Breaking the Bottleneck | Issue 84
[6/30/2025] Image to CAD, On-Device Robotics, The Great Trade Rearrangement, Factory Staffing, & More!
Breaking the Bottleneck is a weekly manufacturing technology newsletter with perspectives, interviews, news, funding announcements, manufacturing market maps, 2025 predictions, and more!
💥 If you are building, operating, or investing in manufacturing, supply chain, or robots, please reach out - adityabreakingthebottleneck@gmail.com I’d love to chat!
🏭 If you were forwarded this and found it interesting, please sign up!
Interview & Startup Series 🎙️💬
Alongside the usual newsletter, I’m excited to launch a weekly interview series, “Friday 5,” starting in July. The series focused on navigating the industrial software landscape, featuring industrial operators, founders, tech providers, and more. If you are similarly enthusiastic and know people whom I should interview, I’d love to hear from you and connect!
Content I Enjoyed Last Week 🗞️🔬 📚
News:
The Great Trade Rearrangement [McKinsey]
Recent US tariff hikes have amplified long-standing trade tensions, particularly with China. These tariffs correlate with geopolitical distance: the more distant a country is politically from the US, the higher its tariff exposure, with China facing the steepest increases. Even before 2025, US imports from China had dropped by 20% ($100B) between 2018 and 2024, while total imports rose nearly 30%, illustrating a realignment toward allied economies. A core concept introduced is the "rearrangement ratio," which measures how easily the US can substitute imports from China with alternative sources. While goods like chocolate bars (with a low ratio) are easily rearranged, products such as rare earth magnets and Christmas decorations (ratios >1.0) are functionally irreplaceable, exposing vulnerabilities in the US supply chain. Electronics, textiles, and consumer goods, key sectors for US imports, are among the most difficult to reorganize, due to both the high volume and concentrated supply in China. The analysis predicts that Europe will become the central node in the new trade geometry, absorbing 30–65% of US imports currently sourced from China, while also receiving a larger share of Chinese exports that would otherwise go to the US. Vietnam dominates ASEAN’s role, especially in electronics, though its impact is narrower in scope. Surprisingly, Canada and Mexico are unlikely to benefit materially, given their existing saturation of exports to the US. If US buyers secure only 25% of the available global supply for key products, the result could be a $100B domestic supply gap. In contrast, global markets would experience a significant oversupply, characterized by falling prices and increased trade interdependence between Europe and China.
“If the United States no longer imported laptops from China, this could result in a supply shortfall of nearly $26 billion.”
Made in China 2.0 [World Economic Forum]
China’s once-explicit “Made in China 2025” industrial plan has quietly evolved into a more decentralized but equally ambitious strategy, “Made in China 2.0.” Though the original MIC2025 slogan vanished from official rhetoric amid international scrutiny, its foundational goals remain intact and have matured into a broader transformation of China’s manufacturing system. Driven by AI integration, green energy, and an ethos of self-reliance, China now leads in lithium batteries (75% of global output), solar modules (80%), and EVs, while making strides in robotics, smart factories, and industrial AI. However, gaps remain in advanced semiconductors and biopharmaceuticals. MIC2025 acted more as a mobilization framework than a detailed policy playbook, setting targets for local governments and signaling a strategic focus. Today, the evolution of that agenda reflects a systems-level coordination where gains in one sector, like batteries, accelerate innovation in others, such as EVs and grid storage, creating a mutually reinforcing industrial engine. This next phase of China’s industrial rise is powered by a convergence of “process knowledge,” short innovation cycles, and a new generation of technically fluent entrepreneurs. These fast-learning firms, such as BYD and Xpeng, combine vertical integration with AI-driven efficiencies to iterate, reduce costs, and establish global standards rapidly. China is increasingly treating AI as infrastructure, viewing it as a general-purpose layer embedded in manufacturing to optimize production, logistics, and customization in real-time. The fusion of software and hardware in China’s dense industrial clusters enables near-instant field testing, fostering a model of “short-cycle innovation.”
How to Stand Out in a Company’s Global Manufacturing Network [HBR]
Pfizer’s Puurs plant in Belgium offers a powerful case study in how a manufacturing site in a high-cost, low-profile location can secure its survival within a competitive global network. Rather than competing solely on cost or volume, Puurs’ leadership spent two decades methodically building distinct capabilities across five key stages: operational excellence, new product introduction, specialization, responsiveness, and knowledge sharing. By the early 2000s, the plant had adopted lean practices and demonstrated substantial improvements in both cost and quality performance. They invested in engineering talent, positioning the plant to lead in the rapid deployment of new products consistently and eventually establishing a pilot plant that evolved into a corporate knowledge center. A pivotal moment came when Puurs was selected as one of only two sites to manufacture the Pfizer-BioNTech COVID-19 vaccine. This decision stemmed not from cost or location, but from Puurs' reputation for excellence in complex aseptic manufacturing, its ability to scale fast, and its embedded culture of collaboration and innovation. The plant doubled production from 100 million to over 200 million doses per month in early 2021, and its success prompted further €1.2 billion investment. Puurs also built a sophisticated knowledge network, sharing insights across Pfizer’s global operations and partnering with industries such as frozen seafood logistics to solve ultra-cold storage challenges. As one executive noted, “We were asked to make the impossible possible,” and they delivered.
Why Factories Are Struggling to Fill 400,000 Jobs [NYTimes]
“For every 20 job postings that we have, there is one qualified applicant right now.” – David Gitlin, CEO of Carrier Global.
President Trump’s push to revitalize American manufacturing faces a significant hurdle: a shrinking blue-collar workforce. With nearly 400,000 factory jobs currently unfilled and many baby boomers retiring, manufacturers are struggling to attract younger workers. According to the Bureau of Labor Statistics and industry economists, the gap in skilled labor has persisted since at least 2017, driven by cultural shifts that have devalued non-college career paths, inflexible immigration policies, and the demanding nature of modern manufacturing work. Many jobs now require advanced training and software literacy, but vocational enrollment is declining, and high school guidance counselors continue to prioritize college placements over vocational training. Compounding the issue, factory wages often fail to compete with more flexible and comfortable service-sector roles, making recruitment even more challenging. Efforts to reverse the trend include initiatives from groups like the Business Roundtable, where CEOs are rethinking hiring criteria, starting outreach as early as high school, and expanding apprenticeships and training for veterans. Carrier Global CEO David Gitlin highlighted that for every 20 job postings, only one qualified applicant emerges, and estimated the HVAC industry will need 400,000–500,000 new technicians over the next decade. However, federal support has been inconsistent. The Trump administration’s proposed budget cuts $1.6 billion from workforce training, undermining its “Make America Skilled Again” initiative despite its goal of creating one million apprenticeships. Industry leaders, while supporting Trump's manufacturing agenda, voiced concerns over restrictive immigration policies and reductions to programs such as the Job Corps. As Rockwell Automation’s CEO noted, bridging the skills gap will require not just policy shifts but also reshaping perceptions, showing young Americans that modern manufacturing is clean, high-tech, and full of opportunity.
Superconducting Motor Could Propel Electric Aircraft [IEEE Spectrum]
High-temperature superconductors (HTS), long considered a scientific breakthrough in search of viable commercial use, may finally be on the cusp of real-world application. Thanks to falling costs, companies are now utilizing yttrium-barium-copper-oxide superconducting tape to construct powerful magnets for fusion reactors and compact, high-efficiency electric motors, particularly in the aviation industry. One standout is Hinetics, a University of Illinois spinout that recently validated a prototype 5–10 megawatt motor with superconducting rotor magnets, backed by ARPA-E funding. Unlike conventional HTS systems, which require complex cryogenic plumbing to cool fast-spinning rotors, Hinetics has developed a novel approach: a miniature cryocooler mounted directly on the spinning rotor, transferring heat via conduction rather than convection. This design eliminates pumps, seals, and rotating fluid couplings, significantly reducing system complexity and weight. Despite the small cryocooler’s limited heat removal capacity (10 watts), the motor achieves an efficiency of up to 99.5%, approximately 4–5% better than leading permanent-magnet motors, without compromising power density. The prototype utilized slip rings for power delivery; however, future versions will likely employ wireless energy transfer, possibly through inductive coupling. The implications extend beyond aircraft to marine propulsion and other high-torque, low-speed systems, signaling what Haran calls “a transformational new technology” now approaching practical deployment.
Other Reads:
Electric Vehicle Outlook 2025 [Bloomberg]
Energy Groups Spend Record Sums on Power Plants to Feed Data Centers [FT]
How Can We Convince the Next Generation of Manufacturing Workers That the Category Is Innovative? [Manufacturing Dive]
Trump Wants America to Make iPhones. Here’s How India Is Doing It. [NY Times]
LG Opens Michigan Factory to make LFP Batteries [Canary Media]
Product Announcements:
Elementary launches VisionStream, which requires no labeling, vision expertise, or line stoppage to detect defects
Apptronik launches Elevate Robotics to focus on automating industrial tasks beyond the limits of the human form.
IBM launches AI-enabled PFAS screener tool that helps identify and eliminate fluorochemicals from its research operations
Hexagon announces a significant release of ESPRIT EDGE, its industry-leading computer-aided manufacturing (CAM) software.
PTC adds supply chain intelligence to Arena PLM, embedding real-time AI-driven component monitoring and risk mitigation insight directly into product development workflows
Blog / Research:
The first structured 3D generative model that jointly synthesizes multiple semantically meaningful and geometrically distinct 3D meshes from a single RGB image. Unlike existing methods that either produce monolithic 3D shapes or follow a two-stage pipeline, first segmenting an image and then reconstructing each segment, PartCrafter adopts a unified, compositional generation architecture that does not rely on pre-segmented inputs.
Gemini Introduces On-Device Robotics Model [Google Deepmind]
“Our model quickly adapts to new tasks, with as few as 50 to 100 demonstrations, indicating how well this on-device model can generalize its foundational knowledge to new tasks.”
Google DeepMind has unveiled Gemini Robotics On-Device, its most powerful vision-language-action (VLA) model designed to run entirely on robotic hardware without needing network connectivity. Designed for bi-arm robots, this local version of the Gemini Robotics model offers state-of-the-art general-purpose dexterity, rapid adaptation, and instruction-following capabilities across a wide range of real-world tasks, with low-latency inference. Crucially, the model performs well in environments with poor or no connectivity and can quickly adapt to new tasks with as few as 50–100 demonstrations. Developers can access the Gemini Robotics SDK, simulate tasks in MuJoCo, and fine-tune the model through a trusted tester program. Compared to previous on-device systems, it consistently outperforms in generalization, multi-step instruction following, and fine-grained manipulation. Despite being pre-trained for the ALOHA robot, it can be fine-tuned for use on diverse platforms, such as the Franka FR3 and Apptronik’s Apollo humanoid. The model enables real-time execution of tasks using natural language prompts, even with previously unseen objects or environments. Its architecture supports high levels of semantic, visual, and behavioral generalization, and it’s the first such VLA model Google is offering for fine-tuning. Google emphasizes safety and responsible deployment, embedding multiple levels of semantic and physical safety checks via a new benchmark and low-level controllers. As DeepMind notes, this release aims to address core challenges in robotics, particularly latency and connectivity, and is expected to accelerate innovation in the field significantly.
DexWrist: A Robotic Wrist for Constrained and Dynamic Manipulation [MIT]
Podcast/Video:
Manufacturing in a Fragmented World [World Economic Forum]
Finance & Transactions 💵
Venture Capital:
PhysicsX - A company that is deploying AI to transform how physical systems are engineered, embedding intelligence across the entire product lifecycle.
$135 million [Series B] - From Siemens, Temasek, Atomico, and Applied Materials
Tacta Systems - A company that is developing Dextrous Intelligence, the nervous system for robots.
$64 million [Series A] - Led by America's Frontier Fund and SBVA and joined by Matter Venture Partners, B Capital, Woven Capital, and others
AllSpice - A company that is building an AI agent to enable electronics teams to collaborate, automate their workflows, and perform hardware design validation.
$15 million [Series A] - Led by Rethink Impact and joined by L'attidude Ventures, GingerBread Capital, and DNX Ventures
Lidrotec - A company that is developing laser systems for cutting microchips and devices (wafer dicing).
$13.5 million [Series A2] - Led by Lam Capital and Goose Capital and joined by Zeiss Ventures
Bonx - A company that is developing a no-code ERP solution for mid-market manufacturers.
$8.6 million [Seed] - Led by 9900 Capital and joined by Kima Ventures, Purple, OSS Ventures, and Dynamo Ventures