The startup Perimeter Compute is emerging from stealth today, Latitude Media has learned, with plans to join the emerging edge computing market in major U.S. cities by using the excess energy capacity found in commercial office buildings.
The company is within the portfolio of Montauk Capital, a venture firm based in New York City that recruits experienced founders to run new startups in the digital infrastructure space. David Hall, a former executive at the AI neocloud Lambda, is Perimeter’s CEO.
“I walked through some buildings in New York and Boston and was astounded that they have everything a data center has: space, cooling, and power,” Hall told Latitude Media in an interview. “And after years of energy conservation efforts, they have reduced their power consumption between 30% and 60%, yet the infrastructure is still designed for that peak load they started with.”
Hall said Perimeter plans to install the latest AI chips in the basements and mechanical rooms of Class A commercial office buildings with between half a megawatt and 20 MWs of spare capacity. That amount of distributed capacity is meaningful in today’s market, where data centers are facing numerous construction bottlenecks including equipment shortages and a strapped power grid. Even half a megawatt of power is enough to serve 240 of NVIDIA’s latest GPUs, which can process billions of user requests per day, Hall said.
Perimeter is the latest entrant into the edge computing market, after two residential energy companies Span and Sunrun launched their own pilots earlier this year, targeting homes with excess capacity largely from battery storage. Hall said those endeavors will be installing fewer chips per building than Perimeter, though the startup Perimeter is similarly targeting the AI inference market. That’s the day-to-day use of chatbots or other programs, which requires less energy than the training of large language models; the latter is what massive AI data center complexes are designed to power.
For more on the question of whether inference will move to the edge, listen to this episode of the Catalyst podcast:
The company could fetch premium prices from hyperscalers because GPUs placed in urban centers can process AI workloads faster than those that need to transmit data long distances. Plus, the actual application of AI is what generates money for companies like Amazon, Anthropic, Google, and OpenAI. So inference compute is in high demand.
“We are getting great signals not just from the OpenAI’s and Anthropics of the world, but also individual tenants of these buildings that are in the financial services, robotics, and healthcare industries,” Hall said.
Perimeter is currently raising capital to buy equipment, which costs about $35 million per MW. Hall declined to disclose how much money Perimeter aims to raise in this first round.
The company is talking to local utilities about the expected increased loads, and hired a consulting firm that already has relationships with them. Some local distribution grids in urban hubs like New York and Boston may already be congested, in part because utilities “oversubscribe” connections.
For example, maybe a local substation can handle 80 MW of capacity at any given time, but a utility approved 90 MW across various homes and buildings because none run at 100% capacity at all hours.
By contrast, GPUs need power all the time, potentially upending a utility’s carefully managed load. “If they have oversubscribed their grid in different locations, we want to protect ourselves,” Hall said.
Perimeter has identified more than a gigawatt of excess capacity in office buildings across the U.S., Hall said. Building owners, for their part, wouldn’t be on the hook for any of the compute costs. Perimeter will pay for the GPU equipment, the installation, and the extra energy the system uses, which a submeter will track. Perimeter will also pay to lease space from landlords and share a portion of the revenue from selling compute with them.
But it’s early days; Perimeter doesn’t have signed contracts yet. It’s still unclear how these contracts with tech companies and building owners will be structured — and whether they’ll ultimately be attractive enough to scale more broadly.
