Integrating Intelligence into Manufacturing

Manufacturing is at the core of any discussion about the future of the United States’ leadership — which is increasingly synonymous with having the technological advantage. With artificial intelligence and automation now recognized as the key unlocks to transform virtually every industry, we need a domestic manufacturing system that can build the critical components powering the AI revolution at scale. The downstream impact is immense — AI and automation are enabling transformative progress in everything from building the most advanced aerospace and defense technology, to reinventing transportation and energy, to digitizing supply chain logistics, and numerous aspects of healthcare. Creating category-defining software and robotics for these industries depends on our domestic capacity to quickly manufacture CPU- and GPU-based servers and other complex electronics that make up our AI infrastructure. However, with swift policy changes and tariffs underway amid rising geopolitical tensions, the reality of our domestic manufacturing shortcomings have come into sharp focus. Having outsourced a continuously larger portion of our manufacturing needs to foreign countries over the past five decades, our domestic capacity is well below what’s needed to wholly produce advanced technology here. Not only is the machinery insufficient and outdated, the U.S. is grappling with historic labor shortages at a time when many electronics require at least some assembly by hand due to unique configurations and product-specific components. The only way to reclaim domestic manufacturing leadership and establish AI dominance is by reinventing the industry. The next generation of manufacturing must be software-driven, adaptable to a wide range of use cases, and leverage the very technologies it brings to life: robotics, automation, computer vision, and AI.

Bright Machines is emerging as a leader of this effort. Since its founding in 2018, the company has been on a mission to transform manufacturing with a software-driven, full-stack automation solution to provide design recommendation and to process, assemble, inspect, and disassemble complex electronics.

The company’s Bright Robotic Cells collectively work as “Microfactories” to perform a diverse range of intricate tasks in the assembly and disassembly of AI components, such as CPU- and GPU-based servers, data storage equipment, and more, while its connected software Brightware Platform OS captures production data to provide visibility throughout the entire process. All of this has made Bright Machines a valuable partner to some of the biggest players in the AI ecosystem, including Nvidia and Microsoft. This has established the company as an enabler of software-driven manufacturing for any organization that needs to assemble or disassemble complex AI infrastructure components.

“AI has been around for a relatively long time, but the techniques and applications for how it is used are just being developed,” says Bright Machines CEO Chris Stori. “Our goal is to be the force multiplier for that development process.” Speed, precision, and brute strength are all part of the value — Bright Machines’ Bright Robotic Cells (BRCs) can assemble 50 servers per hour with up to 98% first pass yield, compared to 90% yield with standard assembly. For GPU integrations, Bright Machines’ automated solution can perform the task with zero scrap operations versus the standard assembly initial first pass yield of about 60% to 80%. “GPU clusters are the engine of AI, so if something fails on just one of them — coolant leaking, overheating, etc. — everything must stop to repair the issue. Our solution is to just integrate GPUs correctly the first time,” says Bright Machines Senior Vice President Sean Murray. This includes relatively straightforward, but challenging tasks that must be performed perfectly in order for the server processor to function, like fastening multiple screws at the exact right angle, and lifting heavy GPUs onto a platform. “We’re not trying to replace humans, but help them with tasks that are just hard for humans to do — like lifting a GPU cluster that weighs 280 pounds and costs as much as a Ferrari,” says Murray, describing the aptly named ‘cobot’(collaborative robot), which assists in GPU integrations.

The Idea Phase Bright Machines is also getting involved at the stage before electronics are even assembled: the idea phase. In partnership with Nvidia and Microsoft Azure, Bright Machines recently launchedBright Designer, a web-based application that allows design engineers to simulate automated assembly of new CPU-and GPU-based server designs within the Bright Machines Microfactories. The application is built using Nvidia Omniverse, a physics-enabled simulation platform and data exchange that can integrate across different CAD and PLM platforms, and hosted on Azure Cloud. “We are a true picks and shovels company for the AI backbone,” says Bright Machines VP of Product Hari Ramachandran. “But we’re also thought partners to the organizations we work with. This completely changes how you can think about building automation and AI into your products.” This takes Bright Machines’ capabilities a significant step further. With Bright Designer, users can optimize their CAD or new product designs for automation that would otherwise require extensive (and expensive) prototyping, and which could still fail when actually in production. Bright Designer identifies any potential issues before they become problematic in the later stages, which significantly shortens the design process. This means Bright Machines’ network of partners — which includes chipmakers, OEMs, ODMs, and contract manufacturers — will ultimately be able to get viable products to market much faster. “This is the number one metric for everyone we work with,” says Ramachandran. From Drills to GPU Clusters Bright Machines has hit its stride since sharpening its focus on AI hardware three years ago. But it was the company’s initial mission to transform the manufacturing process for any type of electronics — from power tools to servers — that illuminated the ways in which it could provide the most value to partners throughout the entire ecosystem. “We used to try to work with everyone — power tools like drills or products like smoke detectors. Then we started moving into higher value, more complex electronics, and along the way we learned what was important to each of those verticals,” says Chief Strategy Officer Sviat Dulianinov. “Then, of course, there was the ChatGPT moment.” The explosion of AI in 2022 spurred Bright Machines’ focus on data centers, which involve frequent assembling and disassembling of servers, and the replacing and recycling of decommissioned products. The company continued working with many Contract Manufacturers, Original Equipment Manufacturers, and Original Design Manufacturers including multinational manufacturer Jabil, and launched their application, which provides virtual design recommendations to shorten products’ time to market and ensure quality control and traceability. As Bright Machines dove deeper into AI infrastructure and with , the company became much more entrenched in the network of partnerships between those OEMs, ODMs, and CMs — and much more involved in the end-of-life process for servers and data centers. Today, Bright Machines Microfactories are deployed in circular manufacturing centers to intelligently sort, separate, and remove components for reuse and recycling. “We’re more than manufacturing. We’re part of the entire life cycle of the product,” says Dulianinov.

The Next Generation of Manufacturing This holistic data loop throughout the life cycle of a product effectively creates a cohesive manufacturing value chain, of which Bright Machines is at the center. Through its network of relationships and visibility across the entire manufacturing ecosystem, Bright Machines is transforming not only the technical aspects of the process, but the very dynamics of manufacturing workflows and partnerships. By providing specialized manufacturing and design services to customers who would otherwise rely heavily on manual labor (from a dwindling supply of skilled talent), or have to build bespoke, automated, software-driven manufacturing in-house (as previous generational companies like Tesla did), Bright Machines has the potential to dramatically change the nature of the full manufacturing life cycle.The company’s closed environment data loop allows for continuous improvements, from design through to automation assembly and disassembly, while the customization abilities of the Microfactories and software allows Bright Machines to partner with companies developing a wide range of AI infrastructure products. As Murray puts it, the company provides a “neutral throughpoint enabler” of intelligent automation, significantly contributing towards building the nation’s AI backbone. As supply chain vulnerabilities and labor shortages bump up against the push to rebuild domestic manufacturing and the quest for AI dominance, Eclipse believes Bright Machines has the potential to be the vertically integrated CM that will enable physical AI on a broad scale. Plus, as Dulianinov puts it simply, “AI infrastructure is too important to be offshored.” Follow Bright Machines on LinkedIn or subscribe on Bright Machines website to get insights on manufacturing transformation to your mailbox. Follow Eclipse on LinkedIn or sign up for Eclipse’s Newsletter for the latest on building the New Economy.