Pretty much every conversation about electricity in 2026 sounds the same. Build more generation. Build more data centers. Wait five years for an interconnection queue to move. It is all supply, all the time, and honestly it is kind of exhausting to listen to.
AP Issa is betting the other direction. As CEO of Encycle, she is basically arguing that the fastest megawatt you can get your hands on is the one you simply never use, and that it has been sitting inside commercial buildings the whole time, hiding on the roof, running when nobody asked it to.
That bet just earned her company a 2026 Sustainability Leadership Award from Business Intelligence Group. And when you look at the numbers behind it, you sort of understand why. Encycle reports more than $12.5 million in customer energy cost savings and 34,308 metric tons of avoided CO2 emissions in a single year, along with 8,300 automated utility demand response events. All of that came out of equipment the customers already owned.
We Are Not Really In an Enron Crisis, But We Are In One
When I asked her whether we are heading into an energy crisis, she pushed back a little on the framing, and then made it sound worse.
"It's not the energy crisis that we saw after Enron," she said. "But we are facing an energy crisis. We do not have capacity on the grid. We just don't."
The math she walked through is the part that stuck with me. Between 2010 and 2019, she said, energy rates in the US only went up about ten percent across that entire stretch. From 2020 through the end of last year, they climbed twenty nine percent. And in just the first six months of 2026, utilities have already filed for $18.6 billion in rate increases that they plan to pull the trigger on going into 2027.
"I don't know if everybody realizes that we're in an energy crisis," she told me, "but one way or another we're all about to feel it."
The demand side is not helping. Gartner now expects global data center electricity consumption to hit 565 terawatt hours in 2026, a 26 percent jump over 2025. Capacity markets are telling the same story in dollars. PJM's most recent capacity auction cleared at a record $333.44 per megawatt-day for a total cost of $16.4 billion.
And here is the wrinkle AP pointed out that I had not considered. Everybody in the utility world spent a decade bracing for electric vehicles. Almost nobody was watching for the thing that actually showed up first.
"Nobody saw the data center train," she said. "That has really taken a problem that they knew that they had to deal with, and it's compounded that problem significantly."
The Case for Commercial HVAC Energy Optimization
So where does a software company fit into all of that? Basically, in the least glamorous part of the building.
Commercial HVAC is a genuinely enormous load that almost nobody watches closely. According to the EIA, cooling accounts for roughly 14 percent of electricity use in US commercial buildings and ventilation adds another 18 percent. That is nearly a third of the meter, running on equipment that most facility teams only think about when somebody complains.
AP was pretty blunt about why. Rooftop units are, well, on the roof.
"Unless it's smoking, leaking, on fire, or someone's actually uncomfortable, no one's doing anything with it," she said. "So the triggering event is usually pretty expensive on the repair side."
Encycle's whole approach to commercial HVAC energy optimization sort of falls out of that observation. There is no hardware to install anymore, which is a big deal. It is an API connection or a file load onto the existing energy management system, and she says they can bring up hundreds of buildings a minute at this point. No retrofit, no capital request, no six month procurement cycle.
Twenty One Years of Data Is the Actual Moat
The part of the conversation I keep thinking about was not about the product at all. It was about what makes the AI work, and it is kind of an unfashionable answer.
"AI is only as good as the data set it's built on," she said, and then compared training a model to raising a toddler. You have to teach it please and thank you, and you have to teach it not to leave the deep freeze door open. What goes in determines what comes out.
Encycle's first installs went in back in 2005, and those early boxes were physical controllers that submetered every single load they touched. Those installs were painful. They were also, in retrospect, an accident of good timing, because they produced two decades of granular data across school districts, big box retail, movie theaters, restaurants, and just about every make and model of rooftop unit in every weather zone you can name.
"We didn't build the AI and then go and try and collect two years of data," she said. Today, she says their virtual submetering lands within one percent of a utility grade submeter, which is not an easy number to hit.
Her framing for anyone selling AI right now was my favorite line of the episode, and it came straight out of her sales background. "Anybody that's been in sales has interacted with Salesforce, and you know, crap in equals crap out. AI's no different."
Nobody Has a Data Problem Anymore
That leads into the thing she thinks the whole energy sector is getting wrong.
"In energy, much like in IT, we don't really have a data problem anymore," she said. "Everybody's got data. There are a lot of companies that had a data lake for at least the last 10 years. Data's not the problem. It's what's the decision that the data drives?"
What that looks like in practice is genuinely useful. Because Encycle has been keeping trend data since 2008, the system can spot a rooftop unit in the process of failing roughly 45 days before it actually dies. Which matters way more than it sounds like it should, because a failing unit keeps drawing power the entire time it is dying.
"You're spending money to just dump hot air into your building," as she put it.
There is also a scheduling angle that surprises new customers almost immediately. They discover their energy management system has a perfectly good schedule programmed, and their rooftop units are just ignoring it, running around the clock regardless of time of day. Customers using trend data instead of waiting for hot calls are seeing a forty four percent reduction in those emergency calls heading into summer, and that is in year one.
What Happens Next Is Not Optional
The uncomfortable part of her forecast is that this stops being a choice fairly soon.
If utilities cannot solve capacity on their side, she thinks they will simply push it onto customers. Real time pricing. Mandatory demand response participation. Capacity charges that show up on the P and L whether you planned for them or not.
But she is not gloomy about it, which I appreciated. "Whenever there's these times of constraints, that is always when we see the best technology really being embraced," she said. "It's an exciting time to be in the market."
For anyone sitting on a portfolio of aging buildings right now, that is probably the takeaway. The grid is not going to get roomier on your timeline. The equipment you already own is the only lever you can pull this quarter.
🎧 Like this conversation? Subscribe to The Winners' Circle Podcast at bintelligence.com/podcast for more conversations with the leaders and companies winning at what they do.