.svg){kind=avatar}
The energy expansion has a new limiting factor, and it is worth being precise about what it is and is not. It is not demand: load growth from data centers, AI, electrification, and reshoring is real and accelerating. It is not gas: the United States has abundant, low-cost molecules and a midstream network built to deliver them reliably. The constraint is the equipment that converts those molecules into power, and the lead time to get it. The world wants to build dispatchable capacity faster than the hardware to build it can be manufactured and installed.
The most visible symptom is the large-frame gas turbine. For the biggest machines, those rated around 250 megawatts and up, lead times have stretched to roughly five years, and in some cases longer. An order placed today is unlikely to deliver before 2030, and the three manufacturers that build the overwhelming majority of these units are effectively sold out into the back half of the decade; one major supplier expects its reservation slots to be booked clear through 2030 by the end of this year. Smaller aeroderivative turbines, in the 30 to 100 megawatt range, can be delivered faster, often in 18 to 30 months, which is why data center developers lean on them for bridge power, but they are smaller machines that do not fill the same utility-scale baseload role. The constraint also runs deeper than final assembly, reaching down to the handful of firms that cast the nickel-superalloy blades and forge the rotors these machines depend on. That is a real and acute pinch point. But it would be a mistake, and an easy claim to poke holes in, to say the entire expansion lives or dies by turbines. It doesn't. Large-frame turbines are the sharpest example of a broader equipment squeeze, and the industry is responding with a whole portfolio of alternatives.
At Energy Capital Ventures®, that broader framing is exactly why we find this moment compelling. Whichever generation technology wins a given project, two things stay true: the equipment still consumes molecules that the midstream network must produce, move, compress, and deliver reliably; and in a supply-constrained world, getting more out of the system already in the ground becomes more valuable. Both are technology-agnostic. Both sit squarely within the Green Molecules® thesis. And both keep the operators who run the natural gas value chain at the center of the solution — no matter how the hardware question resolves.
In this issue of the Green Molecules® Journal, we look at the hardware bottleneck not as a story about one machine, but as a signal about where value accrues when equipment, not molecules, is the scarce input.
For most of the last two decades, the binding questions in power were about supply, price, and permitting. Today, even fully financed, fully permitted projects with firm gas behind them are waiting on two things: a place in the equipment queue, and a connection to load.
The interconnection problem is severe on its own. Waiting to connect a large new load to the grid has become the single slowest step in bringing a data center online, with timelines running from roughly three years in the fastest markets to seven or more in the most congested ones. That is why on-site, behind-the-meter generation has moved from a niche choice to a mainstream one — developers are building their own power rather than waiting in the interconnection line.
But building your own power runs straight into the equipment queue. U.S. gas-fired capacity in development nearly tripled in 2025, approaching 252 gigawatts — an extraordinary vote of confidence in gas as the backbone fuel. The equipment to realize it is not scaling at the same rate. The cost to build a new combined-cycle plant has risen roughly 66% in two years, build times have lengthened, and the large-turbine order book is full for years. The scarce input is no longer capital or gas. It is the machine that turns one into the other, delivered on a timeline that matters.
Turbines are the marquee constraint, but they are not the only way to serve new load, and developers are moving aggressively across a menu of options.
Reciprocating engines, essentially very large versions of the piston engines in trucks and ships, have become the leading alternative. They ramp fast, tolerate heat well, use little water, and proved they could stand up a large data center in a matter of months. Among U.S. on-site gas projects that have disclosed timelines, roughly 55% plan to use turbines and about 29% plan reciprocating engines, with the engine share gaining ground. Solid-oxide fuel cells have crossed from pilots into primary power: in April 2026 Oracle contracted for up to 2.8 gigawatts of Bloom Energy's cells to run its AI data centers, one New Mexico campus entirely so. They convert gas to electricity at high efficiency, install in a compact footprint, deploy in months, and, with next-generation designs targeting still higher efficiency and lower cost, look set to take a growing share. Aeroderivative turbines, lighter aircraft-derived units, can be installed far faster than heavy-frame machines. Trailer-mounted mobile turbines, steam turbines fed by packaged boilers, and other configurations round out the mix.
But the same demand wave is now tightening those alternatives too. Leading engine makers are sold out well into 2027, erasing much of their historical speed advantage, and even fuel-cell capacity is filling fast, with deals the size of Oracle's already spoken for much of the near-term supply. In other words, the squeeze is not a quirk of one vendor or one machine. It is systemic across dispatchable-capacity equipment.
This is why widening the lens strengthens the argument. If the story were only about turbines, it would be a bet on one bottleneck resolving. Because it is about equipment and lead times broadly, the conclusion holds regardless of which technology a given developer picks: usable capacity is scarce and slow, and anything that adds it faster — or frees it up from what already exists — is worth a premium.
There is a second reason the hardware bottleneck plays to the strengths of the natural gas value chain. Look at the menu of dispatchable options: large turbines, aeroderivatives, reciprocating engines, fuel cells, mobile units, and notice what nearly all of them have in common: they run on gas.
The competition among generation technologies is, from the operator's vantage point, a competition among customers for the same molecules. Whether a data center campus is powered by a combined-cycle turbine, a hundred reciprocating engines, or a stack of fuel cells, the fuel must be produced, gathered, processed, compressed, transported, and delivered reliably, on peak, day after day. More dispatchable capacity of any kind means more demand on the pipelines, compression, storage, and delivery infrastructure that midstream operators own and run.
That makes the operator a neutral beneficiary of a debate they don't have to win. The generation technology mix will shift with lead times, economics, and siting constraints. The need for molecules, and for the network that delivers them, is common to nearly every path. In an expansion where the hardware winner is uncertain, owning the thing every option depends on is an enviable position.
It also raises the bar in a familiar way. As gas volumes grow across a wider and more distributed set of end uses, the methane profile, measurement quality, and efficiency of the system will draw more scrutiny, not less. That is not a compliance footnote; in many markets it increasingly shapes commercial access. It is also precisely where Green Molecules® innovation adds value.
In an unconstrained market, efficiency is a way to trim operating costs at the margin. In a constrained one, efficiency becomes a form of capacity. Every unit of energy recovered, every point of heat rate improved, every megawatt freed up without waiting in an equipment queue is effectively new supply — arriving on a timeline the manufacturing base cannot match.
This reframes a whole category of Green Molecules® technology. Compression efficiency, waste-heat and waste-pressure recovery, methane reduction, and process optimization are not only emissions plays — they are capacity plays in a market where capacity is the scarce good. And unlike a new generation asset, they don't require a manufacturing slot years out. They install in months, plug into existing assets, and let operators add or free up power without betting on the hardware supply chain.
Crucially, this advantages the operators who own the network. The pressure letdown stations, compression trains, and gas-handling infrastructure that make up the midstream system are full of energy that is routinely discarded in the course of normal operation. When equipment was cheap and demand was flat, that waste was tolerable. In a world of scarce equipment and surging demand, it looks increasingly like stranded value — capacity hiding in plain sight, owned by the very operators best placed to capture it.
This is where one of our portfolio companies, Sapphire Technologies, fits the moment, and it fits precisely because it is agnostic to whatever generation technology sits downstream.
Sapphire's FreeSpin® In-line Turboexpander recovers the energy otherwise lost whenever high-pressure natural gas is stepped down to lower pressure — at city gates, pressure letdown stations, LNG terminals, and throughout the transmission and distribution network. There are more than four million such letdown points globally, and at nearly every one, energy is simply thrown away as the gas expands. Sapphire's technology captures that expansion energy and converts it into clean electricity, with no combustion, no added fuel, and no water, using pressure already present in the system.
What makes this relevant to the hardware bottleneck is the delivery model. The units are modular, ship in months, and add distributed, continuous capacity to infrastructure already in the ground without requiring a new combustion turbine, reciprocating engine, or fuel cell, and without waiting in anyone's manufacturing queue. It does not compete with the generation menu; it sits underneath all of it, recovering energy regardless of what technology sits downstream. Deployed at a letdown station, a system installs in parallel with the existing regulating valves, so operators keep doing their core job while recovering energy that is otherwise dissipated. A single system can generate up to roughly 2.6 gigawatt-hours of clean electricity per year.
This is not a laboratory concept. Sapphire's turboexpanders are already running in continuous operation at an LNG terminal in Japan, and its largest U.S. deployment — more than 70 FreeSpin units across Tallgrass Energy's pipeline system — pairs recovered pressure energy directly with data center load through a partnership with Evolve Energy, whose facilities consume the clean power the units generate. In other words, one of our portfolio companies is already doing the exact thing this bottleneck calls for: turning wasted energy in existing midstream infrastructure into power for new demand, without waiting on the equipment queue.
That is the Green Molecules® thesis in miniature: not replacing the gas system, and not betting on which generation box wins, but making the system do more — cleaner, faster, and with the value captured by the operators who keep it running.
The hardware bottleneck is often framed as a problem or a brake on the energy expansion. It is more useful to read it as a redirection. When the equipment that converts molecules into power becomes scarce across every category, value flows to everything that gets more out of the system already in place, and to the molecules every option depends on.
More load still requires more generation, and turbines, engines, and fuel cells will all remain part of the buildout. But the constraint makes a second path unmistakably valuable: delivering the molecules that feed every one of those technologies, recovering the energy the system already produces, and adding clean capacity inside infrastructure operators already own.
For Energy Capital Ventures®, this is the Green Molecules® thesis at work. The energy expansion will require reliable molecules and new hardware, but it will also reward the operators and technologies that make the existing network more efficient, more productive, and more valuable while the rest of the supply chain catches up. Companies like Sapphire Technologies show what that looks like in practice: capacity recovered from energy that was already there, waiting to be captured, regardless of which machine sits downstream.
