# Interview: Unison Energy CEO on Data Centers Turning to On-Site Power — retained record

> Retained news record from news:datacenterknowledge.

Canonical: https://highsignal.app/data/records/1b10c062de1f91bf

[← sources](https://highsignal.app/data)


source: [news](https://highsignal.app/data/news) (news:datacenterknowledge)  published 2026-05-08 18:15 ingested 2026-05-09 07:45

[original source ↗](https://www.datacenterknowledge.com/build-design/unison-energy-ceo-on-data-centers-turning-to-on-site-power)

## extract

Interview: Unison Energy CEO on Data Centers Turning to On-Site Power
Mariko McDonagh Meier says interconnection delays are pushing developers toward gas-based microgrids as primary power – not backup – reshaping how AI infrastructure gets built.
Unison Energy, a Connecticut-based Energy-as-a-Service (EaaS) developer backed by Tiger Infrastructure Partners, is positioning itself at the center of the power crunch facing AI infrastructure. The company builds, owns, and operates behind-the-meter generation systems – primarily natural gas-based combined heat and power (CHP) and microgrids – for large energy users across North America.
In January 2026, Unison named Mariko McDonagh Meier as CEO, bringing roughly two decades of experience across utilities, distributed energy, and grid services. Her appointment comes as developers confront mounting interconnection delays and rising uncertainty around when – and if – grid capacity will materialize.
Recent activity includes a CHP deployment with General Mills in Missouri and an expanding pipeline tied to large-scale data center projects. As grid timelines extend, Unison’s model – on-site, dispatchable generation delivered under long-term contracts – is gaining traction with customers seeking faster paths to power.
On-site power is moving from a workaround to a default design choice. US data centers used about 176 TWh in 2023 (4.4% of total electricity demand) and could reach 325-580 TWh by 2028 – up to 12% of US load – according to the US Department of Energy. Interconnection queues now total more than 2 TW of generation capacity awaiting approval, underscoring how far new supply lags demand, according to RMI.
That mismatch is already reshaping how projects get built: instead of waiting for grid upgrades, developers are increasingly building power alongside compute – using on-site generation to secure capacity early, then layering in grid supply later if and when it arrives.
Data Center Knowledge spoke with Meier about why developers are shifting away from grid dependence, how on-site generation is scaling, and where the limits are emerging.
The following interview has been lightly edited for length and clarity.
Data Center Knowledge: You’re relatively new in the CEO role. What’s your background in this space?
Mariko McDonagh Meier: I’m pretty new to this seat, but I’ve been in the electricity and energy space for basically my whole career. Most recently, I was CRO at Convergent Energy and Power, working on solar and storage, including for data centers.
So, I’ve been in the space of providing power to large electricity users for the better part of a decade, just from different angles across the industry.
DCK: What changed in the last year that pushed on-site power from niche to urgent?
Meier: It’s funny – when you say “the last year,” in electricity, that’s such a short period of time. I always think in decades.
But it really has changed in a year. The biggest shift is an acceptance that grid power isn’t going to be it – at least not for the next several years.
Two years ago, it was: “What do we need to do to get grid power? How long will it take? What will it cost? What changes do we need to make?”
Now it’s: “This isn’t going to solve my problem in the next five years. So what else can I do? What are my options?”
That’s the shift we’re seeing.
DCK: What problem are data centers hiring Unison to solve right now – speed, reliability, or cost?
Meier: Yes – all of those, but speed is the first one.
If they don’t have reliable power at the scale they need, they simply can’t build. GPUs are a bottleneck, too, but they’re secondary. Nothing works if you don’t have electricity to run the facility.
There’s also a public perception piece. If they can say, “We’re bringing our own power, we’re paying for it ourselves, we’re not relying on the grid,” that helps reduce opposition. It doesn’t eliminate it, but it changes the conversation.
For our industrial customers, cost is usually the biggest driver. For data centers, it’s speed – getting to power fast enough to actually move forward.
DCK: What percentage of your data center pipeline exists because customers can’t get grid interconnection?
Meier: All of it. That’s the reason those projects are coming to us.
DCK: What are you seeing most – delays, denials, or instability?
Meier: Usually those three, in that order.
It’s, “It will take too long,” or “It’s a no,” or it’s so uncertain that you don’t know the timeline or cost.
And these projects require everything to line up – real estate, power, chips, tenants, financing. All of those pieces have to come together at the same time.
So a big part of what developers are trying to do is de-risk each of those inputs. What we’re doing is helping de-risk the energy piece, which right now is one of the hardest ones to control.
DCK: If utilities fixed the interconnection issue tomorrow, how much of your pipeline would disappear?
Meier: I’ve been in electricity for 20 years, and interconnection has only gotten worse. And if I’m a utility, I’m not fixing it for data centers first. It would be for renewables integration, reliability, and all the other things they’re trying to get onto the grid.
Helping data centers get online just isn’t at the top of that list, so I don’t see that changing quickly.
DCK: When you say microgrid, what are you actually deploying?
Meier: For almost all the data center customers we’re talking to, it’s natural gas – turbines or engines.
The “microgrid” part just means it doesn’t rely on the grid. Gas is dispatchable, and customers have a firm gas supply available.
Solar and storage are a different solution. Solar takes a lot of space. We have a hospital project where the entire system fits in something like a three-car garage. To do that with solar, you’d need multiple football fields' worth of panels. As data centers move closer to load, that kind of space just isn’t available, so the physical footprint becomes a real constraint.
DCK: What scale are you seeing?
Meier: Hundreds of megawatts to gigawatts in total agreements.
Most of these are phased. Phase one might be 50 to 100 MW, but we’re signing multi-phase buildout agreements over time.
So it’s not just one deployment – it’s a long-term buildout tied to how the data center scales.
DCK: Are these systems primary power or backup?
Meier: Primary.
We typically provide the majority of the kilowatt-hours the site uses. That’s how the model works.
We build, own, operate, and maintain the system, and we get paid based on what we produce. So our incentives are aligned around keeping those systems running and delivering energy consistently.
DCK: What’s the fastest you can bring 50-100 MW online?
Meier: Historically, about two years from contract to commissioning. Right now, nothing is typical. It’s really a supply chain question.
It’s not just turbines – it’s transformers, medium-voltage equipment, all the pieces that go into the system. When you’re dealing with 70-week lead times, that two-year timeline becomes harder to hit.
A lot of what we’re doing right now is working very closely with suppliers and trying to get ahead of those constraints.
DCK: What’s the real gating factor today?
Meier: It’s a mix, and it depends on geography.
Interconnection is often the last gating item. It’s not always the riskiest, but it drives timing, and you don’t want to commit to major equipment orders without confidence that it will come through.
At the same time, equipment availability is a constraint.
What’s really changed is how projects are sequenced. You can’t do everything linearly anymore. You have to move things in parallel, and that means taking on more risk earlier in the process.
Different partners have different appetites for that, but that’s the reality of building in this environment.
DCK: What breaks first as you scale past 50 MW?
Meier: Reliability becomes the focus very quickly.
Industry-wide, engines might run at about 92% uptime. We’ve historically been closer to 96-97% across our fleet.
Data centers need much higher than that, so what you’re really doing is layering redundancy on top of that base performance.
It’s the same underlying equipment – you’re just designing the system so that you can meet much stricter uptime requirements.
DCK: When does on-site power beat the grid on cost?
Meier: It’s very geographic.
It’s often cheaper than the grid in places like Massachusetts or California. It’s almost never cheaper in places like Oklahoma.
It also depends on how the customer uses power. A steady load looks very different from a peaky load from a pricing perspective.
For industrial customers using combined heat and power, it’s almost always cheaper because they’re using the waste heat. Data centers don’t have that same advantage, so their economics are different.
DCK: Where do deals fall apart?
Meier: For us, it’s about the counterparty.
We’re signing 20-year contracts, so we need a creditworthy tenant – typically a hyperscaler.
We’re not contracting with the developer; we’re contracting through to the tenant. So if that piece isn’t in place, the deal doesn’t move forward.
Once it is in place, things tend to come together pretty quickly.
DCK: Are customers pushing back on 20-year contracts?
Meier: Not really. That’s standard in electricity.
These are long-lived assets, and the contracts reflect that.
DCK: Your portfolio is mostly natural gas. How do you respond to criticism that this locks in fossil dependence?
Meier: Most of our systems can be retrofitted for biogas. So it’s not necessarily a permanent decision – it’s a cost decision today.
And the broader point is that within the current policy framework, it’s not possible to meet all of this demand with renewables alone.
If we want to keep building and growing, it’s going to take a mix of solutions.
DCK: Is there a credible path to decarbonize without sacrificing reliability?
Meier: With biogas, reliability isn’t the issue – it’s supply and cost.
If everyone wanted to switch tomorrow, the question is whether the supply would be there at scale. That’s the real constraint.
DCK: How much of your pipeline is tied to AI data centers?
Meier: By site count, maybe a quarter to a third. By size, they dominate. A gigawatt-scale data center is very different from a 20 or 30 MW industrial site.
So even if they’re a smaller share of projects, they represent a much larger share of capacity.
DCK: Are customers designing around the grid or treating on-site as a supplement?
Meier: Most are planning for on-site as the primary source.
They’re assuming they won’t get the amount of grid power they need. If grid power comes later, they’ll integrate it over time.
So it becomes a phased approach – on-site first, grid later if and when it’s available.
DCK: Is the current pace of AI-driven demand sustainable?
Meier: There’s definitely a hype cycle.
We’re not going to 10x the grid in five years. But doubling – that’s still a massive number, and that feels much more realistic.
There are a lot of incentives in the market to overstate growth, so the reality is probably somewhere in between.
DCK: Five years from now, will a large data center get built with on-site power by default?
Meier: It will require some amount of on-site power. I don’t think it’s going to be fully off-grid. It’s going to be a combination – on-site generation, grid power, storage, and flexibility.
But on-site is going to be part of that default architecture going forward.

## retained text

No document body was retained for this record — only its metadata was stored.