As data center developers race to secure power, they’re increasingly exploring colocation with combined-cycle gas plants, renewables, and, at least in concept, on-site nuclear. The primary argument is speed. If a utility says an interconnection could take five to seven years, building your own power plant looks tempting.
But Brian Janous, co-founder and COO of Cloverleaf Infrastructure, argues that the economics and operational complexity don’t add up. For most data center developers, he said, the model just won’t work.
However, they may have another option. While it has yet to be implemented at scale, Janous thinks grid optimization — deploying everything from virtual power plants to grid-enhancing technologies — could get developers the power they need even faster than behind-the-meter resources.
“This whole notion that we need [a specific technology like] combined cycle plants running very high-utilization to supply data centers just isn’t true,” said Janous, speaking on the Catalyst podcast with Shayle Kann.
Replicating that same 24/7 output is possible even without building new generation, he said, if data centers instead orchestrate several different sources at once. This is what his startup Cloverleaf does; the company takes an energy-first approach to data center development, investing in land where renewable energy is easily accessed, and building data centers there.
“When we sit down and look at a region, we’re thinking about what distributed energy technologies would make sense here,” Janous said in an interview with Latitude Media last year. “What grid-enhancing technologies might help to accelerate or create more output? How do we partner with utilities to make strategic investments in storage on their system?”
Coordinating existing resources comes with inherent advantages, he said on last week’s episode of Catalyst: “Not only is that going to be faster because a lot of these things already exist or are relatively easy to deploy, it’s also going to be cheaper because there’s less overall infrastructure I have to build and it’s going to end up being more sustainable.”
Listen to Brian Janous’ whole interview on Catalyst:
Janous, who for years served as Microsoft’s VP of energy, believes that leaning on colocated generation as a primary source of power — especially via combined-cycle gas — is full of downsides.
For instance, gas isn’t as easy to access as some of the fuel’s backers would have you believe. “It’s not true that I can just always stick a pipe in the ground and get an unlimited amount of gas to build a data center,” he said.
And he also said that overbuilding generation to guarantee redundancy is very expensive:. “If I’m paying $2,700 a kilowatt, roughly, for that generation and I’m having to overbuild it by maybe even two times, the per-kilowatt-hour cost of that system is extraordinarily high.”
The alternative, in Janous’ eyes, is to make the most of the existing grid — and experiment with load flexibility. Six months ago, Duke University researcher Tyler Norris and his colleagues released a landmark paper on data center flexibility, arguing that if data centers and other large loads could embrace flexibility, it could unlock more than 100 gigawatts of extra capacity on the grid. Citing these findings, Janous told Kann, “it is a capacity problem, not an energy problem.”
Other tools for grid optimization include long-duration battery storage, GETs and VPPs. The latter, Janous argues, could cut down on the need for turbines and other behind-the-meter infrastructure. ”The less infrastructure you have to build, the faster it’s going to be,” he said.
That said, this coordinated mix of energy sources for data centers that Janous envisions remains nascent for the time being — though some companies like Orennia and Gridcare have taken steps to embrace it.
But investors have embraced the Cloverleaf approach; in less than a year, the startup managed to secure $300 million in non-dilutive infrastructure funding largely from NGP and Sandbrook Capital in summer 2024, and the company is already securing agreements to build projects.
“I’m encouraged by a lot of folks who are sort of honing in on this problem and trying to figure out, ‘How do we reduce the friction here?’” Janous said. “How do we help utilities to understand, ‘Hey, here’s a better way to do this that would enable more rapid load growth on your system where you’re not losing out to these sort of off-grid competitors?’”
Despite his optimism, Janous nonetheless recognizes that the behind-the-meter generation approach has the appeal of simplicity; in comparison, the multi-party approach is quite complicated: “The orchestration of this is the real challenge.”
The true breakthrough, he said, would come from software that could determine a data center’s optimal energy stack; the perfect blend of on-site generation, VPP’s, batteries and beyond. “When you find that company that has nailed that perfectly, please let me know because we will be their first customer,” he said.
