Physical Intelligence News — 2026-10-20 (Tue) Morning News

Overview

Industrial and logistics physical AI produced the most concrete headlines: Siemens set a target of up to 80% factory automation with AI, and a picking system was reported at 1,300 clothing items an hour (Reuters, drivesncontrols.com). Humanoids stayed at the preview stage, with a 2027 demo teased and a former NVIDIA executive calling robotics immature (Fierce Sensors, 디지털투데이). Toolchains, chips and world models for physical AI kept being announced (AWS, EdgeCortix, Skyfall AI, GlobalFoundries). 【Humanoids and general-purpose robots】 - OpenAI's Altman teased a humanoid demo in 2027 (Fierce Sensors). - A former NVIDIA executive said robotics is still immature, with parts and safety frameworks not yet fully in place (디지털투데이). - A daily video digest covered ABEJA and Mazak, and said OpenAI explicitly referred to a humanoid (YouTube). 【Robot foundation models and world models】 - Skyfall AI unveiled SKY-001, described as a frontier engineering world model (WebWire). - AWS announced the AWS Physical AI Toolchain (Amazon Web Services). - EdgeCortix scaled its Raiden product for physical AI (Jon Peddie Research). - GlobalFoundries is betting on FDX Fusion for physical AI, according to an EE Times piece attributed to Manfred Horstmann (EE Times). - The president of Kawasaki Heavy Industries said he wants faster development of domestic physical AI and will cooperate with competitors (Nikkei). - Southwest, Worlds, DFW and Alaska Airlines discussed physical AI (Future Travel Experience). - A Japanese-language explainer covered what physical AI is, including its mechanisms, data, world models and robots (Pasquale Pillitteri). - A biped robot that lets children learn about physical AI went on sale in the US for about 50,000 yen (CNET Japan).

Manufacturing and factories

- Siemens targets up to 80% factory automation with AI, a board member said (Reuters, Investing.com, TradingView, Mezha.net). - The plan is backed by a 500 Million Euro expansion in Erlangen (BigGo Finance). - Siemens and Nvidia are developing AI-driven factory automation technology (marketscreener.com). - Machine Design published a piece on how AI, automation and design could expand additive manufacturing (Machine Design).

Logistics and warehouses

- An AI system picks 1,300 clothing items an hour with an accuracy of 99.5% (drivesncontrols.com). - The DHL Logistics Trend Radar found that AI takes action while people remain at the center of logistics (DHL). - Japan AI Center HD unveiled a robot that transports cage carts (LOGISTICS TODAY). - A company transliterated as Activarch announced AI and quantum computing for delivery and inventory optimization (LOGISTICS TODAY). - NE announced a feature that reflects inventory transfers between physical stores in Next Engine inventory, aimed at agentic commerce (Nikoniko News). 【Cooking, housework and food service】 No notable developments. 【Crafts, manufacturing and artisan skills】 - NOVARC will showcase physical AI on robotic welding solutions at FABTECH 2026 (Digital Journal). 【Construction, field work and mobility】 - A headline states "World's fastest AI-powered racing drone outflies the best human pilots" (Live Science). - Another report says the drone zips around a race course at over 60 mph (Yahoo). - A demonstration in Oita City used drones and AI to inspect factory equipment (FNN Prime Online). - TomTom partnered with the Chinese AI autonomous driving company DeepRoute.ai (LIGARE).

Funding and business

- Siemens' 500 Million Euro Erlangen expansion accompanies its automation target (BigGo Finance). - AIBotics acquired NovaCore Labs assets to expand its AI, drone and mapping operations in Jamaica (citybiz). - TomTom and DeepRoute.ai entered a partnership (LIGARE). - Tokyo Boeki Group signed a memorandum of basic agreement with Korea's ROAI in physical AI (Nikoniko News).

Physical intelligence in Japan

- Kawasaki Heavy's president called for faster domestic physical AI development and cooperation with competitors (Nikkei). - Yaskawa Electric's earnings were analyzed for the potential of AI-related capital investment (Shikiho Online). - Japan AI Center HD unveiled a cage cart transport robot (LOGISTICS TODAY). - commissure's "Physical AI研究会 #0" study group filled all seats, and the next session is planned for November 24 (PR TIMES).

Statements by key figures

No notable developments. 【Integrated analysis: what is happening】 Two layers appear in the headlines. One is the enabling layer: AWS's toolchain, EdgeCortix and GlobalFoundries chips, Skyfall AI's world model, and the Siemens-Nvidia work (Amazon Web Services, Jon Peddie Research, EE Times, WebWire, marketscreener.com). The other is the application layer in factories and warehouses. Infrastructure is being announced across several companies at once, which suggests suppliers are preparing for demand before large-scale deployment is visible.

Announcements differ in maturity. Siemens' 80% figure is a target, and the 500 Million Euro Erlangen expansion is a plan (Reuters, BigGo Finance). The 1,300 items an hour at 99.5% accuracy is a reported performance figure, and the headline does not say where it is deployed (drivesncontrols.com). The Oita inspection was a demonstration, and NOVARC's welding is a planned showcase (FNN Prime Online, Digital Journal). The only humanoid item is a teased demo for 2027, alongside a former executive's view that parts and safety frameworks are not ready (Fierce Sensors, 디지털투데이).

For readers, this suggests near-term physical AI progress is more likely in industrial automation and logistics than in humanoids. This is inference from the headlines, not a confirmed trend.

What to watch

Whether the Siemens 80% automation target and the Siemens-Nvidia work produce named deployments or measurable results (Reuters, marketscreener.com). 2. Whether the humanoid demo timing (2027) and the parts and safety gaps raised by the former NVIDIA executive change, and whether the "OpenAI humanoid" statements are backed by details (Fierce Sensors, 디지털투데이, YouTube). 3. Japanese cooperation efforts, including Kawasaki Heavy's call for work with competitors and the commissure session on November 24 (Nikkei, PR TIMES).