machines are already being used in high‑tech sectors such as new‑energy manufacturing and aerospace. Backed by strong government policies and billions of yuan in funding, China’s embodied‑intelligence industry is projected to hit ¥400 billion by 2030 and could top ¥1 trillion by 2035, signaling a rapid shift from laboratory breakthroughs to everyday applications.

Read more

China Leads the Charge as Humanoid Robots Move From Lab to Real‑World Jobs

A new report from the UK’s Alan Turing Institute flags “embodied intelligence” – AI that can see, think and act in the physical world – as one of 2026’s hottest tech trends. Sensors, actuators and natural‑language processing are now being packed into robots, drones and autonomous vehicles, turning AI from a cloud‑only concept into a tangible, on‑the‑ground force. Industry analysts say the breakthrough moment has arrived: large‑scale commercial roll‑outs are replacing laboratory demos. Deloitte and the World Economic Forum both note that physical AI is ready for mainstream use, and market research firm M&M projects the global embodied‑intelligence market to jump from $4.44 billion in 2025 to about $23 billion by 2030 – a 39 % annual growth rate. China is at the forefront. From fully autonomous robots showcased at CES to humanoids performing alongside hosts on the Spring Festival Gala, Chinese firms have captured worldwide attention. Government backing, a ¥100 billion industrial fund and inclusion of embodied intelligence in the 2025 work report have accelerated development, with shipments of humanoid robots reaching roughly 18,000 units in 2025 – most of them made in China. Challenges remain, such as immature large‑model robotics and limited high‑quality training data, but experts predict a “ChatGPT moment” for physical AI within years. If breakthroughs occur, the hype could outpace even the mobile‑internet boom, reshaping logistics, manufacturing, transport and everyday life.

Read more

Green Power Meets Chemistry: New Method Turns Waste Electricity into Eco‑Friendly Glycolic Acid

Researchers at China’s Institute of Physics and Chemistry have unveiled a simple, low‑cost way to convert ethylene glycol—a widely produced chemical—into glycolic acid using electricity from renewable sources. The breakthrough is especially important for China’s north‑west region, where wind and solar farms generate more clean power than the local grid can absorb. Instead of letting that surplus electricity go to waste, the team uses an electro‑catalytic reaction that works in a neutral (non‑acidic) solution, avoiding the hazardous chemicals and high energy demands of traditional glycolic‑acid production. Why it matters: Glycolic acid is the building block for biodegradable plastics such as polyglycolic acid (PGA), which are in high demand as governments tighten bans on single‑use plastics. China currently produces less than 50,000 tons of glycolic acid a year, far short of the projected million‑ton demand. By turning excess green electricity into a valuable chemical, the new process tackles three problems at once—absorbing surplus renewable power, upgrading the region’s massive ethylene‑glycol industry (which accounts for over 80 % of the nation’s capacity), and providing a cleaner feedstock for eco‑friendly plastics. The result is a win‑win for the environment, the economy, and the push toward a circular, low‑carbon future.

Read more