I've been all three sides of this market

June 2nd 2023. I was sitting across the table from a $26B AUM funding team and I realised the deal structure we had put together over the past 6 months was structurally incomplete with the way their fund was designed. Over the next few months they worked with us to get the deal to close. We never did.

The thing I learned in that room is the same thing I keep watching tech companies, site developers, and capital allocators learn one transaction at a time in 2026: the three sides of an energy infrastructure deal are not the same problem viewed from three angles. They are three different jobs, with three different risk languages, and the bridge between them is a layer of work that almost nobody is structurally incentivised to do.

I have done all three. Not in sequence cleanly, the way a CV makes things look. Across overlapping decades, in markets that didn't yet exist as their own asset class. This is what I learned about each side and how they connect.

The technology side

I came in on the technology side. The product was electric vehicle charging as a service, billed as a fixed monthly fee with a small share of the variable revenue, paid for by site hosts and offtaken by drivers. The thesis was that site owners and developers preferred the infrastructure experts to bring the capital to the table and provide the hardware as a service. We closed our first paying deployment in July 2023. By the end of 2025, we were operating in 3 countries with more than 95,000 EV chargers under contract.

The thing I underestimated for the first three years was distribution.

We had the hardware. We had a software platform and operating data. We had a paying customer base. None of that translated into the next deployment without a structural workstream nobody on the team had ever owned: matching a deployment-ready product to a site that had power, access to additional supply, and an investor grade offtake counterparty inside the window the site option allowed.

The pattern was as follows. We would meet a site owner - a hotel chain. They would say yes. We would spend 3 months on the legal documentation. In two out of every three cases, the deal would die for one of three reasons that had nothing to do with the technology: the site owner did not have sufficient supply, the offtake counterparty had a different timeline expectation than the one written into the LOI; or the team on the ground delayed the project for months because they hadn't been part of the original executive sale.

Three counterparties. Four lawyers. 6 months. Zero deployment.

That is the shape of the failure on the technology side. It is also the shape of the failure on the developer side and on the capital side, which I would discover later. The variable causing the failure was not on any single side. It was in the absence of a layer that did the matching work between them.

The site side

As an EV charging as a service business, and lately as a developer of data center sites, I have seen the other side of the challenge as well.

On the developer side, the technology problem disappears almost completely from view. The technology is something you select; it is not something you operate. What occupies the working day is site control, interconnection diligence, offtake counterparty quality, regulatory risk, and the funding stack on the host side of the deal.

With my team in Atlanta we started working on a 490MW deal, which was ideal for a data center. What I learned on the developer side is that the offtake counterparty's clock and the developer's option clock are almost never aligned by accident. When they are aligned, it is because someone has done the work to align them. When they are not, the deal does not close — and the developer carries the cost of the site option in the meantime, which compounds against working capital faster than most developer balance sheets can absorb.

The cost of a dead deal on the developer side is not the legal fees. It is the option that was not renewed, the site that was not held, and the next deal that did not happen because the team's attention was on the dying one. The half-life of a developer-side option is significantly shorter than most non-developer counterparties think it is. By the time the tech-company CFO has finished educating themselves on what a 15-year offtake structure looks like, the developer has typically had to choose between extending the option (at a cost the project economics can't absorb) or letting the site go.

This was the second pattern I started watching for: not the failure of any individual side, but the timing mismatch between the three clocks.

The capital side

I began my journey in finance straight out of college, but it wasn't until 2021 that I started sitting across the table from infrastructure debt and PE money.

The capital side surprised me the most.

I had assumed, having been on the technology side and watched deals die for non-technology reasons, that the capital allocators on the other side of the table had a structural view of why deals were dying. They did not. They had a deal-by-deal view, in which each death looked like a bilateral problem — this developer was too slow, this tech was too early, this site was too marginal — and the structural pattern across deals was not visible to them, because their information flow was filtered through the deal team's incentive to keep the next memo moving.

The 283-investor process taught me this directly. Approaching finance partners, you need to understand what they need from the deal. What is their fund focused on, what is the fund size, minimum cheque size, sweet spot and maximum single exposure. What return do they need to get and what terms can they simply not accept even if it made the deal golden.

What I learned on the capital side was that the mandate-clarity problem and the deal-flow-quality problem are the same problem. Capital with a vague mandate is allocated against vague deal flow. Capital with an operationally specific mandate is allocated against deals that have been pre-qualified to that mandate. The funds with the specific mandate deploy. The funds with the vague mandate sub-deploy and don't raise the next vintage.

This was the third pattern, and it closed the loop. Tech side: distribution failure. Site side: clock mismatch. Capital side: mandate-to-deal mismatch. Three failure modes that look like three different problems and are actually the same absent layer.

What I now know to be true

A few things I would have benefited from being told when I was on the technology side, ten or twelve years ago, and which I now try to tell people who are where I was then.

The technology is the smallest part of the deal. What feels like a technology company is operationally a distribution company that happens to make a megawatt. The hours spent improving the product after the second year of commercial operation produce a fraction of the return that the same hours spent improving the matching workflow produce. This is hard to internalise when the product is the thing you are most proud of.

The site is not the asset. The matched site is the asset. A site without an offtaker matched to its clock is a story, not a deal. The developer skill that matters in 2026 is not site identification — there are more candidate sites than the market can finance — but the speed and discipline of matching an identified site to a counterparty whose timeline fits.

The mandate is the moat. Capital that has articulated its mandate at the operational level — geography, technology, structure, counterparty — has access to deal flow that capital with a categorical mandate does not. The work of articulating the mandate is the work of becoming eligible for high-quality deal flow. Most capital allocators have not done this work because nobody required them to. The market is now requiring it.

Nobody on any of the three sides can fix the failure from their side alone. The tech company cannot solve distribution by hiring a US BD lead. The developer cannot solve offtake matching by extending site options. The financier cannot solve deal-flow quality by hiring more analysts. The failure is structural, in the layer that sits between them. Either someone occupies that layer or the deals don't close.

This is what we are building at SitePower. Not a marketplace. Not a brokerage. Not a sell-side advisor. A matching layer that does the work the three sides cannot do from their own incentives.

Why I am personally building this and not something else

The world needs more energy, and it needs more innovative ways to generate it. The last 150 years has seen the human race take great strides forward, with energy generation being a core foundation of this. But evolution slowed, and in many cases regulatory inertia and the politics around it were the cause. Demand now is so high that energy intensive industry will try anything. And that means we are in an energy revolution, not an evolution. Marginal renewable energy solutions are now moving through innovation cycles faster than before. And we are working to make traditional ways of generating power better. My career in energy has been focused on energy innovation the entire time. Wind in the early 2000's, carbon trading in 2007, EVs in 2009, smart infrastructure with IBM in 2013, EV charging in 2017. Each step larger, more complex.

The reason this work feels right is that it is the only work I have done where the lesson from each previous side becomes the toolkit for the next side, and where the absence of the work is producing visible damage at scale. Every quarter I see deals that should close not closing, and capital that should deploy not deploying, and tech that should reach scale not reaching it. The cause is the same one I watched up close on each of the three sides.

The thing the next eighteen months will produce, in our part of the energy infrastructure market, is a cohort of deals that close fast — sub-twelve-month from first conversation to financial close — alongside a much larger cohort that drag and die. The separating variable is structural. Either the matching layer existed for the deal or it didn't.

We are building the matching layer because nobody else has, because the market needs it, and because I have watched, from each side, what happens when it isn't there.

The founding technology cohort is capped at 15 slots, and pricing steps up once it fills. Talk to us about the founding cohort →"