Breaking the Bottleneck | Issue 31
[11/14/2023] NVIDIA Chips, Factory Acceptance, McKinsey's Resilience Survey, Semiconductors in LEO, & The Beatles
Breaking the Bottleneck is a weekly manufacturing technology newsletter with perspectives, interviews, news, funding announcements, and a startup database. For an older version of a discrete and continuous manufacturing market map click the link here!
Content I Enjoyed This Week 🏭🗞️🔬
News:
Nvidia Develops AI Chips for China While Avoiding US Restrictions [WSJ]
Nvidia is developing a new lineup of artificial intelligence (AI) chips customized for the Chinese market, aiming to comply with shifting U.S. regulations while maintaining access to this significant market. U.S. restrictions had previously cut off Nvidia's chips for the Chinese market, prompting the company to develop chips that meet U.S. rules. These new chips are expected to be available in China by the end of this year and are designed to fall below performance and data-transfer thresholds that require an export license. This development comes amid increasing U.S. concerns about the use of AI chips in cyber warfare, espionage, and military development in China. Nvidia's new chips aim to address over $5 billion in orders that were put at risk by the recent restrictions and are being designed to serve the Chinese market while remaining compliant with U.S. restrictions.
(From SemiAnalysis: One of the NVIDIA China-specific GPUs is over 20% faster than the H100 in LLM inference, and is more similar to the new GPU that Nvidia is launching early next year than to the H100.)
The Next Embedded Frontier [EDN]
Infineon has introduced a new microcontroller, PSoC Edge, designed to provide hardware-assisted machine learning (ML) acceleration for the Internet of Things (IoT) and industrial applications such as smart homes, security surveillance, wearables, and robotics. PSoC Edge offers significant improvements in compute performance for neural network applications, allowing advanced ML tasks to be performed locally. The microcontroller is equipped with a high-performance Arm Cortex-M55 processor along with Arm's Helium technology for enhanced digital signal processing (DSP) and ML capabilities to accelerate neural network processing. Infineon has integrated Imagimob's ML tool suite into its Modus Toolbox software platform to provide developers with essential tools and software support for their ML-enabled applications. This ecosystem of software and tools enables end-to-end ML development, from data entry to model deployment, making PSoC Edge a promising solution for various ML-based IoT and industrial applications.
Factory Site Acceptance Testing and Debugging [Assembly]
Debug, checkout, and startup phases are critical in the automation and assembly machine development process. Debugging involves identifying and rectifying malfunctioning elements, checkout ensures proper machine functionality testing, and startup initiates the production process. Machine builders and systems integrators aim to streamline the debug process with structured methods, documentation, and communication however, the debug process can be stressful due to tight schedules and unexpected challenges. Pressure to deliver machinery quickly is particularly high in industries like lithium-ion batteries and electric vehicle manufacturing. The factory acceptance test (FAT) follows debugging and helps ensure that the automation system complies with contractual specifications before shipping to the customer. Thorough testing and documentation before delivery can prevent post-deployment problems. After FAT, the equipment is disassembled for shipment, and a site acceptance test (SAT) is conducted at the customer's location to confirm that the equipment functions as expected.
Industrial Automation Software Management on AWS [AWS Blog]
This blog post from AWS discusses the challenges and best practices for programming and maintaining automation code in factories and warehouses. In traditional automation code development and maintenance, complex tasks, like "depalletizing" incoming goods in a warehouse, involve a mix of programmable logic controllers (PLCs), robot programming, and control logic for various equipment. The code and its “recipes” must be continuously maintained to optimize automation operations. However, dealing with multiple vendor systems and code collaborators can make implementing change control and managing the code's lifecycle challenging. To address these issues, Amazon Web Services (AWS) offers modern DevOps, Digital Twin, and Virtual Commissioning technologies to develop, test, commission, and maintain automation code alongside Copia Automation’s product. Their Git-based source control backs up PLC and robot code to the AWS cloud and allows engineers to visualize code from various vendors and programming languages, offering tools for logic comparison, code merging, code reviews, and code commenting. The benefits of using AWS for automation code management include reliability, remote accessibility for experts, reduced downtime, systematic code reviews, scalability, flexibility, improved security, and cost savings.
Lavish Tax Credits and Trade Protections Lure Solar Firms to U.S. [NY Times]
The U.S. solar industry is experiencing a significant revival, with companies announcing nearly $8 billion in new investments in solar factories within the year since the passage of the IRA. This legislation includes a combination of substantial tax credits for domestic solar manufacturing and tighter restrictions on foreign products. One of the most notable examples of this resurgence is Suniva, a solar panel manufacturer that had previously declared bankruptcy. Six years ago, an executive from Suniva had warned of fierce competition from companies in China and Southeast Asia, leading to numerous U.S.-based solar companies shutting down. In response to Suniva's concerns and the broader challenges facing the U.S. solar industry, the Trump administration imposed tariffs on foreign-made solar panels in 2018. The situation has changed dramatically in recent times. Suniva recently announced plans to reopen a plant in Georgia, citing the positive impact of tariffs, protective regulations, and generous tax incentives provided by the Inflation Reduction Act. Furthermore, companies like REC Silicon in Washington, which had previously shuttered their operations, and, Maxeon, a Singapore-based solar producer, will begin work on new facilities next year.
Research:
McKinsey’s Industrial Global Resilience Survey [McKinsey]
Evaluating Labor Demand In The Transition To EVs [CMU]
The ongoing shift from traditional internal combustion engine vehicles (ICEVs) to electric vehicles (EVs) has raised questions about whether this transition will be economically as well as environmentally sustainable. In particular, one concern is the impact on manufacturing labor. The paper collected detailed data on the production process steps required to build key ICEV and battery electric vehicle (BEV) powertrain components and the labor required for each process step. The data included information for 252 process steps, which was collected from the shop floors of leading automotive manufacturers and combined with information on a further 78 process steps found in existing literature. After research, the paper found that, in all scenarios, the labor intensity required for the manufacturing of BEV powertrain components is larger than for ICEV powertrain components. Furthermore, the results imply that vehicle electrification may lead to more jobs in powertrain manufacturing, at least in the short- to medium-term.
Semiconductor Manufacturing in Low Earth Orbit for Terrestrial Use [NASA]
Over the past three decades, the deteriorating discovery and growth of crystalline materials (DGCM) capacity in the United States has significantly stalled the domestic semiconductor industry. In-space manufacturing offers technological innovation, advancements, and discoveries unbound by Earth’s gravitational forces. However, more work is required to obtain more data on the applications of promising semiconductor R&D in LEO. This would include the whole range of experiments in the semiconductor production phases such as crystal growth, wafer processing, epitaxial growth, circuit patterning, etc. The benefits of semiconductor manufacturing in LEO are clear. Earth’s gravitational forces pose substantial barriers to quick, high-yield semiconductor production. Beyond the scientific benefits of microgravity, there are substantial practical benefits to incorporating LEO-based manufacturing into the supply chain. Transitioning this industry into space is the only path forward if the United States is to keep pace with the technological arms race unfolding across the globe.
Podcasts:
Operational Excellence in Manufacturing [The Manufacturing Executive]
Chart of the Week:
Manufacturing Deals
Niron Magnetics - A a startup producing rare earth-free magnets used in computers, appliances, automobiles, and other machines
$33 million [Venture Round] - Led by GM Ventures, Stellantis Ventures and Others
Sift - A startup offerting an end-to-end telemetry stack that makes “sifting” through machine data simpler
$7.5 million [Seed] - From Riot Ventures, Fika Ventures, First Resonance, Datum and Duro
Weekly Planned Downtime
The Beatles New Release!