Proceedings of a Workshop—in Brief

Convened March 4–5, 2025

Mid-Scale Manufacturing and Characterization Capacity for Department of Defense Critical Materials Supply Challenges, Part 2

On March 4–5, 2025, the National Academies of Sciences, Engineering, and Medicine's Defense Materials, Manufacturing, and Its Infrastructure Standing Committee hosted the second part of a two-part workshop sponsored by the Department of Defense (DoD). This workshop¹ was aimed to examine U.S. manufacturing and characterization capacity for mid-scale production of sufficient quantities of emerging materials to assess the engineering utility of those materials (i.e., more than the small scale produced in a university laboratory and less than is required for serial production). Strategic and critical materials are vital to national defense and economic prosperity, enabling the United States to develop and sustain emerging technologies and improve its warfighting capability. DoD defines strategic and critical materials as those that support military and essential civilian industry and are not found or produced in the United States in quantities to meet its needs. Hence, accelerating, scaling up, and transitioning technologies to produce or replace critical materials, some novel, are essential to mitigating DoD supply challenges. Mid-scale manufacturing and characterization capacity in combination with modeling and simulation is expected to play a key role in this effort. Key workshop topics included the availability, access, and economic sustainability of both mid-scale manufacturing facilities and experimental facilities that provide extreme environment characterization, with particular attention toward defense-specific applications. The workshop events also explored state-of-the-art approaches used to evaluate data, models, and simulations for scale up. The first part of the workshop² explored mid-scale issues in the fields of semiconductors, pharma, energetics, critical materials, and chemistry and touched on intellectual property (IP) and innovation issues. Participants' presentations and discussions from the second part of the workshop, which focused on specific technology challenges, are summarized below.

ANALYSIS OF PUBLIC–PRIVATE PARTNERSHIPS IN THE CONTEXT OF MID-SCALE MANUFACTURING CHALLENGES

Reviewing critical definitions provided during the first part of the workshop, Angus Kingon, Brown University, and Mark Johnson, Clemson University, described

mid-scale manufacturing as a key step in the commercial development process, focusing on manufacturing readiness levels³ (MRLs) 7–8 and technology readiness levels⁴ (TRLs) 6–9. They reiterated that the decision to commercialize is usually driven by economic opportunity. However, barriers exist to commercial viability, and understanding economic opportunity begins by assessing customer needs. Another driver for commercialization is national strategic need, which might justify increased costs (either subsidized development costs or increased product price). They noted that considering both drivers for commercialization and balancing economic and strategic opportunity are required for a dual-use approach (i.e., both civilian and military purposes).

Kingon and Johnson said that promising innovations (i.e., TRLs 1–3) often are not commercialized in the United States for DoD applications, frequently failing at the point of scaling. The innovator (e.g., researcher), connector (e.g., program manager), and enterprise (e.g., investor) all play important roles in the process, and the decision to commercialize the material, component, product, or system hinges on the time-dependent projected profit being appropriately larger than the projected cost. Although funding is readily available for TRLs 1–3, enterprise funding is often needed for higher TRLs and MRLs, and a business case is generally required for justification.

Kingon and Johnson asserted that public–private partnerships (PPPs), defined generally as "cooperation between public and private actors [to] jointly develop products and services and share risks, costs, and resources that connect these products and services," can help bridge this "valley of death" that can occur when scaling up manufacturing. Kingon noted that PPPs use an ecosystem approach, consider the entire supply chain, and include small and medium-sized enterprises, which are either innovative (i.e., sources of research) or market-focused (i.e., sources of solutions). When a new problem arises, Johnson proposed using a decision loop (observe, orient, execute, and decide) to evaluate if a particular PPP is appropriate. He suggested listing three PPPs with the closest capabilities to a specific need and then identifying the relevancy of those PPPs based on capabilities, technical know-how, and workforce skillsets. Next, he said to list necessary secondary capabilities, and, in conversation with both the PPP and the sponsoring agency, to identify supportive mechanisms, capacity to receive third-party funds, sustainability of the business model, and supportive processes (e.g., competitive broad agency announcement, other transaction authority [OTA], military interdepartmental purchase request, non-government commercial ecosystem). At that point, four possibilities will emerge: the PPP (1) is a good fit; (2) will require significant fixed investment to be a good fit; (3) is a poor but adjacent fit; or (4) is neither a good nor adjacent fit and a new PPP needs to be established.

Kingon and Johnson provided several examples of PPPs for dual-use applications. In particular, the National Science Foundation's (NSF's) Directorate for Technology, Innovation and Partnerships⁵ fosters innovation and technology ecosystems, establishes translation pathways (i.e., processes and mechanisms that help move scientific research and discoveries from the laboratory into real-world applications), and partners to engage the nation's talent. The Department of Energy (DOE) and the national laboratories offer project-based approaches, partnerships, and access to specialized facilities. Furthermore, the Defense Advanced Research Projects Agency (DARPA) extends beyond research contracts (e.g., UltraDense Capacitor Materials Processing Partnership); NY CREATES⁶ (New York Center for Research, Economic Advancement, Technology, Engineering and Science) focuses on bridging the valley of death with technology, partnerships, and shared facilities; and the Manufacturing USA institutes⁷ build powerful networks to advance TRLs. (The example of NextFlex⁸ was discussed at the first part of the workshop on January 29.) The Economic Development Administration's Tech Hubs⁹ are complementary to

Manufacturing USA in that they focus on technology, manufacturing scale-up facilities, and ecosystem and manpower development. For example, the American Aerospace Materials Manufacturing Center,¹⁰ which received $50 million in funding and has 50 members, serves as a testbed and training center to develop a class of next-generation lightweight composite parts in preparation for high-rate manufacturing techniques. Methods will be advanced through TRLs 6–9 for defense and commercial applications. Additionally, the Manufacturing Extension Partnership¹¹ is a national network that leverages federal funds, state investments, and private-sector fees. This market-driven program, with 460 service centers and 1,440 advisors and experts, creates high value for manufacturers, leverages partners to maximize offerings, and transfers technology and expertise to small and medium-sized manufacturers.

Kingon turned to a discussion of a literature review he conducted in economics, sociology, public policy, psychology, and neuroscience for new perspectives on the mid-scale manufacturing problem and the value of partnerships. He found that social psychology literature on entrepreneurial cognition, combined with social network theory from public policy literature, provide key insights. The entrepreneurial journey, he explained, is the process of moving from ideas to validating an opportunity (i.e., evidence that profits will be larger than costs) to establishing, funding, and growing a business. Key cognitive perspectives on this journey include the following: (1) embedded entrepreneurial cognition, which is the action-oriented development of mental constructs that are strongly influenced by the entrepreneur's natural and social environment; (2) grounded cognition, which relates to the tasks the entrepreneur completes (e.g., networking) to realize these needed mental constructs and a business case; and (3) distributed cognition, which recognizes that effective entrepreneurial cognition includes significant contributions from interactions with many stakeholders. He summarized that cognitive processes differ depending on whether the entrepreneur is embedded in the market or in the research community. Overall, he stressed that a cognitive perspective helps to consider how individuals and enterprises within the DoD PPP framework can make connections, develop knowledge, and make decisions to take projects across the valley of death and small businesses to the mid-scale size.

Discussion

A workshop participant asked about the difference between profit/cost ratios and return on investment (ROI). Kingon cautioned against over-relying on ROI when evaluating the practical viability and cost of scaling because ROI is more of a post-analysis tool while profit/cost ratios are better suited for planning. Sharon Belenzon, Duke University, noted that economics literature discusses why the government should pay for research and development (R&D) but not why it should invest in manufacturing infrastructure. He asked about market failures in manufacturing. Johnson replied that without an ROI, no motivation to invest exists without national interest. Belenzon said that government investments could solve a big market problem, even without public interest. Johnson agreed and referenced the Small Business Administration's (SBA's) 504 loan program but urged caution with incentives. Recognizing the government's role, he highlighted DOE's loan program as key for funding nuclear power plants, for example.

TRANSMITTING DATA FROM LOW EARTH ORBITING SATELLITES TO GROUND STATIONS: THE GAP BETWEEN LARGE- AND SMALL-SCALE MANUFACTURING

Jade Wang, Massachusetts Institute of Technology's (MIT's) Lincoln Laboratory, explained that the Lincoln Laboratory, a federally funded research and development center (FFRDC), focuses on satellite communications given increased data generation in low Earth orbit (LEO). Because these data are difficult to retrieve with current technology, the laboratory is considering how optical communications (with higher frequency, shorter wavelengths, and an unregulated/unlimited spectrum) could support higher data rates. Furthermore, higher data rates mean that one can get LEO data down to the ground in shorter timeframes. However, some challenges arise; for example, transmission is not possible in cloudy weather and narrower beams require precise pointing.

Wang indicated that ground-based optical communications technology, run across the world on a spectrum-constrained optical fiber network, can be

adapted for free space optical communications. For example, the TeraByte InfraRed Delivery (TBIRD)¹² mission leveraged advances and investments in fiber telecom equipment terrestrially to take steps to deliver large volumes of data from space to the ground. The mission successfully demonstrated delivery of 200 gigabits per second (Gbps) of data. Key enabling technologies for this success were a high-rate optical modem with a 100 Gbps transceiver; large, high-speed, solid-state drives for storage; and an optical amplifier, all adapted to enable high data rates in a compact package from space to the ground.

Wang elaborated on the challenges of achieving this mission. First, atmospheric turbulence, which causes fluctuations in collected power and coupling loss, can result in lost data. To ensure that the communications architecture would be robust to atmospheric fading, her team created an automatic repeat request code to resend lost packets and frames. Environmental testing was also important to reduce component risk and improve component survival. Once the payload was built up, validation and acceptance testing were critical. Launching on May 25, 2022, TBIRD's payload was so small that it could be carried on the plane. Second, she noted that the team cooperated with the CubeSat bus, jointly developing a precision pointing mechanism and achieving best-in-class body-pointing performance to date at high slew rates, or orientation changing speed.

Wang stated that this technology, demonstrated at TRL 9 for LEO, is now available for government use, and the next step is scaling. Commercial technology advancement is motivated by business drivers, which often require large volumes to generate profits. However, government space satellite communications often demand small-volume, high-performance systems. She stressed that bridging this gap requires technology investment, technology transfer, and innovation to create sufficient markets for commercial engagement. She asserted that this technology could be leveraged to support the Black Hole Explorer mission, downlinking high data rates in medium Earth orbit instead.

Discussion

Gamal Refai-Ahmed, Advanced Micro Devices, Inc., posed a question about steps to pursue industry adoption. Wang said that no one-size-fits-all approach to success exists. Industry knows how to manufacture at scale, and TBIRD has been demonstrated at low scale, so partnership could bridge that gap. She underscored that scaling is a unique field that requires specific expertise. Kingon asked specifically about Lincoln Laboratory's approach to exploring commercial potential. Wang reflected on ongoing conversations, since the mission concluded in September 2024, with commercial entities and partnership with the National Aeronautics and Space Administration (NASA) to consider commercialization opportunities for TBIRD. She specified that timing, based on need, is important. She explained that a dual-use business case to support the internet backbone is premature; the bandwidth needs have not yet grown sufficiently to create a technology pull. Wang reinforced the notion that markets drive the adoption of technology and said, "A great technology is nothing without a business case."

Tom Kurfess, Georgia Institute of Technology, inquired about the origin of commercial off-the-shelf capabilities and how to ensure that they are appropriate for government applications. Wang explained that TBIRD was targeted toward buying at low cost, wrapping, and adapting, and she noted that trusted foundry approaches are critical. She also highlighted a complicated trade-off: pushing to know the details of a supply chain drives toward government needs but away from commercial needs, which differ significantly (e.g., security). She added that if trust at all levels is required, the expense increases and much more cooperation is needed to obtain documentation. However, increasing capabilities development in the United States could simplify that problem. Robert Rudnitsky, National Institute of Standards and Technology, noted that decades could pass between a discovery and large-scale production, but the choices that are made at the start influence later stages. He pointed out that choices to have domestic parts and processes in early-scale work may have consequences downstream as the technology develops.

PANEL ON THE RESPONSE TO THE ANALYSIS OF PUBLIC–PRIVATE PARTNERSHIPS

Rudnitsky conveyed that the Manufacturing USA program¹³ has worked for the past decade to advance manufacturing technology in the areas of electronics, materials, energy

and process efficiency, digital controls and automation, and biomanufacturing by connecting people from industry, academia, government, and national laboratories with ideas and technology. Its four strategic goals include (1) increasing the competitiveness of U.S. manufacturing; (2) facilitating the transition of innovative technologies into scalable, cost-effective, and high-performing domestic manufacturing capabilities; (3) accelerating the development of an advanced manufacturing workforce; and (4) promoting a network of institutes that build long-term support for and from their communities.¹⁴ Each of the 18 Manufacturing USA institutes is a PPP, and each starts with federal funds ($70 million–$120 million) and at least a 1:1 cost match in non-federal resources, according to Rudnitsky. He mentioned that the new Smart Manufacturing Institute has much larger-scale investments, with $285 million in federal funds matched by $1 billion in non-federal funds. Manufacturing USA has nine partner federal agencies that contribute resources, including primary sponsors DoD, DOE, and the Department of Commerce, and 2,900 member organizations, who collaborated on more than 900 applied R&D projects last year. As a case study, he described Manufacturing USA's AIM Photonics, where multi-project wafers allow several designs to be fabricated simultaneously on one wafer, dividing costs between customers, decreasing design times, improving efficiency, and lowering the entry price—all especially important for small and medium-sized manufacturers and for moving mid-scale manufacturing forward.

Stephanie Watts Butler, WattsButler LLC, offered guidance on mid-scale manufacturing based on her experience in the semiconductor industry. First, she highlighted the value of identifying specific customers and applications at the beginning of developing a technology. This commercial view leads to profitable products with long lifetimes and significant revenue. Second, she stressed that work to prevent technologies from entering the valley of death is critical because some technology should be killed during or before this stage. This decision saves money and allows people to pivot quickly to make the right technology. Third, she introduced the frameworks of New Product Development (NPD) and New Product Introduction (NPI). She stressed the importance of thinking in these frameworks where TRL and MRL targets are outputs of given phases; this way of operating helps clarify what needs to be done in each phase of development on the way to complete manufacturing ramp up. She suggested developing in the earliest phase what clarifying questions about the target application, the customer and the market, the size of the market based on the product's features, high-risk technology challenges, cost issues, supply and yield challenges, mission profiles, and manufacturability need to be answered in the later development phases. Fourth, she remarked that the innovator and the enterprise share responsibility for productizing a business process, and a PPP provides the architecture for making these connections. Fifth, she noted that productivity and efficiency of R&D and productization have changed over the past 25 years, and questioned how PPPs have changed as a result, especially with respect to IP and knowledge sharing across the ecosystem. She emphasized that the knowledge that emerges from data must lead to actions and decisions that lead to profit and revenue. Focusing on a reasonable intended application/market/product during R&D will meet the needs just described, even though this intended application/market/product is the one that rarely goes to final ramped manufacturing. She said that cohesive team goals across organizational boundaries are essential to produce the right product, at the right cost, at the right time. She concluded her presentation by reiterating the value of using the NPD and NPI vernacular of commercial off-the-shelf suppliers and not just using DoD language to enable dual-use products.

Mick Maher, Maher & Associates, LLC, agreed that decisions made early in the process affect production and that developers should not continue to pursue all technologies. He stressed that "technology pull" propels technologies forward—whether the customer is DoD or a commercial entity—but highlighted the disconnect between the government's and industry's product development cycles. He observed that technology is typically killed not because of its internal failings but rather because a market does not exist. Therefore, he asserted that evaluating the market in discussions about TRLs and MRLs is critical for mid-scale manufacturing. He mentioned the key role that PPPs play in helping to overcome these challenges and advance technology. Another disconnect for mid-scale manufacturing, he continued, relates to the government's acquisition process for a platform. He explained that in

Milestone A (to determine if a program has met all of its exit requirements of the Materiel Solutions Analysis Phase to proceed into the Technology Maturation and Risk Reduction Phase), before developments have even entered manufacturing, technologies are already being selected as which to keep and which to discard. After Milestone C (to determine if a program has met all of its Exit Criteria of the Engineering and Manufacturing Development Phase to proceed into the Production and Deployment Phase), material and design are locked in and manufacturing focuses on optimizing the process, not on inserting technology. Thus, he suggested that PPPs can have the greatest impact, from a technology perspective, if they engage at Milestone A. He emphasized that consortia among manufacturing institutes, industry, and academia could help advance mid-scale manufacturing and technology hubs and similar ecosystems are critical for workforce development to support production.

Belenzon indicated that over the past 50 years the American innovation system has had a growing division of innovative labor as corporations have withdrawn from scientific research. Today's system is fragmented and work is now distributed across universities, start-ups, large firms, and the government instead of being conducted within a single organization. He said that this division of specialization has worked well in some sectors, such as the field of life sciences, but the field of materials sciences has significant challenges in terms of disconnects between ideas and the market. He noted that people often focus on grants and R&D contracts in terms of government funding, but these represent only a small segment of the influence the government has on corporate innovation; product contracts and service contracts comprise a much larger portion. He stressed that well-funded companies seek government contracts to innovate because of the implicit promise that they will get a downstream procurement contract if they produce the best prototype. But in the past 20 years, the government has decoupled R&D contracts and procurement contracts, which decreases the incentive for R&D companies to solve problems in national defense. In closing, he said that government procurement policy is innovation policy when it provides strong incentives for upstream R&D, where market applications are more limited, but much work remains.

Discussion

Johnson observed that decision making across the valley of death is critical. Although missing a milestone is sometimes considered a failure, the opportunity to learn and pivot can be valuable. Maher said that DoD tends to focus on how changes might impact its annual budget, so people unfortunately often persist as planned instead of pivoting. Instead, he encouraged the use of a gated product development cycle—which includes input from technical, operations, marketing, and program management staff—to enable better decision making and the opportunity to pivot to new pathways. In response to a follow-up question from Johnson about better approaches for program management, Maher noted that although cooperative agreements incorporate the flexibility to pivot, the broader culture that discourages pivoting hinders progress. Reflecting on the problems with this culture, Butler remarked that pivoting also unfortunately could lead to a poor performance review. Instead she urged mature organizational leadership that rewards instead of punishes people for efforts to improve technology results by pivoting away from original project plans.

Reflecting on strategies to better leverage the Manufacturing USA institutes, Johnson wondered if a 1:1 matching with additional private-sector money available for other projects would be effective. Rudnitsky said that the private sector has near-term incentives to invest in later stages of technology and manufacturing readiness, and he suggested contract research as a possible way forward. Johnson asked if Belenzon studied differences in contract mechanisms; he replied that the more important issue to study is who will invest in translational research, because high uncertainty discourages investments downstream. He added that people only invest in manufacturing when a technology can be scaled. In the life sciences, for example, early investors protect bargaining positions as the technology progresses and uncertainty is reduced. In addition to this unique sequencing of investments, he continued, life sciences emphasizes IP rights and multiple pathways to commercialize.

Refai-Ahmed observed that most semiconductor development occurs in Taiwan, and he posed a question about whether public funds could be used to increase defense manufacturing in the United States, perhaps

that faster progress is essential, starting with a deep understanding of a problem to generate actionable data early in the process. For example, a hypersonic window material needs low emissivity for light to transmit. After failing to achieve its target initially, APL eliminated several potential candidate materials that were not key to achieve the main properties and focused only on those that could address the core problem. The next step was to conduct custom tests under the right conditions. APL often combines multiple data streams and creates multiple test methods to better understand the potential performance (e.g., thermal and spectral) of a material in a particular environment before conducting operationally relevant ground tests.

Hunt explained that the final step, transition, is driven by data. APL leverages automation and proxy tests to identify trends and accelerate testing and material qualification. It creates novel methodologies to shorten "long-pole" tests (e.g., developing high-throughput fatigue test methods), which has implications for both qualification and materials design. APL also leverages modeling for rapid qualification, using digital twins to define process control windows, predict local performance, optimize experimental budgets, do uncertainty quantification, and add flexibility in the manufacturing process. In closing, he asserted, "None of us is as smart as all of us"; synergistic and collaborative approaches among academia, industry, government and national laboratories, and university-affiliated research centers (UARCs) and FFRDCs yield the best solutions to advance revolutionary technologies.

Discussion

Tony Rollett, Carnegie Mellon University (CMU), posed a question about focusing on performance enhancement instead of on requirements when designing materials. Hunt said that from an academic perspective, focusing on performance enhancement is the right approach. APL, however, is in a unique position as a UARC, and its R&D has specific mission intent with real-world requirements.

Dan Miracle, Air Force Research Laboratory (AFRL), inquired about major barriers to accelerating innovation. Hunt replied that things do not always fail in the ways anticipated, and faster high-throughput testing would be valuable. Miracle observed progress in the combinatorial field for rapid parallelized synthesis, and he urged increased study of the ability to control and manipulate microstructure. Leila Ladani, Arizona State University, asked about better leveraging the Manufacturing USA institutes and pursuing convergent research to solve problems. Hunt championed collaborative approaches, which also help to avoid duplicating efforts.

CASE STUDY 2: LESSONS LEARNED FROM THE SEMICONDUCTOR SUPPLY CHAIN AND PLASTICS RECYCLING TECHNOLOGY

Robert Allen, National Renewable Energy Laboratory (NREL), discussed two polymer-based materials—photoresists and plastics—and related technologies. Focusing first on photoresists, until 2000, he continued, all semiconductors were made with a phenolic polymer. However, because phenols are optically opaque at certain wavelengths, new polymer systems were needed. Eventually, scientists discovered that the gap between the last lens element and the photoresist-coated silicon wafer needed to be immersed in water. All semiconductors were then made in part under water, which provided a 45 percent resolving power boost. However, because water is a solvent, several problems arose, and a material that enabled water immersion lithography was needed. Allen's team designed a phenolic capability into a non-phenolic system using a polymer backbone—methacrylate polymers—an innovation that moved from inception to production in less than 3 years. He explained that concepts become products much more quickly in the semiconductor industry than in other industries, owing to a well-developed materials supplier network and collaboration among chip companies in early R&D; early investments and roadmapping, including interactions with universities and companies; intense competition; the ease of assessing performance; and heightened market pull.

Focusing next on plastics, renewable energy, and recycling technologies, Allen described the difficulty of moving from the laboratory to piloting. NREL's Energy Materials and Processing at Scale facility, due to be completed by 2027, could help address this issue. He noted specific challenges for plastics circularity: nearly 1 trillion pounds of polymers are manufactured annually. Polymers are ubiquitous because they are easily tailored for performance in specific applications, macromolecules of dissimilar

structure separate on the molecular scale, and economics and the ability to scale are critical. Potential solutions include mechanical recycling, selective dissolution and purification, and deconstruction and reconstruction or upcycling. He also presented challenges for polymer circularity: polymers are diverse, mechanical recycling is unforgiving, advanced recycling is challenging, multi-materials and composites are complex, and supply chains and economics are critical. He emphasized that many polyethylene terephthalate (PET) containers are not mechanically recyclable and proposed that mechanical recycling and chemical recycling could work together, with the help of purification and separation technology innovations.

Allen remarked that most people focus on deconstruction chemistry for advanced polymer recycling or upcycling; however, equally, if not more important, are the input (cost, volume availability, purity, impurities) and the output. In essence, much attention is given to the chemistry that breaks down polymers, but the viability and impact of advanced recycling depend just as much—or more—on the quality and characteristics of both the starting materials and the end products. The Bio-Optimized Technologies to keep Thermoplastics out of Landfills and the Environment (BOTTLE) consortium¹⁵ explores these aspects in greater depth and develops new recycling technologies, with more than 100 scientists and engineers from national laboratories and universities working in deconstruction technology, creating building blocks, and redesigning new materials. Headquartered at NREL, the BOTTLE consortium helps companies with existing problems (e.g., circularizable textiles for Patagonia). He also discussed the value of joint ventures among big brands, engineering companies, and technology developers (e.g., Under Armour, Technip Energy, and IBM scaled and commercialized VolCat, a recycling process developed by IBM).

Discussion

David Furrer, Pratt & Whitney, asked about deconstruction of per- and polyfluoroalkyl substances (PFAS). Allen responded that PFAS are persistent, and finding replacements, developing degradation mechanisms, and collecting sources are key challenges.

Furrer also posed a question about the value of technology roadmapping. Allen noted that the Semiconductor Industry Association's roadmaps contain much knowledge and are precompetitive (3–9 years). With a renewed focus on basic transistor technology and with advanced lithography and packaging technologies, he said the roadmapping process has become more complicated, spurring R&D on the path to consensus.

Johnson wondered about the value of a PPP versus a partnership between private entities. Allen shared that in addition to securing funding, IBM's joint venture enabled access to a PET supply chain, which was critical to success. He emphasized that whether it is public–private or private–private, collaboration is key to de-risking technology for potential scaling. To bridge the gap between R&D and piloting, he advocated for a Sematech/Interuniversity Microelectronics Centre–type facility that start-ups could plug into to de-risk technology, do professional technoeconomics, and provide data to large companies and venture capitalists to secure funding.

CASE STUDY 3: A CONVERSATION ON LESSONS LEARNED FROM CLINKENBEARD'S SUCCESSFUL RISE

James Conley, Northwestern University, referenced the Defense Innovation Board's January 2025 Executive Summary, which highlights DoD's innovation pathway from prototype to production—with universities at one end and contractors, like Clinkenbeard,¹⁶ at the other end, and a complex web of intermediaries focused on experimentation.

Conley asked Reg Gustafson, Clinkenbeard, about the history of its product development and production for defense primes. Gustafson said that Clinkenbeard has worked with Collins Aerospace and Woodward for decades and now works with several additional defense contractors. He stressed that Clinkenbeard does not work directly with the government. He added that some defense contracts are leaner than others; for example, it took Clinkenbeard 3 months to secure a purchase order from a defense contractor to make about 30 parts for a vehicle, whereas non-defense contractors make such decisions within days of receiving a quote. Conley inquired as to why Clinkenbeard does not deal directly with DoD, and Gustafson explained that Clinkenbeard has not been able to win DoD's business

directly. Furthermore, although the company is on the government buyers' list, it cannot access engineering data to respond to requests for quotes. Conley pointed out that this level of friction prohibits dual-use.

Conley posed a question about Clinkenbeard's history of product development and production for Blue Origin's civilian space rocket program. Gustafson stated that Clinkenbeard makes parts and assemblies for vehicles and engines used for space flights, as well as training devices and systems. He added that processes to make parts, given coating requirements, may take 3 weeks each. Combining that with the few weeks it takes to make hardware, then delivery of flight hardware to Blue Origin usually takes 10–20 weeks. Given this experience with Blue Origin, Conley asked about an ideal system within DoD that could allow Clinkenbeard to engage. Gustafson suggested that the government act more like an original equipment manufacturer that finds capable suppliers, onboards them, and issues purchase orders. This approach would use neither OTAs or Small Business Innovation Research (SBIR) programs nor complex acronyms or procedures. He said that with this approach the government could procure hardware much more quickly. Conley also wondered how government contracts could become more streamlined. Gustafson emphasized that the contract itself is a barrier to entry that does not exist outside of the government. Even though Clinkenbeard is attracted to government work, such contracts are prohibitive for small businesses that cannot dedicate considerable time for a complicated contracting process.

Discussion

Karen Thole, University of Michigan, posed a question about the certification required for parts and if the associated cost is a barrier. Gustafson said that Clinkenbeard has spent a significant amount of time and money ensuring that it has the right equipment to meet the certification requirements of defense contractors. Jian Cao, Northwestern University, asked if Clinkenbeard supplies process data from the manufacturing or inspection data from final products. Gustafson responded that this decision depends on the customer, volume, and contractor requirements. He stressed that because Clinkenbeard already supplies data to non-government customers that requirement would not be a barrier to working with the government. Rollett asked if Clinkenbeard has had a request for unavailable material. Gustafson acknowledged that this has happened both for government and non-government projects, especially for raw materials (e.g., European-grade aluminum). If Clinkenbeard's suppliers cannot find a material, the customer is asked to supply it.

Kingon commended Clinkenbeard's capabilities and wondered how the company maintains the state of the art. Gustafson replied that Clinkenbeard staff do research, read trade publications, and communicate with software representatives. Cao asked if Clinkenbeard is a member of any consortia, and Gustafson noted that it recently completed strategic planning with the Illinois Manufacturing Excellence Center. Refai-Ahmed posed a question about Clinkenbeard's added value, especially in terms of competing with China. Gustafson responded that the key value add is that Clinkenbeard handles everything based on requirements in a customer's drawing. Although the pressure to compete is real, he said that Clinkenbeard cannot compete with customers who have facilities overseas.

CASE STUDY 4: LESSONS LEARNED FROM ACCELERATING TECHNOLOGY TRANSITIONS

Rama Venkatasubramanian, APL, indicated that thermoelectric technologies could be a $100 billion worldwide market when fully commercialized, with significant implications of new capabilities across a multitude of DoD platforms. He explained that solid-state thermoelectric devices use electrons in semiconductors to cool regions of interest. These same materials can be used in devices that can also turn heat into electricity, generating power to enable a wide range of missions. These devices are scalable and compact, can be miniaturized, and are produced using semiconductor tools, similar to today's worldwide use of gallium arsenide–based solar cells for satellite power for LED lighting. APL has moved this one-platform technology from basic research to applied research to technology transition throughout the past decade. For example, Venkatasubramanian said that in 2014, APL implemented metal-organic chemical vapor deposition equipment and molecular beam epitaxy to grow thin nano-engineered thermoelectric films. In 2017, APL created portable power units and soon after invested in device fabrication. In 2022, APL invested in spark plasma

synthesis to create a one-step manufacturing process for NASA missions with devices now running continuously at 1000°C for 3 years. He stressed that appropriate facilities and material validation are critical to enable higher TRL demonstrations, as are government funding and multi-organizational collaboration.

Venkatasubramanian noted that APL's Technology Transfer Office works with its scientists and inventors as part of APL's UARC mandate in transferring knowledge and technology innovations to the industry responsibly and with speed, agility, and transparency. Technology partners such as Meta and Samsung have strategically supported ongoing R&D in thin-film nano-engineered thermoelectric devices. Sharing specific examples of dual-use technology transitions, he first presented a transition of DARPA-funded research to the U.S. government. In this case, APL worked to enhance combined thermoelectric–photovoltaic converter efficiency, evaluate distributed power for satellite health monitoring sensors, use on-board fuel sources for on-demand power, and compact cooling to enable focal plane array imaging—all enabling capabilities for DoD space situational awareness. Second, Venkatasubramanian discussed the transition of DARPA-funded research to industry. APL partnered with Meta, demonstrating thin-film thermoelectric capabilities in cooling and energy harvesting for wearables. This built on his team's work on a U.S. Army–funded effort to help amputees with prosthetics to perceive the thermal environment for better quality of life. Third, noting that radioisotope power sources APL is building for NASA could enhance the exploration of outer-planets and beyond the solar system, he described it is working collaboratively to transition the NASA-funded research capability to the U.S. industry. APL has demonstrated that silicon germanium (SiGe) unicouples, made with the latest cost-effective and advanced material synthesis technique called spark plasma sintering, could be used to convert nuclear heat at around 1000°C to electric power for sustained missions and to set the stage for potential transitions to Aerojet Rocketdyne. Fourth, he provided an example of how a transition of NASA-funded research could help DoD: after testing and validation, APL is exploring how SiGe devices could also enable new capabilities in hypersonic systems. This emphasizes the importance of manufacturing and research synergy between DoD and other U.S. government agencies.

Venkatasubramanian underscored that APL's approach to moving technology from research to the field could be replicated in a large collaborative entity. While this is important, he summarized the value of maintaining the continuum of R&D for DoD mission relevance: APL focuses on creating materials and innovations that can be validated, reduced to practical demonstrations, scaled, manufactured, and then transitioned in collaboration with industry and DoD system integrators.

Discussion

Erik Svedberg, National Academies, asked how best to enable smooth technology transitions. Venkatasubramanian urged humility with agility, noting that technology transition is challenging but can be done. He suggested setting reasonable expectations of product-level demonstrations and timelines, validating, and pivoting when appropriate. In response to a question from Johnson, Venkatasubramanian described his past work as a learning experience and reiterated that APL has both the independent R&D and collaboration with other sectors (e.g., Space Explorations, Asymmetric Operations, Air & Missile Defense) for technical innovations to be tested ahead of actual transitions to DoD. He noted that DoD could play a larger role in helping to enable collaborations to transition technology. Kingon posed a question about APL's culture as compared to a university. Venkatasubramanian cautioned university researchers, particularly those motivated to realize their innovations for real-world operational needs, to avoid both "irrational exuberance" from their high-impact journal publications and distraction from fleeting technology pulls and to work with sustained enthusiasm and an entrepreneurial spirit to explore all avenues for funding.

KEY TAKEAWAYS FROM DAY 1 PRESENTATIONS AND DISCUSSIONS

Workshop planning committee chair Julie Christodoulou, Office of Naval Research (retired), summarized key takeaways from the first day of the event. She observed that several effective collaborative models with significant activity over the past decade were discussed, including the Manufacturing USA program, other PPPs, and other communities of innovation. She noted that such consortia are particularly valuable for providing new equipment and developing the workforce. Some participants discussed

entrepreneurial perspectives as well as the differences in market-driven and technology-driven innovations. One workshop participant pointed out that the product is more important than the technology in many cases and that thinking about technology from the perspective of a marketer is critical. The value of commercial off-the-shelf technologies for demonstration was also discussed, as well as implications for future systems of decisions made at the prototyping level. She highlighted that participants were reminded about the permanency of some research and the value of national laboratories. She added that organizations such as APL, which, given its more corporate laboratory–like structure, pushes boundaries in combinatorial and accelerated high-throughput testing of unique systems, are important. One workshop participant also addressed the "woes" of government contracting and the fact that working with civilian contractors is more straightforward.

CASE STUDY 5: LESSONS LEARNED FROM DEVELOPING AND IMPLEMENTING DIGITAL TWINS

Rollett expressed that although various definitions exist, in 2024, the National Academies described a digital twin in part as "a set of virtual information constructs that mimics the structure, context, and behavior of a natural, engineered, or social system," and advocated that "[verification, validation, and uncertainty quantification] be deeply embedded in the design, creation, and deployment of digital twins."¹⁷ Thus, a digital twin is a model of an intended or actual real-world physical product, system, or process that serves as a digital counterpart of it for purposes such as simulation, integration, testing, monitoring, and maintenance. Digital twins are valuable for metals additive manufacturing in particular because they encode knowledge about processes and mechanical behavior. He asserted that laser powder bed fusion is now mature enough to build a digital twin that represents connections between microstructures and fatigue.

Rollett shared that his team has been developing a digital twin to support metals additive manufacturing of flight-critical parts. This work is being conducted by the Institute for Model-based Qualification and Certification of Additive Manufacturing (IMQCAM),¹⁸ a NASA Science and Technology Research Institute led by CMU. IMQCAM will build and implement a probabilistic digital twin to connect fatigue behavior to materials process via microstructure. The goal of the project is a model-centric workflow that closes critical gaps between current capabilities and efficient qualification and certification of parts by metals additive manufacturing. The team initiated its work with Ti6Al4V but will transition to Inconel 718. He said that a sufficient knowledge of thermophysical properties enables melt pool prediction, which combined with thermal histories and track pattern enables microstructure prediction, which in turn enables properties prediction. He added that IMQCAM's end-to-end digital twin connects at least 20 models, does not have a linear modeling path, and incorporates the following three critical considerations. First, process modeling is complex, with the need to understand feedstock and melt pool variation. He stressed that decisions are made about the format for transmitting data so that information flows through the entire digital twin, and key process variables should be identified and ordered to model the process. Second, multi-scale modeling helps to predict the location of cracks. Third, the digital twin–assisted qualification and certification process starts by building a complex graph of the digital twin; uncertainty quantification is developed for every link in the chain.

Rollett expressed his hope that these digital twins, currently focused on TRLs less than 4, will be used to model real components; partner companies could develop a battle-hardened version for use. He noted that this work could increase trust in digital twins and metals additive manufacturing. For a material in a demonstration to be highly trusted, he continued, the TRL would align with the highest materials maturity level.

Discussion

Brent Carey, MACH-20, asked how to increase trust in digital twins. Rollett noted that thought leadership will make a difference, encouraging brave people in companies to use digital twins and provide feedback. Refai-Ahmed inquired how industry in particular can advance the use of digital twins. Rollett indicated that companies are hesitant to share their data; the best solution is to build trust through remote validation by making instances of each model accessible to a company and then asking the company to try the model with its own data. The hope is that the company will then share its data. In response to a

question from Butler about using digital twins to support manufacturing scale-up and transfer, Rollett said that partner NASA tested IMQCAM's digital twin, successfully supporting a use case for the same part in a different machine.

Richard Vaia, AFRL, posed a question about the feasibility of analyzing economics to show that a digital twin is more cost-effective than another approach. Rollett replied that technoeconomic analysis is not included in IMQCAM's digital twin; an add-on would be valuable. Kingon added that companies would benefit most from seeing a global use case because they make global, not part-level, decisions. Carey proposed conducting a case study with sensitivity analysis. Johnson noted that in addition to performing technoeconomic analysis, engaging expertise in the diffusion of innovation is critical.

CASE STUDY 6: LESSONS LEARNED FROM ADVANCED ALLOY DEVELOPMENT AND MANUFACTURING

David Alman, National Energy Technology Laboratory (NETL), explained that NETL executes programs for DOE's Office of Fossil Energy and Carbon Management and implements programs for other DOE offices. Historically, NETL played a large role both in commercializing zirconium and titanium and in developing melting technologies for reactive metals. He mentioned that although NETL is not a user facility, its equipment has been used to assist others with upscaling, particularly for alloys and their applications, through cooperative research agreements and cooperation with funded research partners. Key NETL facilities for materials engineering and manufacturing include the Advanced Alloys Signature Center, Functional Materials Development Laboratories, Carbon Materials Manufacturing Facility, and Advanced Sensors Development Laboratories.

Alman focused primarily on the Advanced Alloys Signature Center, which has a liquid metals processing laboratory, a thermomechanical processing laboratory, and severe environment corrosion and erosion research laboratories. NETL's alloy development research relies on an integrated materials engineering approach and focuses on aluminum, refractory alloys, copper, steels, superalloys, and high entropy alloys. He explained that NETL is interested in alloys in particular because fossil energy conversion involves a harsh environment; furthermore, the components are very large and capital-intensive to build, and thus have to last a long time. The facility allows for complete alloy development, with design and discovery, assessment at condition, and manufacturing methods and scales that readily translate to industrial practice. Current capabilities in alloy fabrication include melt processing (e.g., air induction melting, vacuum induction melting, and vacuum arc remelting/electro-slag remelting) and thermomechanical processing (e.g., heat-treatment furnaces to 1600°C, a 500-ton press forge that will soon be replaced by a 1,500-ton press forge with data acquisition and control capabilities, and 2- and 4-high roll mills). The facility uses the feedstocks that an industry manufacturer would use.

Alman also discussed NETL's work in material selection and development with a DOE-funded program on advanced ultra supercritical steam turbines. An alloy that could withstand the 760°C steam temperature and be welded to dissimilar metal components was needed. The result was a demonstration of thick-walled castings of wrought gamma prime–strengthened nickel superalloys—an enabler for steam turbines. The next step was to work with project partners on upscaling, for which NETL designed heat treatments appropriate for a commercial facility. Alman shared another example of NETL's success with the development of a refractory brick to double the service life of slagging gasifiers. This material was licensed to HarbisonWalker and is now found in nearly every slagging gasifier worldwide. Lastly, he highlighted a project initially funded by a small U.S. business for which NETL assisted with prototyping and upscaling manufacturing of a platinum-modified stainless steel for coronary stents.

Discussion

In response to a question from Cao, Alman said that AI and machine learning tools are being used to design the microstructures of high-strength, low-alloy steels and to determine how to control those microstructures during processing. Alman added that customers want to see ample data to be comfortable with models. For example, when NETL puts a brick into an active gasifier, a company has to be confident that the design and material will not negatively impact the function of its gasifier that is producing a commercial product.

Rollett asked if NETL considered creating a physical database of its materials. Alman acknowledged that a centralized library of materials for industry is a good idea; for example, DOE recently issued a call that includes the development of a library of steel materials that can be recycled. Kurfess inquired how NETL's capabilities can best be leveraged. Alman replied that an understanding of available tools for upscaling is important, and networking with organizations to leverage one another's equipment and capabilities is key. Kurfess added that better coordinating resources could accelerate scale up and the digital thread could enable connections. In response to a question from Johnson about knowledge curation, Alman suggested creating databases of capabilities and their applications for the government.

CASE STUDY 7: LESSONS LEARNED FROM AN ENTREPRENEUR'S PLAYBOOK

Shaun Walsh, Peak Nano, remarked that Peak Nano transitioned from R&D projects to producing products over the course of 18 years. He noted that when making this type of transition, the priorities of the people receiving the message are incredibly different. He explained that technology itself is not the biggest innovation driver; one has to be able to think about a problem differently and change a model or process to succeed. For example, the company Anduril brought virtual reality into intelligence environments by creating a platform to which people can add, making 10–15 attempts before succeeding and securing investment. In another instance, Apple negotiated digital rights management, which changed the way people in the music supply chain got paid. Peak Nano's success came via using an AI system for optics design; instead of changing how optics work, it changed how fast optics are implemented.

In terms of funding, Walsh mentioned that venture capitalists tend to cluster, investing in the same opportunities as their peers; for example, more than one-third of venture capital funding in 2024 supported AI. He said that venture capitalists want to know about the market demand and how a new approach is better and easier than existing solutions. They want to know if a company can scale and if the supply chain is well understood. They also want to know how people will consume a product, how they will buy and at what prices, if a sustainable business can be built, and if the development is in a category for which partners are needed. He then shared the playbook for successful entrepreneurship. He suggested thinking about a project as a product from the beginning, "testing before jumping" by talking to as many people as possible, engaging in competitive analysis (i.e., why your roadmap is best), understanding transformation alignment (i.e., fitting your technology into a company's roadmap), having go-to-market champions, and doing phased scaling. He stressed that how a product is sold is more important than the technology itself. He also underscored the value of both finding partners who will provide introductions to the right market fit and getting the right market scale. He advocated for using a "test matrix" (that aligns markets with champions, transition potential, regulatory issues, competition, patents, disruption, price, channels, manufacturing, and supply chain) as a strategy to eliminate areas that will not be profitable. Using this matrix, Peak Nano narrowed its initial focus to drones, fire control, and night vision.

Discussion

Ladani asked why it took 18 years for Peak Nano to move from research to product and if the process could have been accelerated. Walsh explained that the company failed in four markets before completing the test matrix; had it engaged in that analysis sooner, the company could have saved 5–6 years. The company also tried to build manufacturing technology in-house; instead, it could have engaged partners earlier in the process to save another 2 years and $25 million. He also suggested finding partners to initiate sales motion, which could have saved the company another 2 years. Cao asked if Peak Nano's attempt to manufacture in-house actually helped with product design. Walsh indicated that third-party manufacturers have different perspectives, including a better understanding of scale, feasible volume, and cost. Because Walsh's team did not have that insight initially, redesigns were needed.

Larisa Cioaca, Duke University, wondered how Walsh's company evaluated private versus government markets. He replied that decision criteria were not driven by who would provide funding. Peak Nano's first customers were commercial and then the company was led to government accounts via these commercial partnerships. Refai-Ahmed asked Walsh to elaborate on how the company sustained

support throughout the 18-year transition. Walsh remarked that the company did contract R&D and consulting. When an NSF grant was awarded to develop the material, Peak Nano's design teams and researchers worked with other plastics companies and licensed the IP to them in parallel. After about 10 years, the company was product-ready and sought venture capital funding to purchase manufacturing equipment. Johnson said that Peak Nano is a rare example of a company that secured venture capital funding, and he inquired about lessons learned. Walsh reiterated that having a go-to-market champion makes venture capitalists feel safe. Kurfess inquired about strategies to scale up technology, and Walsh commented that many large institutions have significant R&D grants and venture firms that will make connections to productize solutions, because they need the innovations to stay ahead.

PANEL ON NEW METHODS FOR MID-SCALE IMPLEMENTATION

Bill Bonvillian, MIT, stated that between 2000 and 2010, the United States lost one-third of its manufacturing workforce and more than 60,000 manufacturing plants were shuttered, although not as a result of improved productivity, which has been declining for more than 15 years.¹⁹ U.S. investment in manufacturing capital plants and equipment has also declined. Furthermore, he said that U.S. manufacturing declined in 2020 to 16 percent of world manufacturing output while China's rose to 31 percent. In 2024, the United States had a $1.2 trillion trade deficit in manufactured goods, causing social disruption and vulnerabilities in economic security. To restore manufacturing leadership, he advocated for increased productivity and efficiency with new production technologies and processes. He noted that for decades, the United States focused on R&D-led innovation, while Germany, Japan, Korea, and China developed manufacturing-led innovation. He encouraged the United States instead to focus on both and take action soon, as it is on the verge of breakthroughs in manufacturing with digital production, bio-fabrication, and photonics, for example. He also highlighted a lack of new technology adoption, especially among small and medium-sized manufacturers, which hinders the ability to compete. He offered several solutions to these problems: (1) scale up financing with new tax and financing incentives, (2) leverage Manufacturing USA institutes as scale-up facilities and network their advanced technologies (e.g., provide integrated technology packages that can be readily introduced by industry), (3) use DoD's procurement system to support advanced manufacturing, (4) enhance workforce education, and (5) develop portfolios of new manufacturing technologies.

Cioaca discussed government R&D contracts as platforms for experimentation. She mentioned that 25 years of research is available on the effect of SBIR grants and R&D contracts on the performance of firms. Although much of this research considered only one dimension of federal government engagement—the amount of money available—she noted that while money may address some technological uncertainty, market uncertainty also plays a key role. Start-ups with low market uncertainty and low technology uncertainty typically have ample sources of funding, whereas those with high market uncertainty and high technology uncertainty are unlikely to secure venture capital funds and could leverage government support to test market potential. She pointed out that DoD operates 442 procurement offices—that is, 442 potential customers with different requirements. Firms that engage with multiple DoD contracting offices are likely to get differentiated feedback and consider different applications of their technology. Thus, she studied the effect on start-up performance of experimenting with technologies and markets. She analyzed data from 5.9 million high-tech start-ups (established 1989–2019) in their first 10 years of activity across all types of government financial support and agency interaction. Her key finding was that engaging with multiple government contracting offices increases start-up performance, while conducting extensive R&D may actually decrease it. She noted that only about 106,000 firms successfully engaged with the government in their first 10 years, though they did not depend on the government (i.e., demand accounted for about 20 percent of their total revenue). Start-ups that received R&D contracts generated a similar flow of granted patents to those that did not, but their patents exhibited greater technological and economic value. Furthermore, start-ups that secured R&D contracts from multiple contracting offices had greater variability in patent quality. Start-ups that engaged simultaneously with two or more offices were more likely to have both a higher project failure rate and higher product sales to the government. She found that the number of distinct offices awarding contracts, not the dollar value,

was positively related to start-up employment, sales growth, and successful exit.

Rudnitsky described approaches for mid-scale manufacturing, noting that the best solution depends on the problem at hand. One pathway is adapting existing large-scale technology and sharing resources among multiple partners (e.g., multi-project wafers). Another pathway is implementing small-scale approaches for small volume production (e.g., running additive manufacturing devices in parallel). Yet another pathway is to advance a new technology (e.g., adapt robotically controlled sanding to work in smaller areas). An alternate pathway is recognizing that engineered solutions are available and combining technologies. He added that engaging different minds and experiences helps view a problem differently, for which the Manufacturing USA institutes are well suited. The final technological pathway he presented addressed scale up with the use of pilot lines, given that valleys of death exist both in TRLs 3–7 and beyond TRL 7. He urged that if a technology exists, buy it; if not, be prepared to build it. For instance, DoD has paid both for factories and for the development of new technologies.

Carey detailed his previous experience at Owens Corning, where he managed an innovation portfolio and educated leaders about the best projects to invest in based on the highest value to their businesses. He said that the relationship between market and technology discussed by Cioaca is key to investment decisions; successful business portfolios reserve a portion for new markets and new technologies to remain competitive. He suggested using a stage-gate process; although complex innovation for DoD may not fit neatly within this system, adopting its structure and rigor may be valuable for decision making. For example, an idea is generated in Stage 0 and scoped in Stage 1, and a business case is created in Stage 3. If the technology will be a good investment, it will then pass through each subsequent stage. He also noted that objective third-party advisors could serve as "technology brokers." Both these advisors and the stage-gate process could enable mid-scale manufacturing.

Discussion

Johnson advocated for increased productivity in U.S. manufacturing. He noted that some U.S. regions are experiencing growth in manufacturing, often via foreign direct investments. Bonvillian said these foreign investors are bringing productivity improvements from their companies to the United States. He explained that small and medium-sized U.S. manufacturers, who comprise about 46 percent of U.S. output, have the largest productivity problem—these firms are risk-averse, do not do traditional R&D, and do not have rich capital depth to install new processes. He added that tools to bring advanced manufacturing and productivity gains to these manufacturers would be beneficial. Ladani asked about other steps to incentivize U.S. innovation. Bonvillian reiterated that U.S. R&D agencies do not have a history of supporting new manufacturing technologies and processes; investment could address this problem as well as scale-up issues. Manufacturing USA institutes are an important model, he continued, because they connect industry, larger firms, universities, community colleges, technologies, and local economic development support, but the federal government could also invest on a larger scale. He added that workforce development is key to improved productivity in the United States.

Reflecting on Cioaca's research, Kingon wondered how the manufacturing ecosystem could better engage companies like Clinkenbeard. Cioaca underscored that only 2 percent of all high-tech start-ups over a 31-year period successfully engaged with the federal government. The other 98 percent were either unsuccessful or did not seek to engage. She added that a better understanding of the conditions that enable firms to engage successfully with the federal government as well as strategies to enable complementary public and private funding would be useful. Refai-Ahmed questioned if protocols could facilitate the transition from public to private funding. Cioaca noted that the least likely combination of funding in her research was venture capital and government funding, as the two are "like oil and water," and a project to quantify this friction is under way. Venkatasubramanian described the speed of decision making within the U.S. government funding agencies as the main friction; venture capitalists can decide to invest within only months of seeing a technology roadmap. He said that "time is money" for small companies and urged the government to move faster. Carey commented that people who understand DoD mechanisms learn to work through them, but such operations are foreign to venture capitalists.

He suggested another perspective that the two could remain separate entities with different sets of incentives.

BREAKOUT GROUP DISCUSSIONS: GENERATING HEURISTICS TO SUPPORT DEPARTMENT OF DEFENSE DECISION MAKING IN MID-SCALE MANUFACTURING

Before the workshop concluded, participants separated into small groups to discuss strategies to enhance dual-use, innovation, partnerships, and data for mid-scale manufacturing. Members of each group presented the highlights of these discussions, which are summarized below.

Optimization of Dual-Use to Support Mid-Scale Manufacturing

Cao's group discussed two suggestions for small and medium-sized businesses. First, Cao said that conducting a detailed science and technology assessment during the early technology development stage for proposal funding to analyze the potential for dual-use and corresponding ROI is critical. Second, she noted that visiting a local SBA office will help educate small and medium-sized businesses, with customized coaching about how to do business with the government as well as about related incentives. The government could also provide a playbook, she continued. Her group also discussed the value of an independent third party with expertise in contracting to help these businesses.

Strategic Mid-Scale Manufacturing

Kingon's group considered the key role that DoD plays in fostering manufacturing innovation but noted that investments are critical. Furthermore, thinking holistically about business cases could improve mid-scale manufacturing. Kingon remarked that the appropriate time to refine a business case and create a roadmap that includes strategic options and partnerships is when a project reaches TRLs 3–4. He also said that the business case for identifying a warfighter challenge and looking for a solution is different from the business case for identifying a technology platform and exploiting dual-use benefit; questions remain if DoD is optimized for both. Venkatasubramanian commented that the steps involved in taking technical innovations from TRL 3 to TRL 6 are complex in technology maturation, partnerships, and business acumen; creating a more flexible engagement and contracting process, with more freedom to operate, could greatly increase the chance of ultimate success. Carey urged the government to rethink how it bridges this mid-TRL gap and how a PPP could provide incentives for and reduce the risk of dual-use ventures.

Kingon and Johnson stated that many of the stakeholders that Manufacturing USA was built for are not aware of the institutes' capabilities. Johnson advocated for a systematic approach to raise awareness and build bridges via a trusted, unbiased intermediary. Kingon's group also contemplated if the Manufacturing USA institutes could offer more value for large companies (e.g., with shared pilot facilities) and how small and medium-sized companies benefit from consortia differently than large companies. Adam Rawlett, U.S. Army, mentioned that PPPs that encourage feedback and build knowledge are most effective, and Morgana Trexler, APL, highlighted opportunities to ensure that people communicate across boundaries and that organizations build a network to advance critical technologies.

Data Challenges and Opportunities for Public–Private Partnerships and Manufacturing

Christodoulou's group discussed how metadata are difficult to define, record, and report, and are often missing. The group examined the use of round robins and generic components as demonstration articles to establish if data and models are appropriate. In particular, the group considered challenges with business models that make models and/or additive manufacturing equipment available to competitive industries while protecting proprietary data and ensuring trust—for example, Velo3D²⁰ and Carbon3D,²¹ which developed an incentive-based data sharing scheme (e.g., incoming data are fed back to an algorithm and firmware is updated to improve processing capabilities on a regular basis). Christodoulou noted that findable, accessible, interoperable, reusable data are critical for all partnerships and simple practices could be impactful; for example, requirements for data storage could be clearer and authors could provide machine readable data with their publications. Furrer added that data flow and data ownership are key issues and questions remain if medium-sized companies can be more agile, perhaps sharing data in a controlled way so the customer knows what to do with a product.

Cioaca explored links between patents and products. She described the proposed use of virtual patent markings via product data sheets or product webpages; expanding this technology could illuminate the chain from science to product. Hodge added that this is also an opportunity for the public to support science (and government funding) by better understanding how science enables life-changing products.

Christodoulou also explained that common, realistic metrics for assessing PPP performance would be helpful to guide the formation of PPPs and their maturation. Johnson pointed out that PPPs could enable interoperable data structures. He referenced Germany's Platform Industry 4.0 PPP to harmonize data structures, with particular attention toward "sovereign data spaces" (i.e., interoperability frameworks) and the €156 million annual investment from the German government.²² Cao asked if any other countries have a similar model. Johnson described an International Advisory Board that includes several countries working together to create a mission and governance. Furthermore, the Open Data Alliance is implementing interoperability standards to enable data sharing. He advocated for the U.S. government to be part of all of these conversations.

DISCLAIMER: This Proceedings of a Workshop—in Brief was prepared by Linda Casola as a factual summary of what occurred at the workshop. The statements made are those of the rapporteur or individual workshop participants and do not necessarily represent the views of all workshop participants; the planning committee; or the National Academies of Sciences, Engineering, and Medicine.

WORKSHOP PLANNING COMMITTEE MEMBERS:Julie A. Christodoulou (Chair), Office of Naval Research (retired); David E. Aspnes (NAS), North Carolina State University; Jian Cao (NAE), Northwestern University; Jason R. Hattrick-Simpers, University of Toronto; Angus Kingon, Brown University; Thomas R. Kurfess (NAE), Georgia Institute of Technology; Morgana Trexler, Johns Hopkins University Applied Physics Laboratory; Rudy Wojtecki, Applied Materials; and Pablo D. Zavattieri, Purdue University. The National Academies' planning committees are solely responsible for organizing the workshop, identifying topics, and choosing speakers. Responsibility for the final content rests entirely with the rapporteur and the National Academies.

REVIEWERS: To ensure that it meets institutional standards for quality and objectivity, this Proceedings of a Workshop—in Brief was reviewed by Susan N. Houseman, W.E. Upjohn Institute for Employment Research; Leila Ladani, Arizona State University; Ajay P. Malshe (NAE), Purdue University; and Morgana Trexler, Johns Hopkins University Applied Physics Laboratory. Katiria Ortiz, National Academies of Sciences, Engineering, and Medicine, served as the review coordinator.

SPONSOR: This Proceedings of a Workshop—in Brief is based on work that was sponsored by the Army Research Office and was accomplished under Grant Number W911NF-23-1-0409. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Office or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein. Any opinions, findings, conclusions, or recommendations expressed in this publication do not necessarily reflect the views of any organization or agency that provided support for the project.

STAFF:Erik Svedberg, Samantha Koretsky,Maddi Nicol,Amisha Jinandra, Joseph Palmer, and Michelle Schwalbe

SUGGESTED CITATION: National Academies of Sciences, Engineering, and Medicine. 2025. Mid-Scale Manufacturing and Characterization Capacity for Department of Defense Critical Materials Supply Challenges, Part 2: Proceedings of a Workshop—in Brief. Washington, DC: National Academies Press. https://doi.org/10.17226/29151.

For additional information regarding the workshop, visit https://www.nationalacademies.org/our-work/mid-scale-manufacturing-characterization-capacity-accelerating-scale-up-and-transition-of-technologies-to-mitigate-dod-critical-materials-supply-challenges-a-workshop.

Division on Engineering and Physical Sciences

Copyright 2025 by the National Academy of Sciences. All rights reserved.