DAE!🚀🚀 Blockchains are the financial and economic rails of the future. Almost 1mm autonomous AI agents have already transacted using x402 protocol. Exponential growth from a year ago. @50TFunds portfolio companies at the forefront of this structural change. Early Days!

Ultra-thin, ultra-precise. ⚪️Jade Ultra & 🔴Obsidian Ultra DIY Mouse Skates D6.5mm Universal Dots—now at just 0.85mm​ thick. 🚀Slimmer than our standard 1mm Ultra skates by 0.1–0.15mm, these are engineered for the perfect LOD tuning. ⚡️Elevate your aim today.

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.

Panasonic Industry, a subsidiary of Panasonic Holdings (6752 JP), will triple prices for some of its general-purpose copper-clad laminates shipped from Jan. 1, 2027, Taiwan’s Commercial Times reports. The increase applies to all R-1705 specifications and to listed products at least 1.1 mm thick in the R-1766, R-1566 and R-1755 series. It covers production at Panasonic’s plants in Japan, China, Taiwan and Thailand. Panasonic said changes in market conditions and demand patterns had made current prices difficult to maintain for some products. The company previously raised prices for its glass-epoxy multilayer copper-clad laminates by 30%, effective for shipments from May 1, 2026. Kingboard Laminates (1888 HK) also raised prices by 15% for laminates and prepreg made with 2116-grade or thinner glass fabric, effective for new orders from Oct. 13. It was the company’s eighth increase of 2026. Kingboard cited tight supplies of glass fabric and yarn.

We took part in the research preview of MHS from @AnthropicAI Here is Claude Code running a real SO-ARM101. Nothing in this was trained - no policy, no teleoperation, no demonstrations. The agent measured the workspace itself and wrote the motion. The calibration is the interesting part. No checkerboard, no camera intrinsics: the arm is its own ruler. Torque drops, a human rests the closed gripper on 16 dots the software draws in the camera view, and the robot reads back where they are. 4.1 mm position accuracy, 3.0 mm placement. Best run so far: 12 bricks placed with all four colour groups formed. A full hands-off run, start to finish, is what's next. Research preview today, open source coming soon.