.png)
Why inventors lose and deployers win
I've had this article on the back burner for a while. Recently, I've engaged with a very interesting startup that matches exactly the thesis I've been mulling over, and it finally pushed me to put these thoughts down.
Here's the thing.
Innovation doesn't always mean inventing something new, it's not always about pushing technological boundaries. Sometimes, the most valuable innovation is taking something that already works and wrapping it in a business model that makes it accessible at scale. A clever financing structure that turns CapEx into OpEx? That's innovation.
Hence: sometimes value is not in inventing, it's in deploying. The companies doing this aren't winning innovation awards, but they're the ones building genuinely valuable businesses.
These businesses go by different names: system integrators and EPC+O (engineering, procurement, construction + operations) companies. This article is about these often overlooked players, who are quietly building billion-dollar businesses while the "innovators" burn through R&D budgets chasing marginal efficiency gains and racing each other to the bottom.
Let's dive in!
(A small note: while EPC+O and system integrators' models differ, I'll use the terms somewhat interchangeably to describe companies that excel at deployment rather than technological advancement in the article)
Lessons from solar and wind
The cleantech boom of the past two decades tells a story that runs counter to Silicon Valley's narrative. It's not a tale of brilliant inventors creating breakthrough technologies in garages, but rather one of methodical system integrators who took existing technologies and turned them into industrial-scale businesses. This history offers a crucial lesson for today's deep tech investors, one that many seem determined to ignore.
Over the past decades, the solar and wind industries experienced nothing short of a revolution. Between 2010 and 2024, the global annual solar photovoltaic installed capacity exploded from 17 gigawatts to nearly 600 gigawatts annually; wind power followed a similar trajectory, with 121 gigawatts of new capacity added globally in 2024 alone. Yet beneath these impressive headline numbers lies a paradox that should give every deep tech investor pause: while deployment volumes soared to unprecedented heights, the companies actually manufacturing the core technologies (the solar panels and wind turbines) found themselves trapped in a financial nightmare. The world's top 10 solar module manufacturers shipped a record 500 gigawatts of modules in 2024, nearly doubling their previous year's volume. Their reward? Collective losses of $4Bn as revenues declined and margins evaporated. The International Energy Agency (IEA) has noted that the net profitability of the solar PV manufacturing sector has been historically volatile and has underperformed other industrial sectors, leading to numerous bankruptcies despite significant policy support. The wind industry painted an equally grim picture. Siemens Gamesa reported losses exceeding 600 million euros in the first half of 2022, while Vestas hemorrhaged nearly 900 million euros in just four months. Consider the cautionary tales of companies like Solyndra, Evergreen Solar, and countless startups that promised revolutionary advances (higher efficiencies, novel materials, etc.): they attracted hundreds of millions in funding and government subsidies, but most still went bankrupt or stagnated, undercut by the rapid cost declines of scaling existing tech and by execution challenges.
These aren't (weren't) necessarily struggling startups or poorly managed companies. These are (were) the titans of their industries, the supposed winners of the cleantech revolution. So, where did all the value go?
The answer reveals a fundamental truth about how value creation works in capital-intensive technology sectors. While manufacturers bled cash, a different breed of company was building empires. NextEra Energy, a Florida-based developer and operator of renewable projects, briefly surpassed ExxonMobil to become America's most valuable energy company, reaching a market capitalization of over $146Bn. They achieved this not by inventing a revolutionary new solar cell or turbine blade, but by mastering the unglamorous work of project development, financing, and operations. Wind farm developers mastered practices to evaluate sites, line up off-take contracts, and deliver projects on time and on budget, ultimately achieving steady returns while turbine manufacturers often struggled with commoditization. Solar and wind projects deliver so reliably because they're mostly very modular projects comprised of standardized units (solar panels, wind turbines) that can be easily replicated.
Let's summarize: on one side, you had companies pouring billions into R&D, chasing incremental efficiency improvements in solar cells or marginally better turbine aerodynamics; on the other, you had companies like NextEra, Sunrun, and others who took these commoditized components and wrapped them in sophisticated business models that solved real customer problems. The winners' competitive advantage was not rooted in a portfolio of patents and invention, but in a mastery of the complex choreography of industrial-scale project execution.
This was no accidental divergence: it was the inevitable result of how hardware technologies mature. As deployment scales up, manufacturing becomes increasingly standardized and concentrated in regions with cost advantages. China's strategic investment of over $50Bn in its domestic photovoltaic supply chain exemplifies this pattern. In 2024, China controlled almost 95% of global production for key components like polysilicon and wafers. This dominance drove solar costs down dramatically, making it the cheapest form of electricity generation in history, but it also transformed solar panels into a global commodity, crushing manufacturers' margins worldwide.
Ultimately, the real innovation and the real value creation happened downstream. Companies like Sunrun didn't invent new photovoltaic technology. Instead, they pioneered the residential solar Power Purchase Agreement, a financial innovation that removed the single biggest barrier to adoption: upfront capital costs. By offering solar installations with zero money down and monthly payments lower than traditional utility bills, they transformed a complex technology purchase into a simple service offering.
Looking back, the system integrators who dominated the renewable energy boom shared several key characteristics.
First, they embraced standardization and repeatability. Early renewable projects were often custom-engineered, one-off endeavors: expensive, difficult to finance, and slow to build. The winners systematically turned these bespoke projects into standardized, bankable assets. They developed repeatable construction methodologies, streamlined procurement processes, and created standardized plant designs that could be replicated across multiple sites.
Second, they mastered global supply chains without falling into the trap of vertical integration. Rather than trying to manufacture every component themselves (a strategy that proved fatal for many), successful developers built resilient networks of suppliers. They became experts at sourcing, logistics, and quality assurance, benefiting from intense price competition among manufacturers without being exposed to the brutal economics of commodity production.
Last, most importantly, they drove financial innovation. The single most important invention in the renewable energy revolution was arguably not a piece of hardware, but a financial contract: the Power Purchase Agreement (PPA), which fundamentally changed the customer value proposition. It shifted the transaction from a high-CapEx sale of an asset (a solar or wind farm) to an OpEx-based sale of a service (predictable, low-cost kilowatt-hours). This masterstroke removed the single largest barrier to adoption for utilities and corporate customers: the need to commit hundreds of millions of dollars of upfront capital to an unfamiliar technology. By financing, owning, and operating the asset themselves, the integrators absorbed the technology and operational risk, presenting the customer with a simple, off-balance-sheet energy bill. This function (the transformation of a complex, risky technology into a simple, bankable service) is the absolute core of the integrator's value proposition and the primary engine of their financial success.
This pattern of value migration from innovation to integration isn't unique to renewable energy. It's a predictable consequence of how industrial technologies evolve: as any hardware technology matures and scales, the basis of competition shifts from novel features to cost, reliability, and bankability. Value doesn't disappear in this transition, but rather migrates: it flows away from component manufacturers toward entities that can manage system-level complexity, aggregate (fragmented) demand, drive down soft costs like permitting and customer acquisition, and most crucially, absorb end-to-end project risk. Therefore, the financial struggles of component manufacturers should not be seen as a failure of the technology, but as a clear signal of a maturing industry where the locus of value creation has decisively shifted downstream.
I believe that this historical pattern serves as a powerful predictive model for what will inevitably occur in other (deep tech?) verticals as they move from the lab to industrial scale.
History repeating itself
In the previous section, we've discussed what has happened. In this section, let me paint you a picture of what is happening.
Across Silicon Valley and beyond, founders are crafting pitches about revolutionary breakthroughs: the next quantum leap in battery chemistry, the carbon capture material that pulls twice as much CO2 from the air, and so on. Many VCs, armed with billions in dry powder and dreams of finding the next unicorn, are writing (have written) checks with enthusiasm. The narrative is compelling: deep tech will save the world through scientific innovation.
There's just one problem. We've seen this movie before, and we know how it ends.
Yet here we are again, watching the same dynamics unfold with eerie precision. The overwhelming majority of deep tech funding flows toward companies promising singular, paradigm-shifting breakthroughs. Investors chase the next novel molecule, the exotic material that promises to redefine the laws of physics. This focus on component-level innovation, while captivating, systematically ignores the hard-won lessons from the renewable energy revolution.
Maybe the root of this problem lies in a fundamental mismatch between how venture capital works and what deep tech actually needs. The VC model is engineered for software-like growth curves and massive, winner-take-all exits. So naturally, the pitch that wins funding isn't about evolutionary implementation and system-level cost reduction: it's about fundamental breakthroughs that will reshape entire industries (--> "moonshoot" narratives). This creates powerful incentives that shape behavior in ways that are actively counterproductive for industrial technology development and deployment, as well as a vicious cycle: founders pitch groundbreaking science to attract VC funding --> VCs fund these science projects in search of outlier returns --> the ecosystem becomes saturated with highly promising but commercially immature component technologies --> these companies become masters at raising the next round of financing but remain utterly unprepared to meet the demands of real industrial customers.
Individual components can be technically brilliant yet make a negligible impact if the balance-of-system costs remain prohibitive. Industrial markets are not necessarily won by the best technology but by the best solution. Industrial customers are structurally predisposed to favor proven, reliable, "good enough" technology with robust operational track records over novel, high-performance alternatives that are unproven at scale. Their procurement decisions are governed by a mandate to minimize risk and ensure operational continuity.
We're funding the creation of a vast library of technological components without adequately supporting the crucial work of assembling and deploying them in the real world: it's like building thousands of powerful engines without anyone working on the cars they're supposed to power.
The irony is that we know exactly what works. The renewable energy sector has already shown us the playbook. The companies that generated massive returns weren't those with the most advanced technology: they were the system integrators who took proven components and deployed them at industrial scale with innovative business models. Yet the deep tech investment community seems determined to ignore this clear historical precedent.
The system integrator model exists precisely to bridge this chasm, so let me talk about it.
The real innovation: business model + (standardized) deployment
There's a multi-trillion-dollar opportunity (ok, perhaps this is an exaggeration!) hiding in plain sight in the (deep) tech sector, and it has nothing to do with inventing the next breakthrough technology. Instead, it's about taking the thousands of proven technologies that already exist (currently trapped in pilots, demonstrations, and small-scale deployments) and turning them into solutions that industrial customers will actually buy. This is the domain of the system integrator, and it represents a quite underappreciated and underfunded opportunity.
To understand why system integration matters so much, you need to first understand what (industrial) customers actually want: a cement plant operator doesn't dream of becoming an expert in amine solvents for carbon capture; a food processing facility has no interest in underwriting the technology risk of a novel thermal battery; you get the gist. What they want is embarrassingly simple: a guaranteed outcome at a predictable cost. They want someone to make their emissions disappear, their energy costs drop, or their processes more efficient, without them having to think about the underlying technology at all. They want OpEx, not CapEx.
This gap between what technologists build and what customers buy is where system integrators thrive. They're not trying to push the boundaries of science. Instead, they're masters at assembling proven components into solutions that actually work at industrial scale. Their innovation is in the field, where they figure out how to deploy technology reliably, affordably, and repeatedly.
The integrator's first skill is what we might call "mining the technology shelf." The deep tech world is full of components that work perfectly well but haven't found their way to widespread deployment. Maybe it's a CO2 capture process that's been proven in gas processing for decades but never packaged for broader industrial use. Perhaps it's a thermal storage system that works great in one pilot but hasn't been standardized for mass deployment. Or maybe it's a set of robotic solutions. The integrator's job is to identify these orphaned technologies and figure out how to make them accessible to mainstream markets. This approach is the antithesis of the typical deep tech startup playbook. Instead of spending years and millions trying to improve a technology's performance by 10%, integrators ask a different question: how can we take something that already works and more seamlessly and frictionlessly deploy it at scale? The answer almost never involves breakthrough science. Rather, it involves mundane but crucial improvements in "packaging", standardization, and business model innovation.
The power of this approach becomes clear when you look at how costs actually decline in industrial systems. While component manufacturers struggle to achieve incremental improvements through R&D, integrators can drive dramatic cost reductions through system-level optimization. This is the difference between improving the efficiency of a solar cell by 1% (hard, expensive, and ultimately limited by physics) versus reducing the total installed cost of solar power by 20% through better project design, streamlined permitting, and improved installation processes (achievable and repeatable), as well as just pure economies of scale. Going back to a previous example, the solar industry provides a perfect template for how this works. The dramatic cost reductions that made solar the cheapest form of electricity generation didn't come from breakthrough improvements in photovoltaic efficiency. They came from turning custom engineering projects into standardized, modular products that could be deployed efficiently at massive scale. The same 250-watt panel design was manufactured millions of times and used in everything from residential rooftops to utility-scale solar farms.
Standardization is crucial, but standardization alone isn't enough. As hinted earlier, the real magic happens when integrators combine modular technology with innovative business models that remove adoption barriers. This is where the concept of "as-a-service" becomes transformative. Instead of asking an industrial customer to make a multi-million-dollar capital investment in "unproven" (to them) technology, integrators/EPC+Os offer to install, own, and operate the equipment themselves. The customer simply pays for the outcome, whether that's tons of CO2 captured, megawatt-hours of clean electricity, or units of industrial heat delivered. This shift from selling products (CapEx) to selling outcomes (OpEx) fundamentally changes the risk equation. The integrator absorbs all the technology risk, operational risk, and financial risk. The customer gets a simple contract with guaranteed performance and predictable costs. It's the industrial equivalent of software-as-a-service, but for physical systems that actually do things in the real world.
The as-a-service model also addresses one of the biggest challenges in deep tech: crossing the "first-of-a-kind" valley of death. Industrial customers are notoriously conservative. They want to see successful deployments at facilities similar to their own before committing to new technology. But someone has to go first, and system integrators solve this chicken-and-egg problem by financing early deployments themselves (not necessarily out of their own pockets, though): by shouldering the risk of these initial deployments, they create the track record that unlocks mainstream financing for subsequent projects. Once again, this is exactly how the solar industry broke through. Companies like SunEdison and Sunrun didn't wait for others to finance solar projects. They raised dedicated funds and put solar systems on their own balance sheets. Once these projects generated reliable performance data and cash flows, traditional financiers became comfortable with the technology. The integrators had de-risked solar by absorbing the first-mover risk themselves.
The economic power of the integrator model compounds over time through system-level learning curves. With each project deployed, integrators gather invaluable data on every aspect of the process. They learn which suppliers are reliable, which permitting strategies work best, how to optimize design and subsequent construction sequences, and a thousand other details that collectively drive down costs and timelines. This operational learning creates a powerful competitive moat and significantly higher cost reductions through better project execution. And unlike technological improvements, which often hit physical limits, operational improvements can compound significantly more.
Consider what this means for competitive dynamics. A new entrant with breakthrough technology faces an uphill battle against an established integrator deploying slightly inferior technology but with superior deployment capabilities. The integrator can likely deliver a working solution faster, cheaper, and with less risk than the technology developer can commercialize their innovation. And if the new technology truly is superior, the integrator can simply license it and plug it into their existing deployment platform. This is why I think that the biggest value creation opportunities increasingly lie downstream from pure technology development: given the choice between tech risk and deployment/supply chain risk, deployment wins every time in my book. Why? Because it has a "linear" relationship with capital: it's predictable, solvable, and scalable. Tech risk, on the other hand, can consume years and millions with no guarantee of success.
The blueprint is clear. The solar and wind industries have shown us exactly how value migrates from component innovation to system integration as technologies mature.
That said, understanding the opportunity is one thing, executing on it is quite another.
A playbook for value creation
Building a successful system integrator (from a "VC-backability" perspective) requires mastering a fundamentally different playbook than the typical venture-backed startup.
The goal here is to quickly get to N-of-a-kind, where each unit is no longer seen as an "unpredictable science project". However, someone has to pay for that first commercial-scale deployment, and it won't be your customers (they won't likely pay CapEx for your n-th, either). The integrators who succeed are those who creatively assemble the capital to cross this chasm themselves. Mind you: the first project is not just a profit center, it's your one shot at building credibility. When a carbon capture system runs profitably for twelve months at an actual factory, demonstrating 90% uptime and measurable cost per ton, that operational data becomes the proof that unlocks further customer contracts and project financing for the next hundred deployments. Nothing de-risks revenue (and therefore financing) like a creditworthy customer committing to buy your output for the next decade!
Now, how can this capital be found? Remember what we said earlier. For industrial customers, the ability to adopt new technology often hinges on a single factor: avoiding capital expenditure. The integrator's superpower is transforming multi-million-dollar equipment purchases into predictable operating expenses while absorbing technology risk, operational risk, and financial risk, thus presenting customers with simple contracts for delivered results (e.g. heat as a service, carbon capture as a service). These are multi-year contracts with predictable cash flows, and effectively an investment asset class which might be of interest for many players in the financial markets (e.g. those looking for a fixed annuity). Therefore, each project can typically be structured as an SPV (Special Purpose Vehicle), a legally separate entity that owns the assets, holds all contracts, and serves as the borrowing entity (attracting a stack with a mix of senior debt + equity from infra funds). The beauty of the integrator model is that you don't need to provide this equity yourself: this new age system integrator is asset-light, and it's the OpCo managing the assets for the SPV. Therefore, you earn fees throughout the project lifecycle: development fees for origination and structuring, ongoing operations and management fees, software licensing fees (discussing this down below), and performance-based carried interest when projects exceed targets (the last of which makes this a potentially cash-cow-like business). This capital-light approach enables rapid scaling: traditional developers are constrained by balance sheet capacity, but fee-based integrators can develop multiple projects simultaneously, limited only by execution capacity rather than capital availability (in fact, the project execution itself is also outsourced to contractors, and not carried out with internal resources).
That said, success in the integrator model demands avoiding a critical trap: becoming a glorified engineering consultancy (since we're thinking of this as a VC-backable opportunity, an important clarification; venture-scale returns demand a pathway to non-linear growth, which is impossible if every project is a one-off, bespoke engineering effort). The key is relentless standardization: "productizing the project." This means architecting systems from standardized, modular hardware blocks that can be mass-produced and configured for specific customer needs with minimal site-specific engineering. The benefits compound dramatically: standardized designs slash engineering costs per project; modular construction reduces deployment timelines; factory production drives down unit costs through manufacturing scale; most importantly, standardization enables true learning curves: each deployment feeds improvements back into the standard design, benefiting all future installations. All of this has implications we'll cover later in the article. (point to make: standardization extends beyond hardware into financial structuring as described above: integrators can develop templated approaches to project finance that can be replicated across deployments).
It's hard to generalize (as the system integrator model can be applied to a variety of industries with different hardware that gets "integrated"), but let me spend a couple of words on the importance of software, which represents perhaps an underappreciated aspect of system integration. At its most basic level, software solves the integration challenge itself: the new system can't simply be bolted onto the existing infrastructure, it also needs to speak the same digital language/be able to interact with the environment. Beyond basic integration, software transforms static hardware into dynamic, responsive assets. Consider thermal energy storage: the hardware might not be mind-blowing technology (e.g. quite simply, heated bricks), but the software orchestrating when to charge from the grid, how to optimize across volatile electricity markets, and when to discharge heat to meet customer needs can mean the difference between a "meh" and a "great" return on investment. Additionally, the true power emerges as these software capabilities scale across fleets: each deployed asset generates operational data (patterns in electricity prices, correlations between weather and grid dynamics, hardware performance under different conditions, etc). Machine learning algorithms can extract insights from this data, continuously improving optimization strategies. Eventually, this software layer becomes valuable enough to license as a standalone product. Integrators can offer their platforms to third-party asset owners (e.g. on a per-megawatt basis), creating high-margin recurring revenue streams that scale without additional capital deployment. For robotics, software can bridge the notorious 'sim-to-real' gap that plagues physical AI deployment: by building high-fidelity simulation environments from actual operational data, integrators can train and validate robotic policies in virtual environments that truly mirror deployment conditions. This approach transforms the traditional painful process of on-site robot training (with its countless edge cases and failures) into a systematic, data-driven methodology where robots arrive pre-trained for specific industrial environments, dramatically reducing deployment timelines and de-risking customer implementations.
Besides, software also serves critical internal functions that accelerate deployment timelines. Take TES or carbon capture: integrators can build proprietary simulation tools and digital twins that can model system performance across different industrial applications, compressing what traditionally takes 4-6 months of OEM feasibility studies into days of automated analysis. Such internal platforms would enable rapid project evaluation, independent technology selection, and standardized system sizing - all crucial for escaping the trap of bespoke engineering. In robotics integration, proprietary software platforms could transform how manufacturers approach automation altogether. Advanced AI-powered tools now could/can allow manufacturers to describe automation needs in plain English and receive complete 3D work-cell designs, ROI calculations, and deployment timelines, all before committing any capital, compressing what traditionally would take months of feasibility studies and engineering consultations into automated proposals generated in minutes, democratizing access to automation for manufacturers who lack in-house robotics expertise (delivered as a service by the system integrator, who will then execute asset-light on the project).
The ultimate evolution is becoming a true platform business where standardized hardware, sophisticated software, and templated financing structures combine to create a flywheel of deployment and value creation.
Ground zero for integrators: a sector example
Which sectors offer the most compelling opportunities for system integrators?
I'd say, wherever complex industrial customers need guaranteed outcomes rather than promising technologies, system integrators are stepping in to bridge the gap. What makes these opportunities particularly compelling is their shared characteristics: each sector features proven core technologies held back by deployment challenges; each serves conservative industrial customers who prioritize reliability over novelty; and in each case, the winning business model transforms capital expenditures into operating expenses through "as-a-service" offerings.
Let me discuss one example (inspired by a fairly recent conversation with a startup, and mentioned earlier in the article) with you: TES (thermal energy storage).
Industrial process heat represents an interesting opportunity. This sector, requiring temperatures between 100°C to 1000°C+, accounts for roughly 20% of global energy use and similar CO2 emissions. Yet it's received far less attention than electricity or transportation, perhaps because the challenge seems so intractable. How do you replace the instant, controllable heat of fossil fuel combustion with intermittent renewable electricity? Thermal energy storage. TES uses simple, proven materials like refractory bricks, molten salts, or high-performance concrete, heated by electric resistance (technology literally centuries old), to store energy as high-temperature heat. What's fascinating about the thermal storage market is how it exposes the exact dynamics we've been discussing. Multiple hardware manufacturers have developed functional thermal storage systems: the technology convergence is striking, with most using variations of resistive heating and high heat capacity materials. Yet these OEMs consistently struggle with commercial deployment. They're technology companies at heart, with teams focused on improving heat retention or charging efficiency by marginal percentages. What they lack is the project development expertise, energy market knowledge, and financial structuring capabilities needed to actually sell heat to industrial customers. Sounds like a fantastic opportunity for a system integrator!
As a matter of fact, some early-stage startups are already tapping into this opportunity. Their real innovation comes in three layers. First, sophisticated software that determines when to charge from the grid, capturing periods of low or even negative electricity prices. Second, these systems can evolve from passive consumers to active grid participants. By registering as schedulable loads with grid operators, thermal batteries can access additional revenue streams, e.g. frequency regulation. This transforms them from simple storage devices into multi-revenue assets that help balance increasingly renewable grids. virtual power plants. Third, and most crucially, the business model innovation. These players offer Heat-as-a-Service, installing thermal batteries at customer sites with zero upfront cost. Industrial customers sign long-term (15-20 year) contracts to purchase clean heat at predictable prices, often structured to compete directly with volatile natural gas costs. This transforms a multi-million dollar capital project with uncertain returns into a simple utility service with guaranteed performance. The integrator doesn't need to provide this capital, which can be raised by financing partners looking for infrastructure-like assets to deploy capital in. This capital-light model enables rapid scaling without balance sheet constraints.
Why wouldn't the OEMs themselves directly offer heat-as-a-service? It's a fair question: after all, who better to deploy thermal storage systems than the companies that build them? Yet time and again, we see OEMs struggle when they attempt to become project developers. The reasons are both structural and cultural.
First, there's a fundamental DNA mismatch. OEMs are hardware companies at their core. Their leadership teams, organizational structures, and company cultures are built around technology development and manufacturing excellence. Project development (with its focus on site selection, permitting, stakeholder management, and financial structuring) is viewed as a necessary evil rather than the core business. When push comes to shove, resources and attention flow to what the organization values most: building better hardware. In turn, this cultural divide manifests in talent acquisition: top project developer executives and energy traders don't want to join organizations where their work is secondary to the "real" business of technology development - they want to be at companies where project development is the hero function, not a support role. Therefore, OEMs consistently struggle to attract and retain the specialized talent needed for successful project execution.
There's also a business model conflict at the heart of the matter. OEMs want to sell equipment: transitioning to long-term service contracts requiring 15-20 year operational commitments fundamentally changes the company's risk profile, capital requirements, and investor expectations. It's like asking Boeing to become an airline: technically possible, but organizationally traumatic. In this regard, the financial engineering required for as-a-service models presents another barrier: structuring SPVs, managing relationships with infrastructure funds, navigating tax equity markets, and optimizing capital stacks across multiple jurisdictions requires specialized expertise that hardware companies rarely possess.
The geographic complexity of energy markets creates another significant challenge for OEMs. Each market requires deep local expertise: understanding grid codes, trading protocols, regulatory requirements, and market dynamics. OEMs might have to spend years building capabilities to participate in just one (!!) regional grid market. Replicating this effort across dozens of markets globally is not economically viable: the learning curve is too steep, the investment too large, and the distraction from core hardware development too significant.
This isn't to say OEMs can't offer as-a-service models at all. Some have found success with hybrid approaches, particularly when serving a narrow customer segment or geographic market where they can build deep expertise. But scaling these efforts globally while maintaining hardware innovation proves overwhelming for most, and the companies that try to do everything often end up excelling at nothing. The smart OEMs know that integrators are not threats but essential partners in market development: after all, every project an integrator deploys represents hardware sales for OEMs.
(Note: to be clear, I'm not advocating thermal energy storage as an ideal opportunity for new-age EPC+Os companies. The sector faces its own market-specific challenges that will, in my opinion, severely limit its venture-scale potential, but it serves as a useful illustration of how the integrator playbook works in practice.)
The case against system integrators
Let's pump the brakes for a moment and examine why the system integrator playbook might not be the goldmine it appears (from a VC investment perspective).
I've used solar and wind as industries in which the model worked wonders. The uncomfortable truth is that solar and wind possessed unique characteristics that enabled their success: characteristics notably absent in most deep tech verticals targeting industrial applications. The solar panel is perhaps the perfect product for this model: completely standardized, infinitely replicable, with universal specifications. A 400-watt panel works identically whether installed on a warehouse in California or a solar farm in India.
The same cannot be said for a carbon capture system at a cement plant versus a steel mill, or a thermal storage system serving a chemical processor versus a food manufacturer. Each industrial facility is a unique organism with its own heat requirements, space constraints, operational patterns, and legacy infrastructure. This brings us to the fundamental tension at the heart of the integrator model: venture-scale returns require standardization and repeatability (-> efficiency gains), but most industrial reality demands customization. You can't simply decree standardization into existence when the underlying hardware (which you have no control over) might not be modular, and the deployment environments are inherently heterogeneous. The result? Linear scaling, if each project requires nearly as much engineering effort as the last. Yes, you might achieve some efficiencies, e.g. your tenth cement plant project will go smoother than your first. But you're fundamentally constrained by the need for specialized engineering resources that don't scale, timelines that don't change, etc.
The hardware dependency problem compounds these challenges. System integrators are ultimately at the mercy of their OEM partners' design choices. If the thermal storage manufacturer builds monolithic systems rather than modular units, the integrator can't magically create modularity through software and financial engineering. They're stuck deploying what amounts to custom equipment for each installation. Solar and wind succeeded precisely because they avoided these pitfalls: they deployed standardized products (panels and turbines) in greenfield locations without complex integration requirements. The grid connection point was standardized, the permitting process became templated, and the technology was inherently modular. (Most) Industrial process applications share none of these advantages, but other verticals might, , or they at least have the potential to fully productize and automate feasibility + system design (e.g. such as robotics; should I write something about it? Let me know!).
From a VC perspective, system integrators in most verticals might not be building scalable technology businesses but rather sophisticated engineering consultancies with better marketing. The margins might be higher than traditional EPCs, but the fundamental economics remain stubbornly linear. This doesn't mean the system integrator model is worthless, far from it. But we need to be honest about which verticals possess the necessary characteristics for venture-scale returns versus those better suited for project finance or private equity models focused on steady, linear growth.
Alright, time to wrap this article up!
Conclusions
The (deep) tech ecosystem today might be trapped in the same pattern that defined cleantech two decades ago: pouring capital into revolutionary breakthroughs while overlooking where value actually materializes. The solar and wind industries showed that while component manufacturers chased incremental efficiency gains and fought brutal margin compression, system integrators quietly built large and profitable businesses by taking those commoditized technologies and turning them into deployable solutions.
Should we just chase system integrators, then?
That's not my point. We absolutely need continued technological innovation: after all, the components that integrators deploy have to come from somewhere, and pushing the boundaries of what's possible remains crucial for addressing many of humanity's and industry's challenges. I just want to be cautious in creating a funding environment so obsessed with breakthrough science that we're systematically ignoring the less glamorous businesses that actually deliver these technologies to market. The companies doing the "boring" work of standardization, deployment, and business model innovation often generate the outsized returns VCs seek.
Yet, I want to be clear about the limitations of this model, too. The system integrator model isn't a universal solution for venture-scale returns. It thrives only where genuine standardization is possible, where you can turn custom engineering projects into repeatable products: a solar farm is fundamentally modular; a carbon capture retrofit for a decades-old chemical plant might not be. Without this repeatability, the risk is that you might be building a sophisticated engineering consultancy, not a scalable platform (growth remains linear). Where the conditions do exist, the opportunity is/will be massive: industrial customers still want predictable, as-a-service solutions rather than science projects. The companies that can identify these pockets and wrap proven hardware in software + innovative financing will capture enormous value.
All in all, the lesson here, in my opinion, isn't to abandon breakthrough innovation for integration, or vice versa. It's to recognize that different sectors demand different approaches, and that value creation in industrial markets might not always follow Silicon Valley's preferred playbook. Sometimes the billion-dollar opportunity lies in breakthrough solid-state battery technology. Sometimes it lies in being the best at deploying commercial-scale energy storage with today's lithium-ion systems.
More appetite for this type of content?
Subscribe to my newsletter for more "real-world" startups and industry insights - I post monthly!
Head to Foundamental's website, and check our "Perspectives" for more videos, podcasts, and articles on anything real-world and AEC.
#DeepTech #VentureCapital #SystemIntegration #IndustrialTech #ProjectDevelopment #HardwareSaaS #IndustrialAutomation #B2BInnovation #Startups
Related Perspectives