Breaking the Bottleneck | Issue 46
[4/15/2024] Thintronics, Microsoft's & MITs AI in Manufacturing Report, Sanctuary AI + Magna, & More!
Breaking the Bottleneck is a weekly manufacturing technology newsletter with perspectives, interviews, news, funding announcements, manufacturing market maps, and a startup database!
🗺️ Check out these market maps for new discrete/process manufacturing tech!
💥 Have you seen any interesting startups recently? I’d love to chat with them!
🏭 If you were forwarded this and found it interesting, please sign up!
Content I Enjoyed Last Week 🏭🗞️🔬 📚
News:
US Startup is Taking On Japenese Semiconductor Powerhouse [MIT Tech Review]
In the fiercely competitive semiconductor industry, a Berkeley-based startup, Thintronics, is challenging the long-standing dominance of Japanese company Ajinomoto in the dielectric film market—a critical component used as insulators in computer chips. For over 30 years, Ajinomoto has commanded more than 90% of this market, benefiting from a proprietary material derived from a by-product of MSG. Thintronics aims to disrupt this market with advanced insulating materials that promise faster computing speeds and lower energy costs, aligning with the goals of the $280 billion CHIPS and Science Act intended to revitalize American semiconductor manufacturing. Despite receiving a $20 million Series A funding round and a US National Science Foundation grant, Thintronics faces significant hurdles, including entrenched industry relationships and stringent technical standards. The company's efforts come at a time when US tech giants are increasingly interested in diversifying their supply chains, potentially making them more receptive to new domestic suppliers. Stefan Pastine, CEO of Thintronics, remarked, "Inventing new materials platforms and getting them into the world is very difficult," underscoring the challenges ahead in an industry known for its conservatism.
The EV Battery of Your Dreams is Coming [WSJ]
In the next five years, the electric vehicle (EV) market is poised for significant advancements in battery technology. The imminent advancements primarily involve enhancements to existing lithium-ion batteries rather than complete technological overhauls. This strategic focus allows these improved batteries to be manufactured using existing facilities and supply chains, leveraging the billions already invested in battery manufacturing, such as the $30 billion in 2023 alone. Innovations include BMW’s "Neue Klasse" platform, debuting in 2025, which promises batteries with 20% more energy density and up to 30% faster charging speeds. This improvement is facilitated by new cylindrical cell shapes and chemical adjustments. Further advances are expected from SES AI, a Massachusetts-based company working with major automakers like GM, Hyundai, and Honda. By 2026, SES AI plans to introduce batteries with solid lithium metal instead of graphite, potentially increasing energy density by up to 50%. This could enable even entry-level EVs to achieve 300-mile ranges, with high-end models pushing past current limits. Additionally, StoreDot, an Israeli startup, is developing batteries that could charge extraordinarily quickly—potentially adding 100 miles of range in just three minutes by 2028. This technology involves replacing graphite with silicon, enhancing both the energy density and charging speed. While these developments are promising, the industry also recognizes challenges, such as the physical expansion and contraction of solid-state batteries during use. Companies like QuantumScape are actively working to address these issues, aiming to introduce their solid-state batteries for vehicle testing soon, although large-scale production may still be several years away.
AI Chip Trims Energy Budget Back by 99+ Percent [IEEE Spectrum]
Researchers at Tsinghua University in Beijing have developed a groundbreaking photonic microchip called Taichi, which uses photons instead of electrons to run AI tasks, boasting significant energy efficiency improvements over traditional electronic neural networks. According to a new study, Taichi can perform advanced AI functions with the same effectiveness as electronic devices but consumes only a thousandth of the energy. Taichi leverages a hybrid design that combines diffraction and interference-based approaches in optical neural networks, allowing for reconfigurable computations while maintaining compact and energy-efficient operations. This chip showcases an impressive 13.96 million parameters with energy efficiency reaching about 160 trillion operations per second per watt and area efficiency at nearly 880 trillion multiply-accumulate operations per square millimeter. These metrics signify that Taichi is over 1,000 times more energy-efficient than the latest NVIDIA H100 GPU and vastly outperforms previous optical neural networks in both energy and area efficiency.
Sanctuary AI Enters Strategic Relationship with Magna [Robot Report]
Sanctuary AI is now collaborating with Magna to integrate general-purpose AI robots, specifically the Sanctuary Phoenix models, into Magna’s manufacturing processes. The Phoenix robots are designed with human-like dexterity in hands and arms, featuring advanced manipulation capabilities, perception features, and AI models that control the robots. Sanctuary plans to begin commercial production runs of the Phoenix robots at Magna's facilities, using these deployments as real-world training grounds to gather data and refine the robots' autonomous behaviors.
How Neural Concept’s Aerodynamic AI is Shaping Formula 1 [TechCrunch]
Neural Concept's core product, Neural Concept Shape (NCS), is a machine-learning-based system that enhances aerodynamic design across various industries, including automotive and aerospace. The technology simplifies the computational fluid dynamics (CFD) process, allowing engineers to quickly generate and test aerodynamic models without the traditional reliance on physical prototypes and wind tunnels. The application of NCS in Formula 1 is particularly noteworthy. Teams like Williams Racing use the software to refine their aerodynamics, gaining competitive edges under the sport's stringent testing and budget regulations. NCS significantly accelerates the aerodynamic testing process, reducing typical simulation times from an hour to just 20 seconds, offering teams a crucial advantage. Outside of motorsports, Neural Concept is making strides in the broader automotive industry, applying its technology to enhance the efficiency of electric motors and optimize vehicle heating and cooling systems. The company is also exploring applications in crash testing, although it currently only simulates individual components rather than whole cars.
New DOE-Backed Projects Tackle Factory Emissions [Canary Media]
The Biden administration committed $6 billion to decarbonize heavy industrial sectors, allocating about $500 million specifically to projects addressing industrial heat generation. This funding will support the deployment of various low-carbon technologies, including electric heat pumps, electric boilers, solar-thermal arrays, and thermal energy storage systems. Jeffrey Rissman, industry program director at Energy Innovation, emphasized the importance of these government-backed projects, stating, "We have an understanding of the technologies that can produce all the goods we rely on without greenhouse gas or other pollutant emissions. But there hasn’t been a demand to commercialize many of these technologies and adapt them into products that industries can simply install." He highlighted that these first-of-a-kind projects can "prove these technologies work" and "drive down the cost of industrial machinery so it can be rolled out more widely." The funding will impact both the suppliers and consumers of clean industrial power. On the supply side, Skyven Technologies will receive up to $145 million to expand the market for its industrial heat-pump systems in the U.S., which are designed to electrify steam production more cost-effectively than natural gas boilers. On the demand side, Kraft Heinz plans to receive up to $170.9 million of the Department of Energy (DOE) grants across 10 facilities, integrating a mix of these innovative heating solutions. The focus for Kraft is on enhancing the efficiency and sustainability of heating processes crucial to manufacturing a range of products, from chemicals to consumer foods like macaroni and cheese.
Research:
Taking AI to the Next Level in Manufacturing [MIT Tech Review]
The survey included 300 manufacturers that have begun working with AI. Some key findings include:
Most of these (64%) are currently researching or experimenting with AI. Some 35% have begun to put AI use cases into production. Many executives who responded to the survey indicate they intend to boost AI spending significantly during the next two years.
The chief constraints are shortages of specialist skills, cited by 49% of executives in engineering and design and 47% in factory operations. In both areas, companies that are deploying use cases feel this crunch more keenly than others.
Manufacturers' most common use cases involve product design, conversational AI, and content creation. Knowledge management and quality control are the most frequently cited at the pilot stage. In engineering and design, manufacturers chiefly seek AI gains in speed, efficiency, reduced failures, and security.
Respondents clearly state that AI use-case development is hampered by inadequate data quality (57%), weak data integration (54%), and weak governance (47%). This number is higher in the chemical industry at 75% with almost as many (54%) citing the need to improve data integration.
Chart of the Week:
In March, U.S. manufacturing activity expanded for the first time in 16 months, with ISM’s Manufacturing PMI at 50.30, up from 47.80 in February and 46.30 a year ago.
Podcasts:
Business Breakdowns - Intel
Manufacturing Deals🏭💵
Collaborative Robotics- A company leading the development of practical collaborative robots (cobots) in sectors including manufacturing, healthcare, & retail/e-commerce.
$100 million [Series B] - Led by General Catalyst and joined by Bison Ventures, Industry Ventures, Lux Capital, and insiders Sequoia Capital, Khosla Ventures, Neo, and MVP Ventures
Varda - A company helping manufacture unique products in microgravity and then bringing them back to earth.
$90 million [Series B] - Led by Caffeinated Capital and joined by Lux Capital, General Catalyst, Founders Fund, and Khosla Ventures.
Oden - A provider of AI-driven solutions for manufacturers, transforming complex data points into action for the front line every day.
$28.5 million [Series B] - Led by Nordstjernan Growth and joined by Flat Capital, Recurring Capital Partners, Atomico and EQT Ventures
Basemark - A Finnish maker of AR and computer vision tools for automakers.
€22 million [Series B] - Led by ETF Partners and joined by Tesi, Constructor Capital, Business Finland and the European Innovation Council
Archetype AI - A company building a foundation model to help understand the physical world
$13 million [Seed] - Led by Venrock and joined by Amazon Industrial Innovation Fund, Hitachi Ventures, Buckley Ventures, Plug and Play Ventures
Planned Downtime 🏭🧑🔧
Conan on Hot Ones