Last updated August 21, 2024
For those who operate in the physical world, it’s difficult to ignore the surge of companies aiming to reinvigorate critical industrial sectors. I’ve noticed that many of these companies are tackling similar looking industries, with similar economic dynamics, and similar technological strategies. I have found the macro “why now” for industrial revitalization to be clear and well documented, but have struggled to pull the thread through to the true operational problems the companies in this space are trying to solve. What does it actually mean to “rebuild the industrial base,” and how can any startup impact such an enormous effort? The below is an ongoing attempt to understand these “new industrialists,” explain what makes them work (or not work), and whether they fit the venture model that has enabled them to proliferate.
Most people have an intuitive understanding that much of the manufacturing in the world has been outsourced to other countries, or in other words, the US doesn’t really “make stuff” anymore. This isn’t entirely true. What is true is that our industrial output decelerated in the 21st century and has completely stagnated for the last 15 years.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXeK1hLekoy9GcfAfUItX47uA1yoFZz5Z98n71_tUgF4Gm3cITsdx13KEw3LPIogYkC_y7ulnUIeQpYsuagudwXgJ-_BvENWRE5NPLTKEao9HfWSJXSTYPDQkfgZjaXVWm8FZsoJZVzx4KJNxY1_E3hGRZY?key=aTBDVTBh2-Z1tYhUsCZzTA
Moreover, most leading indicators point not only to stagnation, but imminent decline. Manufacturing entrepreneurship is on a steep and steady 50 year downward trend.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXczX9mQ2I9r7Ld0NXQfMDxmPwSQmXmdoykW0mdPjohehDdoXOnZmmLaWrctKdtLIE1z2aKIm_vvmkwK19gfbUunThW6rZfQ8iFMV4rv_JdPIhRw5c4nTggPumcbqcGg6frn_6mvQOM1sfuoVk0logRwNbxb?key=aTBDVTBh2-Z1tYhUsCZzTA
And investment in our industrial capacity has also flatlined for the last 15 years.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXcy9ehNqPAxeG086ADLRUpXbmmrxjMlqkWkrLr0ViCtCARBg0rE4oxmEQIuLQhsAh9WZMAqMrDf8_wXwpHFDciY6IN3P4Zz2S1mPJ2Oc8vYoVLV5e_FUCx6R1xVgzjRBZef2Miw5Uzsaz1Yegqv51-s3unm?key=aTBDVTBh2-Z1tYhUsCZzTA
Industrial underinvestment has hurt some sectors more than others, such as metal parts manufacturing, which supplies key components to aerospace, defense, and automotive industries.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXdZBDcRRBUEm3jqJlr7KKHK-wGyKfOEXL1ocQn1xQofYY7803X1kX15yX5CHAY_6Swe16_5j8fHhavVF1WvO6IQpTH5IwFVgJpM5ajjIvAWTUD8d10bTJ4PTFpSF60OkGCWsMdTAPbdYiOessCi6Eobu62H?key=aTBDVTBh2-Z1tYhUsCZzTA
Further, like many of our traditional industries, our manufacturing workforce is aging into retirement, taking with them key skills and knowledge built up over decades, exacerbating our industrial decline. Between 2000 and 2016, US manufacturing employment decreased by 5 million people. Over 95% of U.S.-based manufacturing companies employ fewer than 500 people, and many will likely exit the market as their founders retire, driving a skilled labor shortage expected to exceed 2 million in 2030.
Of course, none of this is news. Harvard Business Review wondered whether the US could ever regain industrial prominence back in 2009, detailing the fragility of the Kindle 2(!) supply chain as proof.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXdkdmqIEDEm40n2LC1LfDHk6X-2mn19OY-dQG432rsKbG5FPdK7D3RuWtw7hjJQePPgfJFEZBG_rIoJ24-W3h14EO6KgTa8EZqXJnw-71lQYC3_dGcLtR_cFfJvpNHRIGkZHlcXhtCnVhTXpbr54BH-7clm?key=aTBDVTBh2-Z1tYhUsCZzTA
The long term trend has resulted in an increasingly outsourced industrial manufacturing sector, and a complex, globalized supply chain to connect new sources of manufacturing output to the large US consumer base.
A good supply chain always goes unnoticed until it breaks down. Most people experienced this during the pandemic when they ran out of toilet paper, or when we collectively needed to ration N95 masks. Maybe you noticed used car prices were at historical highs because new car production came to a halt. But for the most part, a few years post-pandemic and the consumer’s interaction with supply chains is back to normal…for now.
However, this is undoubtedly not the case for business owners. High quality supply chains are the backbone of globalization, and thus the lifeblood of most physical goods businesses. Safe and open seas enable global free trade, allowing global specialization where discrete regions benefit from labor or technological advantages in the supply chain. The result is a network of components manufactured and assembled in the most cost competitive geographies at stage of production.
The operative phrase is “cost competitive.” It is not intuitive, but shipping raw materials and intermediate components all over the globe is actually incredibly efficient. This study (as highlighted in a recent Brian Potter piece) demonstrates why growing cotton in the US, shipping it to Bangladesh, only to have the final T-shirt shipped back to the US for sale is the most economically rational thing to do.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXcYpNLz896_I1lhtfEbcnTYNRVjXSEMk7oCvkCoZHAADHXMG1QbzBD9h6vFvTLsTuLEQYpU9xFlaYOIDzqkdnvdHk-q_x7NcI4hLgzImgiukDSWFAWriRevB6jeFsaR0z9QDhElLulkkeqzs8Z_2DDDKxg?key=aTBDVTBh2-Z1tYhUsCZzTA
Shipping across the world and manufacturing in Bangladesh represents a tiny fraction of the overall cost of the T-shirt due to the low cost of ocean freight and the cost of labor in Bangladesh.
Efficient? Yes. Resilient? Not so much. It is hard to fully understand the risk even in this simplified supply chain. The hypothetical T-shirt manufacturer is vulnerable to all of the geopolitical and environmental risk in Bangladesh, along with the challenges of procuring and maintaining equipment, labor and energy. Then there’s also the complexity of unpredictable congestion at global and domestic ports, and the challenge of interfacing with a supplier with minimal incentive for transparency.
I first experienced the critical importance of the global supply chain at Bowery Farming, where we built highly automated indoor hydroponic farms. These farms required a diverse range of materials, from hairnets to custom control panels. I remember our second farm was delayed for months because a few parts shipped from Europe were stuck at port. This happened multiple times, for multiple farms, and multiple suppliers. Sure, sometimes delays were for key robotics parts, but other times it was just steel or plastic trays.
You can imagine the level of disruption this causes when you’re building a large capex project, particularly one that is developing a relatively new or innovative technology. You’ve already invested millions, you’re staffing up a team, training them, and you’re trying to pre-sell the product. So the clock is ticking on your returns math, and your sales team can’t go into meetings with confidence around timing or volumes. Not to mention you’re a venture backed company with venture backed expectations of blistering growth.
That’s…a lot of risk.
It's the type of risk that not only changes the execution and financial health of the business, but it’s also the type of risk that can metastasize into declining morale, a potentially more insidious issue, as you keep pushing back dates for reasons that feel out of your control.
In short, we live in a world of efficient supply chains, but not particularly resilient ones. For companies building nationally critical products, or businesses developing new and innovative technologies, supply chain resiliency can be the difference between success or failure.
The geopolitical risks inherent in a globalized supply chain are well documented, so I won’t go too in depth here. I think the most succinct explanation of where geopolitics and supply chain resiliency intersect is the notion that “designed in California, made in China” will likely not last another 10 years. China alone has more manufacturing output than the US and Europe combined, output that no longer pertains to low end items but instead to high value products like electric vehicles, solar cells, batteries, and even semiconductors.
As an investor evaluating the US, you might say that we (or the world writ large) has an existential supplier risk. This supplier might uncover an epidemiological crisis (Covid), face a rapidly diminished labor force (aging population), choose to move upmarket into value added products (semiconductors), or decide that our values are diametrically opposed to theirs and conclude they’d be better off without us as a customer altogether. All four could potentially happen at the same time.
China recently concluded its Third Plenum, a conference held ever five years to map out the country’s economic priorities and strategy. The Third Plenum resolution emphasizes self sufficiency in not just the industrial sectors of old, but also “new quality productive forces” whereby China will be the leading source of science and technology innovation, particularly in AI, semiconductors, automotive, and aerospace.
China’s ambition to eclipse the US in all technological development is clear and consistent. The US goal to “de-risk and diversify” from Chinese manufacturing has not been as clear or consistent, but may be the only item Republican and Democratic parties can seem to align on. This posture has most recently materialized in the CHIPS Act and the Inflation Reduction Act, and is clearly articulated in Jake Sullivan’s latest speech on US industrial policy.
There are, of course, other factors breaking down international supply chains. Russia’s invasion of Ukraine, increased piracy in the Middle East, and the rise of populism (and thus anti-globalism) across the developed world, and the retrenchment of the US navy all play a role. However, onshoring and nearshoring manufacturing is largely a phenomenon driven by an understanding that a supply chain dominated by one vendor is a fragile one, particularly when that vendor views itself as your successor in the market.
It is sometimes difficult to fully internalize the consequence of geopolitical developments that challenge much of how the world has worked for the last 80 years. But that is exactly the point. Everyone alive is so far removed from a world without a seamlessly integrated global economy that we can easily overlook the signs of change.
Most of my experience over the last several years has contextualized the push for American reindustrialization as a response to, as described above, fragile supply chains reliant on an actor with an opposing world view. However, I have yet to see economic growth placed at the center of this argument. Perhaps we are too far beyond the days of manufacturing-led growth to imagine it, but resilience and growth can actually be complementary rather than accepted tradeoffs.
In 1969, Jane Jacobs published The Economy of Cities, her second book after her debut work Death and Life of Great American Cities laid the foundation of modern urbanism that we take for granted today. Written prior to the stagflation that defined the US throughout the 1970s, The Economy of Cities ultimately is a treatise on how entrepreneurship, particularly in manufacturing, is critical to economic growth, and how its absence ultimately leads to economic stagnation and decline.
Core to Jacobs’ theory of economic growth is the combination of entrepreneurship and a term she calls “import replacement.” The recipe for growth works as such:
Jacobs believes that the entrepreneurial drive to replace import work is the force behind America’s greatest economic centers. She notes that we are taught in school that New York City grew rapidly once the Erie Canal opened in 1825, but that if the canal indeed incited economic growth, Jersey City would be our nation’s economic capital, not New York. After all, it had equal access to the Atlantic Ocean and had the added advantage of being on the mainland, a benefit Alexander Hamilton believed would one day make it the biggest city in the world. Instead, just prior to the opening of the canal, NYC had begun to eclipse Philadelphia, then the nation’s most productive city, in manufacturing output. Growth was not preordained, but rather developed from within.
Further, Chicago underwent rapid import replacement in the 1830s and 40s, leading to 7x increase in population in the mid 19th Century. After World War II, 1/8th of all new businesses from 1945-1950 were started in Los Angeles, coinciding with its largest period of population growth in the 20th century. Jacobs reinforces this dynamic at the company level as well, noting that the Dodge brothers began their careers by building engines for Ford before developing their own automobiles, work which they then exported. Motorola began by producing batteries and car radios for Chicago area police at the height of Al Capone and Prohibition Era crime before becoming a leading radio manufacturer for the government in World War II.
We can find Jacobs’ theory of economic development in our own world of software, whereby many companies are developed to address the needs of rapidly growing industries, previous employers or internal company operations. Notable examples are Paypal with Ebay, Palantir with Paypal, Salesforce with Oracle, Netsuite with Oracle, Yammer with Geni, and Slack with Tiny Speck. So many of the most prominent B2B SaaS companies today are solving problems created by work that would have been considered new or non-existent 10 years ago.
As I’ve detailed below and in keeping with Jacobs’ theory, it is not surprising that many of the new industrialist companies have emerged to address “import work” at two key companies, SpaceX and Anduril, and are largely concentrated in the same geographic area.
Why does any of this matter? The emergence of new industrialists is not simply about addressing fragile supply chains, or even the “national interest” in the context of a powerful adversary, but rather an opportunity to build a more resilient and sector diverse economic base, and develop the foundation for new areas of economic growth.
Put simply, new industrialist companies are aiming to increase the total capacity and economic viability of manufacturing processes critical to our industrial sectors where demand is projected to outpace supply over the next 10 years. Instead of selling SaaS or acquiring existing facilities, they are building and operating their own manufacturing facilities. Each of these companies sells well known products into established markets, focusing their innovations on the process, not the finished product.
Before going any further, it’s worth mentioning that I spent 5 years at a company that, although structurally different from the new industrialists, was founded with the mission of leveraging technology to rethink an industrialized process. As a result, I have a healthy skepticism about claims of traditional industries being “outdated.” At Bowery, we built large indoor vertical farms growing pesticide free produce shipped to the grocery store within a day or two of harvest.
However, the more time I spent with “outdoor” farmers, particularly when I was leading our seed breeding effort, the more I learned that farming was quite technologically advanced, but perhaps not in the mostly software defined way I was used to. Did I meet 60 year old guys who could look at 100 leaves of arugula and describe the health of each one individually? Yes. Are there any 20 year olds today aspiring to build up a lifetime’s worth of intuition like that? No, probably not. But I did see automated, computer vision enabled tractors and planters, environmental sensors and data, and warehouse management systems smoothly orchestrating workflow. I also observed shockingly efficient human-in-the-loop methods of harvesting and weeding for particularly difficult to mechanize crops. Yes, there are plenty of farmers that are still farming the same way they may have 50 years ago, but the notion that the entire industry had it all wrong was…all wrong.
I am on net optimistic about this emerging wave of industrial focused companies, but I have also developed a few heuristics for evaluating these new businesses.
It’s difficult to keep track of the phrasing du jour for companies building physical products or software for physical industries. 5 years ago it was Hard Tech, 2 years ago it was Deep Tech, and now I’m hearing Frontier Tech. Whatever the term, New Industrialist companies are a subsector of this emerging industry, and the markets that they operate in have a specific set of defining criteria:
All New Industrialist companies are manufacturing products destined for mature end markets. These are tried and true subcomponents of finished products in key industries such as construction, robotics, aerospace, and automobile manufacturing. These subcomponents have been in production for decades, and can range from rebar, to wiring, to control panels.
Buyers care about quality, cost, and speed.
New Industrialist companies are either attacking markets where the existing manufacturers in the space are either a long tail of mom and pop shops, or a collection of highly diversified, often international, behemoths. In the former category, employee headcount can often exist in the range of 5-100, and in the latter, from 5,000+.
Somewhat unintuitively, many of these businesses are tackling industries where the machinery most responsible for production automation is efficient and has a robust supplier and manufacturing base. Said another way, many of the critical processes that could be automated, have been automated. In instances where there are exceptions to the rule, the complexity of the product necessitates manual handling.
However, most of the automated equipment in these industries is siloed, engineered to perform a specific task, and lacking any connectivity or communication with other systems.
Due to the complex yet siloed nature of the existing machinery in these markets, labor is often far more specialized in a typical manufacturing facility. Specialists develop deep expertise in operating specific machinery, and hone their intuition for optimizing output over time. As a result, the quality of output can vary depending on which specialist is operating which machine during a given production run, a dynamic particularly pronounced in markets dominated by mom and pop shops.
Unsurprisingly, the average age for operators in the markets of the New Industrialists skews older. This is common amongst mom and pop shops, where both owners and operators in some industries have an average age in their late 50s and early 60s. For this reason, several private equity firms, such as Re:Build, have identified this dynamic as an opportunity to roll up several manufacturing businesses and find efficiencies through combining and simplifying back end processes.
New Industrialist markets are often characterized by low NPS competitors, largely due to unreliable delivery times and a lack of communication or transparency into timing and cost. This is a particularly salient dynamic amongst operators who are used to operating at tech company level speed, with instant and accurate quotes and delivery estimates. Communication can be slow and comparatively manual, requiring multiple phone calls and email exchanges trading PDFs of critical documentation. For mom and pop shops, the same person usually handles admin, customer service, sales, and sometimes even operations. Operators are stretched thin and don’t have time to invest in customer service. At large enterprises, many buyers are too small to warrant best in class customer service.
While these industries are often somewhat automated, most of the new industrialist entrepreneurs believe that manufacturers have underinvested in software as the critical connective tissue orchestrating and streamlining the soft costs of operations. This can range from opportunities in QA/QC computer vision to cookie cutter verticalized systems of record.
Similarly, the new industrialists meet the above market dynamics with almost identical value propositions and high level operational strategies, regardless of sector or finished product.
There’s a saying in manufacturing that you can get something cheaper, faster, or with better quality, and you can usually get 2, but you can’t get all 3. There’s always a tradeoff. The new industrialist companies operate in markets with mature products and minimal delivery time transparency and communication. Most of the companies below focus primarily on delivering products 10x faster. Nearly every one of their websites has language to the effect of “get X in weeks not months.” Maintaining the bar for quality is a given, but in the markets that these companies are tackling, speed matters more than price. As one former SpaceX procurement manager recently noted, “A reliable manufacturing partner is a better business partner than a pool of eight from which you take the lowest bidder each time but don't have fantastic delivery records.”
Aside from speed, the second most critical factor contributing to low NPS is lack of transparency and communication. More specifically, this includes ease of information sharing, pricing and timing transparency, and proactive communication around delays and mishaps. All of these features can be bucketed under high quality customer service. Like any high performing CX org, best in class customer service is delivered through a mixture of software and service. Buyers are stretched thin and are trying to evaluate their options thoroughly yet quickly. As a result, most of the new industrialist companies highlight their ability to easily consume design files and provide rapid price quotes and delivery timelines.
Each company focuses very narrowly on a specific end product vertical. However, within each vertical, they also promise the ability to iterate through designs and testing, working hand in hand with customers through the initial product development phase. Again, this approach serves to fill the customer service gap in the existing market and fits the market naturally given the largest manufacturers will always focus on large volume runs, and many of these new industrialists will begin by servicing the growing number of new aerospace, defense, and automotive startups. However, to build large businesses, these new industrial companies will need to learn to scale to large scale manufacturing runs.
Despite claims of automating 50 year old processes, the reality on the ground is that most, but not all, of the new industrialists are driving operating efficiencies by connecting or enhancing existing automation equipment rather than completely inventing their own. Many of these industries have already automated the low hanging fruit and have developed networks of sophisticated, high quality equipment vendors.
To be clear, working with and integrating existing automation is a positive sign and demonstrates an understanding of the often non-obvious tradeoffs between effort and impact in manufacturing.
Soft or indirect costs can be thought of as labor costs incurred for all of the work outside of the actual physical manufacturing of a product, but that still should be included in COGS. This may include production scheduling, data entry, machine programming, machine setup, maintenance and QA/QC. Although soft costs may take different shapes across each vertical, almost all of the companies below are driving operational efficiencies first by building software to handle previously un-productized soft cost workflows. This might look like developing an ERP or Manufacturing Execution Software (MES) in house, or building proprietary QA/QC software.
All companies have processes and internal knowledge. However, the software industry has an almost overwhelming buffet of tools with which to document and disseminate internal information. Further, the labor markets for software engineering, design, and product management are so mature that there exists a near infinite amount of documentation around best practices free on the internet.
The manufacturing markets of the new industrialists benefit from no such ecosystem. Specialized knowledge is passed down through an apprenticeship model that is suffering from a dearth of apprentices. As a result, nearly all of the new industrialist companies purport to codify key operating knowledge. However, most of the companies below are too early in their development cycle for me to accurately assess how this effort is accomplished in practice.
The notion ultimately is to obviate the apprenticeship model altogether, and enable anyone to quickly learn to become a machine operator. This significantly widens the potential labor pool, reducing the barriers to scale.
At the highest level, the new industrialist thesis is predicated on 2 core beliefs: 1) demand for key subcomponents in the aerospace, defense, robotics, and automotive sectors will far outstrip domestic supply in the next 10 years, and 2) technology can be used to upgrade existing methods of manufacturing to expand production capacity and drive better economics.
Operationally, this is the north star for the new industrialist model, and as we’ll see below, will be necessary for venture scale returns.
We now have a high level understanding of the types of opportunities driving techno-industrial entrepreneurship, and how the new industrialists are tackling them. However, before diving into the specific list of companies, it’s worth taking a step back and laying out the common pitfalls that these companies may encounter.
Although each market has its own unique characteristics, I have seen several common strategies in this space that are worth avoiding.
The easiest pitfall to miss in industrial tech is that of market risk. Companies operating in the industrial world will point to a large TAM as evidence of a promising opportunity without being realistic about what technology is required to be cost competitive with each pricing tranche within the TAM. Most industrial tech companies start at a price premium with the plan to drive cost down over time, and as a result, their actual serviceable market is materially smaller than it might originally appear.
That’s okay. But most of the new industrialist companies focus on a specific niche within their sub sector of manufacturing and materials. They purchased specialized machinery, trained teams, and built workflows to service that niche. To expand one’s manufacturing capacity to a new product set or new material is not trivial and may require a significant investment in capex and training.
In industrial tech, market expansion is not a matter of adding a new feature at zero marginal cost. In SaaS, we ask ourselves if this new product or that new feature is one that our customer base wants and needs relative to all of the other products we can build. In the industrial world, we largely know what products our customers want. Instead, we must evaluate each new product or material category by asking the question, can we expand at the same level of profitability and product quality at which we operate today?
At Bowery, I led our effort to expand into strawberries (from leafy greens and herbs). From the outside, the move to strawberries seemed straightforward - you still need water, light, nutrients and airflow, with a new addition of pollination. However, to grow strawberries economically, we would have required a significant redesign of the system we had spent the last 5 years building and tweaking. This shift would have required significant capex investment, and even more time for training. While the strawberry category was certainly part of our TAM, it would have been a critical error to project this revenue expansion without making commensurate adjustments to the operating model at worse economics relative to the existing business line.
And sometimes, one step back doesn’t yield two steps forward.
Almost every technology company with enormous venture funding at some point feels the pain of scaling too quickly. Most companies experience this pain in the form of degrading talent quality. For industrial tech companies, this problem is far more complex and existential. Due to the significant capex investment required for scale, expanding one’s footprint before reaching facility level profitability (or at least a clear path to UE profitability) can be catastrophic. It seems obvious but is worth repeating, that when you build a facility that is losing money, the more product you sell, the quicker you run out of cash, and changing course is not simply a matter of laying off staff (who is going to run the facility?), but also a further capex investment in retrofitting an already broken system.
Industrial tech companies that scale too quickly can often find themselves having dug a hole that is simply too deep to crawl out of. And for anyone who has scaled a manufacturing project by several orders of magnitude before, you know that almost everything that can go wrong will go wrong. It is not as simple as just “doing it bigger.” 25x requires a completely different set of operational considerations, project management, and supply chain management. It requires operational manufacturing expertise that the company hasn’t truly built a muscle for yet.
As I mentioned above, industrial companies compete on price, quality, and speed. The new industrialist companies almost entirely spike on the speed vector and differentiate from competitors by creating a far superior service. However, it is important to remember that because these are mature markets, one cannot exceed on speed but underperform on quality or price. Both must at the very least be maintained relative to competing products in order to even have a chance of breaking into the market.
Techno-Economic Analysis (TEA) is a concept I’m familiar with from experience in the climate world, and is often used in the context of new technologies or science. Put simply, TEA is a “modeling framework that aims to help forecast a new technology’s ability to compete in an industrial market.” If you’re interested, Elliot Herschberg has a great overview of the framework and how it is used in biology.
In the context of industrial tech, putting together a TEA is the process of mapping out every single process, workflow, and piece of equipment such that you have a complete end to end picture of how the product is made, and then ascribing economic values to each part of the process. Typically, the greatest effort is in manually measuring the labor attribution of every single process in the facility, from setting up the machinery to manually entering data.
The best companies will do this work, and use the TEA as a roadmap for where technology can best be leveraged to drive operating efficiency.
From the outside, it's easy to misconstrue “slow and outdated” for “un-automated.” Anyone who has seen an industrial operation before understands that while there are certain processes that remain manual, much of the key operational bottlenecks are indeed automated with high performing and sometimes expensive machinery.
Entrepreneurs are fundamentally optimists when it comes to technology, and as a result, one of the most common pitfalls of founders entering an industrial setting is to attempt to not only automate every step of a production line, but to do so using proprietary technology. Admittedly, automation is cool and provides a compelling story to investors and the public, even if it is superficial.
However, a burgeoning company can make unrecoverable mistakes early in its life by investing too much in automation. Unlike a bug or a poor performing new feature, automation is not only a massive investment in capex, but also in engineering and operating time. Recovering from a failed implementation of proprietary machinery requires a reorganization not only of the tools being used on the shop floor but also the entire process that was built around the original piece of equipment. Additionally, new robotics can often end up being more expensive relative to less productive but cheaper and more adaptable manual labor. Lastly, the hard truth of manufacturing is that investing in a 70-80% automated solution has a much higher ROI than the 100% solution. The remaining 20-30% is often the most difficult part, and simply not worth the incremental gains.
Start manual, observe what can be automated, and use as much off the shelf machinery as possible. Seek to only automate the next biggest production bottleneck.
In the physical world, stuff breaks. Sometimes product lines go down and need to be fixed, suddenly reducing supply for a crucial order. Yield, the amount of product you get out for a certain amount of resources you put in, drives the economics of a facility along with throughput. Sometimes one production run yields better than another.
When modeling the long term financials of a facility, it is common to forget the idiosyncrasies and edge cases of the physical world that impact unit economics. Coming from the software world, it is easy to overlook maintenance opex and capex, production buffers, and yield loss throughout a budget. If your model isn’t thoughtful about these dynamics, make sure to add them in.
It is tempting to “pro forma” every assumption when building projections for a physical manufacturing start up. In the beginning, it feels like an intellectually honest exercise. After all, it is unlikely that your operating model assumptions will be representative of the at-scale version of your facility one day. Labor costs may be cheaper in a different location. The production scheduler tool is halfway built but you know it’s going to drive long term savings.
The pro forma approach may make sense in the early days of operating or when painting a medium term vision to investors, but when overused it can be a crutch that blinds operators from taking stock of immediate term operating deficiencies. Put simply, pro forma is a way of kicking the can down the road. If you say “when X happens, our performance will look like Y” enough times, you begin to believe it is the truth.
New industrial companies, unlike SaaS businesses, are about reaching unit economic profitability quickly and with as little capital as possible, and then scaling that profitable machine as fast and as large as the market can handle. Companies that operate under pro forma assumptions for too long rob themselves of the discipline of reaching UE profitability quickly.
There have been two prominent, well funded venture backed companies that I would have bucketed under the “new industrialist” moniker were they operating today. Both companies were founded and closed their doors fairly recently, and can provide us with a number of real world lessons for the next cohort of industrial companies to implement.
Tempo Automation was founded in 2013 to empower any electrical engineer to think of an idea and iterate on it quickly with rapid design and manufacturing capabilities for printed circuit boards (PCBs). PCBs function as the backbone of any electronic device, functioning as both a physical skeleton and central nervous system that connects electronic parts of any system. For this reason, they are indispensable components to variety of industries, from consumer goods (phones, TVs), to industrial equipment, robotics, medical devices, aerospace and automobiles.
Tempo Automation specialized in the rapid prototyping and low-volume production of PCBs using a software-automated manufacturing platform. Historically, this process had been a black box, whereby specs were traded back and forth slowly via PDFs in email, and communication around timing and manufacturing delays was either unclear, inconsistent, or altogether absent.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXcFXawGu8Qhxw2XY4zR9pg_wzYMkengzGu5HoiWhp6FOoE3Nh45D-bbWsbwu8AVVLnlWylcrj2DuqXpY1jzXJI2Sk6BwhNu2w_UeLt77WgyfMN5qQoTCJVN-Nci7o-GAD17rPZxZt7QjXCc-1xZuV3RkHI?key=aTBDVTBh2-Z1tYhUsCZzTA
The idea was that by simplifying or automating every step of the PCB design and manufacturing process (above), Tempo could help electrical engineers prototype designs quickly (batches of 5-20 boards), and as a result accelerate the timeline to validating and then manufacturing new PCB designs. Tempo’s goal was to win on speed, quality, and transparency by taking out every single manual step and creating an unbroken digital thread from design upload through delivery.
Tempo described its product offering and differentiation as such:
https://lh7-rt.googleusercontent.com/docsz/AD_4nXdlgVaA-PKe9XzUkB1LVD9lHQF1C5pJoONGqc1qVnsN9RQ6Sq8vxdHBCy01MIc9toIWWwk9l9RgPVaSUv65IAYgzQ9kU7IiAzL3uu8j89GyDyRRVL5WLQK-2ZLDpqd-KlUx-kNaMmbDZxzyNjaYlvjYYb0?key=aTBDVTBh2-Z1tYhUsCZzTA
The idea was to service the industries where time to market and high precision was paramount, such as in the aerospace and automotive sectors. Today, a typical PCB manufacturer takes 20 days to ingest designs, manufacture and deliver a new board. Tempo claimed to only need 5.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXcW6X1IDb_HVl4cDsKtsVXDkWryjJZ9SXM7Zz8XmjFL6Ew4UuE_VpMCKSUEBmhADWlwDgzU2yhxuzQJ7-Nph_zy7cM9jLAhHbjbJizFOBVRQiMGUr4K5c6ealKvvuUc3LyZZSjZ2a66Q_b61ZfcnBTTezY?key=aTBDVTBh2-Z1tYhUsCZzTA
According to Tempo, the small scale commercial PCB assembly market was $290Bn alone, dominated in the US primarily by ~1,100 mom and pop shops, each doing an average of <$50M in revenue.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXfZe_iILTtgcfOVlYPQ_mocnppQ5LJbmn9kSw-wI8sfe0P4lL8qKq6iywJbVHe6k6ezDlod1ErGEqEysbJKYLDR4rSCQex1Y8n-5htv6K7z7FJnv60rKdjp9beYkU1TXsCss8qM3LWFOx1Y7oB_46ldECAg?key=aTBDVTBh2-Z1tYhUsCZzTA
If this all sounds very familiar to the descriptions of the new industrialists above, that’s because it is.
Tempo went public via SPAC in November 2022, providing us a brief window into its operations and performance before declaring bankruptcy a year later.
At a high level, Tempo seemed to have a strong customer base. Although 90% of their revenue came from just two customers, NASA and Lockheed Martin, one might assume that if they could satisfy the largest buyers in the aerospace and defense sectors, that they should be able to win other business.
https://lh7-rt.googleusercontent.com/docsz/AD_4nXdRRNA7QQIE4uQzVB2t2ptkD2xiveuRfq-CpRNPyEwWcUpKMxZUcHbMfNI_0guQCnhd5MGTxM8lf_LFNGSOQpolniOXV-TRxdH-_5AgeA1t0kn05fU_4-J3QttPQKKOttdCSfKlIBA3SDGnHLvPWe5_4v_Z?key=aTBDVTBh2-Z1tYhUsCZzTA
However, a glimpse at their finances told a less promising story. At the time of the S1, the company was doing just about $12.5M in ARR at an 11% gross margin with enormous R&D (91% of revenue) and S&M (78% of revenue) spend. Further, revenue and gross margin had actually declined for 2 years straight, which the company largely attributed to the COVID era semiconductor shortages. This is all on the back of ~9 years of operation, ~$75M in equity funding and $54M in debt financing prior to the SPAC.
Although disruptions in semiconductor availability certainly negatively impacted Tempo’s ability to produce, the economic model was upside down regardless.
In a last ditch effort, the company attempted to roll up other companies within their value chain to acquire revenue and further vertically integrate. These acquisitions ultimately fell through, but it is unlikely that they would have changed a fundamentally broken model.
Before diving in, I’d like to note that building any business is extremely hard, and building one in manufacturing and scaling it at venture speed is particularly so. Anyone who has struggled through building understands that both failure and success are rarely the result of one or two key decisions, but rather many small actions over a long period of time. Businesses are complex organizations of people, and only those within Tempo can fully understand what drove the company’s outcomes. However, I’ll attempt to tease out a few conclusions from the information I have, and relate them to the pitfalls mentioned above.