Reciprocating Perspectives on Manufacturing Technology

This week: Hyundai Metaplant, Real-World AI for Manufacturing, AI-Powered Pallet Jack, titanium idiot index, heating up electron transfer reactions, Scalable PVC Recycling, Embodied Foundation Models

Nov 09, 2025

∙ Paid

Shop Talk

Capturing this week’s zeitgeist

I keep this weekly digest free because I am thrilled with the continued community growth of industrial technologists and venture capitalists. I have one ask: if you find value in the time and effort I put into curating from thousands of sources each week (accessible below the paywall), support this publication with a subscription 👇

Financial support isn’t the only way to help, and that’s perfectly fine. You can still provide tremendous support by helping this publication grow. If you found value in this issue, please forward it to a friend, share it on social media, or post it in your organization’s chat.

Assembly Line

This week’s Industry 5.0 breakthroughs and frontier technologies of the built world.

🏭🚙 Inside Hyundai’s Massive Metaplant

Hyundai’s Metaplant leverages a suite of advanced technologies to create a hyperefficient EV manufacturing process, driving business value through precision, cost savings, and human-robot collaboration. The factory relies on AI-controlled robots for tasks like welding and 300 automated guided vehicles (AGVs) that manage an AI-based just in time logistics system, which eliminates waste and stockpiling. A key value is using collaborative robots for heavy, repetitive tasks, such as installing doors, which they perform with perfect precision, freeing human workers to focus on high-skill craftsmanship. This high-automation strategy, supplemented by technologies like Boston Dynamics’ Spot robots for weld inspection, allows the plant to reduce labor costs and material scrap, rival global competitors, and pursue its goal of 100% renewable energy.

🧠🎛️ Building Real-World AI for Manufacturing

Jonathan Wise, Chief Technology Architect at CESMII and veteran leader across Rockwell Automation, Microsoft, GE, and Amazon, joins The Manufacturing Intelligence Podcast to unpack what it really means to be “AI-ready” in manufacturing and what it takes to build real-world AI for manufacturing that works.

Drawing from decades of experience and over 50 digital transformation projects, Jonathan discusses the hidden complexity beneath data structures, why unified namespaces aren’t enough, and how AI success depends on the right semantic and ontological foundation. He also offers a refreshingly honest perspective on standards adoption, hallucination risks in LLMs, and how to bridge both technological and generational divides on the factory floor.

🧠🎛️ How Enterprise Industrial Leaders Are Applying AI to Systemic Manufacturing Problems with Emily Nguyen of Palantir Technologies /AI in Business/

🧠 AI-Powered Pallet Jack Detects Warehouse Obstacles

Manufacturing and logistics operations are no exception when it comes to charging digital transformation. The automation techniques being adopted by MiR in its latest offering, the MiR1200 Pallet Jack, is setting the pace in future minimal requirements in warehousing and logistics in the U.S. market.

💿 Johns Hopkins ChemEs Crack Crucial Chip Challenge

✍️ Author: Seán Crevan Ottewell

The research team had previously shown that new resists made of amorphous zeolitic imidazolate frameworks (aZIFs) could withstand the higher-powered radiation sources being tested. Working with Swiss, Chinese and other U.S. research organizations, the group developed a new methodology called chemical liquid deposition (CLD). Using this, researchers can study various combinations of aZIFs easily and quickly, according to a Sept. 11 article in Nature Chemical Engineering.

At the heart of the process is a 3D-printed reactor with a well-stirred mixing region, leading to a laminar flow coating region. Aqueous solutions of zinc nitrate and 2-methylimidazole (2-mIm) linker are fed into the reactor separately at 2 ml/minute via a two-channel syringe pump. The two solutions merge and enter the 0.24 ml well-stirred section for vigorous mixing by a stir bar. After a few seconds there, a moderately reactive, non-aggregated liquid mixture forms, from which smooth and continuous aZIF films can be continuously deposited by CLD under laminar flow conditions in the reactor’s second section. The flow field in the microreactor is designed to allow the results of film depositions to be analyzed quantitatively.

New Product Introduction

Highlighting new and innovative facilities, processes, products, and services

♻️ Breakthrough U.S. Demo Plant Advances Scalable PVC Recycling

The Vinyl Institute (VI), a U.S. trade association representing the leading manufacturers of vinyl, is proud to support Plastic Back in its launch of a first-of-its-kind pilot project, aiming to enhance PVC recycling rates and showcase that PVC can be efficiently transformed into valuable raw materials and reintegrated into industrial supply chains. The project, to be commissioned in Q1 2026 at a Freepoint Eco-Systems’ facility in Hebron, Ohio, will test and scale a novel chemical recycling technology developed by Plastic Back. Supported in part through funding from the VI’s VIABILITY grant program and the BIRD Energy program managed by the US-Israel BIRD Foundation, the pilot will process PVC-rich waste streams and serve as a collaborative platform—bringing together waste handlers, petrochemical companies, brand owners, and technology providers to demonstrate a scalable circular solution for PVC.

🧠🤖 GEN-0 / Embodied Foundation Models That Scale with Physical Interaction

For years, foundation models in robotics have primarily used vision-language pretraining as the stepping stone towards scaling robotics, allowing us to transfer the benefits of semantic generalization from existing large multimodal models. But what’s been missing is how to effectively scale large multimodal model training in the domain of robotics itself—to establish scaling laws that corroborate the consistent (and predictable) improvement of robot intelligence with more compute & data, as has underpinned progress in other domains e.g. LLMs. This requires an architecture, training procedure, and data engine that pushes new sensorimotor capabilities, provides behavioral generalization, and grows with the vast and ever-expanding experience generated by interacting with the real physical world.

To this end, we’re introducing GEN-0, a new class of embodied foundation models built for multimodal training directly on high-fidelity raw physical interaction. Its architecture builds on the strengths of vision and language models while also going beyond them—natively designed to capture human-level reflexes and physical commonsense. One core feature is Harmonic Reasoning, in which the models are trained to simultaneously think and act seamlessly.

Read more at Generalist AI and discuss on X

AgiBot deploys its Real-World Reinforcement Learning system /The Robot Report/
Figure and Agility Robotics spat /X/
RPT-1: SAP Launches a Relational Foundation Model for the Enterprise /The New Stack/

TradingView chart — **It matters what you build**: Caterpillar to Double Gas Turbine Capacity to Feed AI Boom /Bloomberg/ while Eaton’s Electrical Sector and Electrical Americas data center orders are each up approximately 70% and revenue up approximately 40% versus 3Q24 demonstrating continued strong momentum in this rapid growth market /Eaton/ and Owens Corning Swings to a Loss Amid Weak Residential Demand Trends /WSJ/

Business Transactions

This week’s top funding events, acquisitions, and partnerships across industrial value chains.

📑 Factories as a Product with CynLr /Wevolver/
📝 This time is different for Physical AI? /X/ Intelligent robotics: The new era of physical AI /BVP/ while DEEPX Unveils ‘Physical AI’ Vision at World Economic Forum /PR Newswire/
🏚️ K-Scale Labs is ceasing operations due to financial constraints /YouTube/ open-sources all IP /Humanoids Daily/
📚 What we’ve learned about next generation hardware manufacturing /Shapeways/

🇨🇭🧠 mimic robotics raises $16 million to deploy frontier physical AI across industries

mimic, a Zurich-based robotics company, has raised $16 million in funding led by Elaia, alongside Speedinvest, to deploy its frontier physical AI across industries, enabling robots to handle complex, dexterous tasks that conventional machines cannot. This heavily oversubscribed seed round also included participation from Founderful, 1st kind, 10X Founders, 2100 Ventures and Sequoia Scout Fund, bringing mimic’s total funding to over $20 million. The new capital will accelerate the development of mimic’s foundation AI model and humanoid robotic hands, and advance deployments with leading global industry players.

mimic builds frontier physical AI models trained on real-world human demonstrations, using innovative methods to overcome the data scarcity problem in robotics. Skilled operators wear mimic’s proprietary data collection devices while performing their daily work on factory floors, capturing detailed movement data from live production settings without disrupting operations. These demonstrations are then used to train AI models via imitation learning, enabling mimic’s humanoid robotic hands to reliably reproduce human technique. The company’s physical AI models ensure that robots autonomously react to changing positions and orientations of objects, handle disturbances and self-correct their actions, seamlessly operating in environments designed for humans.

Read more at GlobeNewswire and discuss on X

🇮🇹 Adaptronics raises €3.15M to revolutionise robotic manipulation /Tech EU/
🇰🇷 This New Artificial Muscle Could Let Humanoid Robots Lift 4,000 Times Their Own Weight /ZME Science/
OpenMind, Robostore launch academic training program for humanoid robots /SiliconANGLE/
Reinforcement learning implementation for Unitree robots, based on IsaacLab /GitHub/
Humanoid Robotics Industry Awards 2025 - Winners /YouTube/

🇺🇸 Mine Vision Systems Announces $12.5 Million Series A to Accelerate Underground Mining Decision Platform

Mine Vision Systems (MVS), the leader in vision-driven mapping and analytics for underground mining, announced a $12.5 million Series A financing led by Rockwell Venture Capital with a significant commitment from Condire Investors. The capital will fuel rapid execution of MVS’s product strategy and expansion of its decision-making platform for underground operations, building on strong industry momentum in 2025. Proceeds from the round will accelerate R&D and product integration, expand deployments of FaceCapture and adjacent analytics, and strengthen customer success in key mining regions.

🇺🇸 IndustrialMind.ai Raises $1.2 M Pre-Seed to Bring the AI Engineer to Manufacturing

IndustrialMind.ai, an AI company founded by former Tesla manufacturing AI leaders, announced a $1.2 M pre-seed round to advance its mission of helping every factory become best-in-class. Investors in the round include Antler, TSVC, Plug and Play, and an angel investor. The funding will accelerate development and customer deployment of IndustrialMind.ai’s “AI Engineer,” with capabilities ranging from drawing-to-process automation to real-time monitoring and root-cause analysis, improving yield, increasing throughput, and shortening new-product-introduction timelines.

🇪🇺🇩🇪 Industrial AI Made in Europe: Digital Sovereignty Through Partnership and Innovation

SAP and a group of European and global partners launched the Industrial AI Cloud project today, marking a major milestone on the path to implementing the vision of “Industrial AI Made in Europe.” At the heart of this effort, SAP is working in concert with Deutsche Telekom, NVIDIA, Siemens, Deutsche Bank, Perplexity, PhysicsX, and Agile Robots. Together, the AI Cloud project partners plan to help shape Europe’s digital future and facilitate the application of advanced AI technologies in European industry, beginning with a partnership between Deutsche Telekom and NVIDIA totaling around €1 billion.

The initiative, which is backed by the German Federal Ministry for Digital Affairs and State Modernization, is designed to transform Germany into Europe’s AI hotspot and enable European companies to access secure and sovereign computing resources, as well as help them accelerate innovation and unlock the full potential of industrial AI — responsibly, collaboratively, and on Europe’s terms — drawing on Europe’s vast corporate data reserves.