Onsite Power for Business: Who Owns the Bottleneck Wins
When a shared resource becomes the constraint, the durable cash flows tend to accrue not to the glamorous demand story but to whoever controls or relieves the bottleneck. Electricity is the current instance of a very old pattern.
The quiet shift underneath the loud story
The loud story is easy to summarize: artificial intelligence needs staggering amounts of electricity, data centers are being built at a furious pace, and much of the developed world is suddenly short of power. It is a good story, and it is largely true. Like most good, true stories in investing, it is also crowded — nearly everyone is already positioned on the obvious end of it.
The quieter story is more useful. A growing number of businesses — data center developers, reshored manufacturers, hospitals that cannot tolerate an outage, industrial operators running around the clock — have stopped waiting for the electric utility to connect them to the shared grid. The wait has simply become too long. In many regions, hooking a large new facility into the grid now takes years rather than months, because it requires new high-voltage transmission lines, new substations, large custom-engineered transformers, and a long march through environmental reviews, right-of-way negotiations, and regulatory hearings designed to protect ordinary ratepayers from bearing the cost of corporate expansion. The process is known as the interconnection queue, and it has become one of the more consequential chokepoints in the modern economy.
So instead of sitting idle in that queue, businesses are installing their own power generation right where the electricity is consumed — natural gas engines and turbines, fuel cells, and solar-plus-battery systems that sit next to the factory floor or the server hall. These are not the diesel backup generators that keep emergency lights on during a storm. They are industrial-grade systems designed to run a facility continuously, independent of the broader grid, as the primary source of power.
Equipment providers in this space have noted a meaningful shift in how customers are approaching these conversations. The question is no longer simply whether onsite generation makes economic sense in the abstract. It has become whether the business can get reliable power at all, and on what timeline. When the answer from the utility is not for several years, a business will pay a premium to make its own electrons. Demand is being pulled forward not because onsite power suddenly got cheaper, but because the grid itself has become the bottleneck.
That single observation points toward a durable idea about where value actually accumulates when a shared resource runs short — an idea that matters a great deal to anyone stewarding long-term capital and deciding where to place it.
When a shared resource runs short, the economics reorganize
This pattern repeats across industries and across decades, which is what makes it worth understanding rather than treating as a one-time curiosity.
Every economy runs on a handful of shared resources that everyone needs and no one individually controls: electricity, water, bandwidth, port capacity, rail lines, radio spectrum. Most of the time these are abundant enough that no one thinks about them. They are background infrastructure. The excitement — and the headlines, and usually the investor enthusiasm — flows to whatever is being built on top of them. The apps, not the fiber. The AI models, not the substations.
Then something changes. Demand surges, or supply gets constrained, or both at once. Suddenly the shared resource everyone assumed was infinite is the thing in short supply. And when that happens, the economics quietly reorganize themselves around whoever can control or relieve the bottleneck.
Consider what happened when internet traffic exploded a generation ago. Enormous attention — and enormous speculative capital — went to the websites and the software companies, the demand story. Much of that capital was eventually incinerated. But a good deal of the durable, unglamorous money went to the businesses that owned the pipes and the connection points: the fiber routes, the data-center real estate, the network interconnection hubs. The demand narrative was thrilling and volatile. The bottleneck was dull and far more reliable, because everyone chasing the demand story had to pay the toll to get there.
Electricity is now going through its own version of this. Three forces have collided simultaneously: the electrification of vehicles and heating has lifted baseline loads; the reshoring of heavy manufacturing has brought concentrated industrial demand back onshore; and the build-out of large-scale computing for artificial intelligence has created new loads measured in hundreds of megawatts at single sites. A modern hyperscale data center campus can draw as much power as a mid-sized city. The grid was planned and built for slow, predictable, single-digit annual growth. It was not built for this.
The constraint is not the ability to want power — it is the ability to deliver it on a timeline that a business can plan around. So value starts moving toward whoever can relieve that constraint. Onsite generation is one answer: skip the grid queue and make power at the point of use. That is the ownable insight, and it is worth stating plainly:
In a genuine shortage, the durable cash flows tend to accrue not to the glamorous demand story, but to whoever owns or relieves the bottleneck.
Why this matters to a long-term owner of businesses
At The Stark Fund, every theme gets examined through one lens. Over long stretches of time, the return on owning a business is driven by the growth of its free cash flow — the actual cash a business generates after paying for everything it needs to keep running and to keep growing. Not the story around it, not the quarterly headline, not the narrative heat. The cash. A business whose free cash flow grows steadily and durably across decades compounds — this year's gains build on top of last year's — and that compounding is where nearly all long-term wealth in owning businesses actually comes from.
So the useful question about any hot theme is never "is this exciting?" It is "where in this chain does durable, growing free cash flow actually sit, and how protected is it?"
Seen through that lens, the onsite power shift sharpens into two distinct kinds of beneficiary — and it is worth separating them carefully, because most commentary blurs them together.
The demand story is real, but its economics are exposed
The businesses driving the electricity shortage — the AI developers, the hyperscale operators, the model builders — are genuinely growing. But their economics are ferociously competitive and capital-hungry. They are spending staggering sums, competing directly with one another in a frantic race for scale, and it is far from settled which of them will earn durable returns on all that spending. Some will. Many will spend enormous amounts of cash simply to keep pace, because their rivals are spending too. That is the nature of a land grab: the collective spending can be rational for the industry as a whole and brutal for the free cash flow of any single participant.
For a long-term owner, "growing fast while burning cash and fighting hard to keep it" is not the same thing as "compounding free cash flow." The two can look identical for a few years and diverge sharply over a decade. Betting on the eventual winner of a fast-moving technology race is a legitimate activity, but it requires picking the right horse in a field where the horses keep changing, and the penalty for guessing wrong is severe.
Two quieter positions that don't require picking the winner
The first is the equipment-and-service chain: the firms that make, install, and maintain onsite generation systems. They sell into all the competitors at once. They do not need to pick the winner of the AI race or the manufacturing reshoring race; they need the race to keep happening. And their customers are, if anything, less price-sensitive than usual, because power is the one thing standing between a completed facility and its revenue. A data center that cannot power its servers earns nothing. A factory that cannot run its lines produces nothing. Paying a premium to get powered now — rather than waiting years in a queue — is straightforward math for a developer who has already committed billions to the facility.
The second position is subtler and often overlooked: the operators who convert power from a dependency into an owned advantage. A manufacturer that can run three shifts without interruption — independent of grid outages or price spikes — produces more units per dollar of fixed plant and protects margins that would otherwise erode during high-price events. A data center operator that can bring servers online on its own schedule, rather than waiting years in a utility queue, starts earning revenue earlier and compounds that timing advantage over the asset's life. In both cases the extra free cash flow comes not from cheaper energy but from higher utilization, avoided delay, and protected margins. A business that treats power as a strategic input rather than a pass-through bill is effectively lengthening the duration and predictability of its own cash flows. That is a real and underappreciated form of competitive advantage.
Both positions share the same attractive feature: they benefit regardless of which particular business wins the underlying demand story. That is the kind of setup that can produce durable, growing free cash flow — an essential product or capability, a customer base that spans the whole theme rather than betting on one horse, and pricing power that comes from being the relief valve on a genuine scarcity.
The most durable cash flows in the equipment chain often hide in plain sight: the long-term service and maintenance relationship. Selling a multi-megawatt prime power system is a profitable transaction. The true economic engine is the service contract attached to that equipment over its operating life, which can stretch for two or three decades. Industrial power systems require continuous monitoring, specialized maintenance, and proprietary replacement parts. A commercial facility will not rip out a functioning, multi-million-dollar onsite power system to save a marginal amount on a service contract with an unproven third party. This creates a stream of recurring, high-margin cash flow that arrives reliably across economic cycles — far more valuable to a long-term owner than the lumpy revenue from new equipment sales alone. When evaluating businesses in this space, the vital question is not merely what equipment they sell, but what share of their revenue comes from these embedded, recurring service relationships.
The hard questions before believing the bottleneck story
The "own the bottleneck" idea is powerful, which is precisely why it needs scrutiny. A theme this attractive draws capital, and capital competes away easy returns. The discipline is to ask what would have to be true for the cash flows to actually last — and to take the counter-arguments seriously rather than wave them off.
Is the bottleneck permanent, or just a timing window?
This is the first and most important question. Some bottlenecks are structural and lasting. Others are a temporary mismatch that resolves itself — and resolves against whoever over-invested at the peak.
Grid delays are real today. But grids do eventually get built. Utilities are investing in transmission, regulators are under political pressure to speed up interconnection processes, and the very shortage that makes onsite power attractive is itself a powerful incentive for the grid to catch up over the coming years. If a business's cash flows depend entirely on the grid staying broken, that is not a moat — it is a window, and windows close.
The stronger version of the onsite-power case does not depend on the grid remaining paralyzed forever. It rests on the idea that independent generation is earning a permanent structural role in how large commercial operators think about reliability and control — not as a stopgap, but as a standard feature of serious facility design, the way mission-critical operations maintain multiple suppliers and redundant systems for anything they cannot afford to lose. The vulnerability exposed by recent grid constraints has permanently altered the risk calculus for facility developers. Even when transmission capacity eventually catches up, it is unlikely that operators of mission-critical facilities will return to relying entirely on a single shared public resource.
This is visible in the growing adoption of "islanding" capability — the ability of a modern onsite system to disconnect from the public grid instantly during an outage or a period of extreme pricing and continue running independently. A facility that can island itself protects its margins and its operations whether or not the grid queue ever clears. That is a more defensible claim than "the grid will stay broken," and it is the claim worth examining carefully.
Where does the pricing power actually live?
Onsite power is a supply chain, not a single business, and not every link in that chain is defensible. Basic components — generic cabling, standard enclosures, commodity solar panels — can be replicated quickly. A surge in demand for those items simply invites new competitors who will aggressively compete away the margins. The durable positions involve something genuinely difficult to replicate: proprietary engineering, regulatory certifications that take years to secure, deeply entrenched customer relationships, or an installed base that customers will not abandon. Heavy electrical infrastructure — industrial gas turbines, high-voltage switchgear, sophisticated microgrid control software — carries formidable barriers to entry. A new competitor cannot stand up a facility to manufacture industrial-grade power equipment overnight; the process requires decades of specialized engineering knowledge, massive capital for precision manufacturing, and the safety certifications that industrial customers require before they will trust a system running at the heart of their operations.
What happens when capital floods in?
New entrants, expanded capacity from incumbents, and private funding all chase the same contracts when a theme becomes visible. The businesses that retain advantage over a full cycle possess something difficult to replicate: a technology lead that keeps extending, a cost position rivals cannot match, or switching costs that keep customers in place after the initial sale. The service contract is often the clearest expression of switching costs in this industry — once a customer's operations depend on a particular system and the team that maintains it, the friction of changing providers is high enough to make the relationship genuinely sticky. Without one of those durable advantages, a bottleneck supplier can enjoy strong results for a few years and average results thereafter, which is a fine outcome for a trader and a disappointing one for a long-term owner.
What to look for when allocating capital
The useful question is never whether a theme is exciting. It is where in the chain durable, growing free cash flow actually sits and how protected it remains over a decade or more. Onsite power for business illustrates the distinction between riding a demand wave and controlling the scarce input that enables it. The former produces volatile cash flows as competitors fight for share. The latter can produce steadier, more predictable streams because every participant in the wave needs the same solution.
Owners evaluating opportunities in and around this theme are well served by asking a handful of concrete questions. Which businesses have secured their own power position, converting reliability into higher utilization and more predictable revenue? Which equipment and service providers have embedded recurring cash flows that compound across cycles, rather than depending entirely on new-project activity? Which links in the supply chain carry genuine barriers to entry — proprietary technology, regulatory approvals, long-standing customer relationships — rather than simply benefiting from today's shortage? And which positions benefit whether the grid eventually catches up or not, rather than depending on the bottleneck remaining permanently in place?
The answers point toward businesses whose growth is tied to solving the constraint rather than simply participating in the broader story — and whose cash flows are more likely to compound steadily across decades than to spike and fade with the news cycle.
The pattern itself is not new. Whenever infrastructure lags a wave of technological or economic change — ports during trade surges, fiber during the internet build-out, spectrum during wireless growth — the parties that control the scarce link tend to capture a disproportionate share of the value created downstream. Onsite power for business is the current instance of that principle. The businesses worth owning for the long run are the ones that will still be collecting the toll long after today's headlines have moved on.
Nothing in this piece constitutes a recommendation to buy or sell any security. This is educational market commentary.