The next people to ask about your cameras will be underwriters
A database built for a different kind of buyer
On September 3, 2026, Verisk announced the launch of a US Data Center Exposure Database. It covers more than 2,500 US data center facilities with building-level detail: geocoding, footprint, construction type, capacity, and redundancy, mapped onto a 90-meter geospatial grid. The stated purpose is to let insurers and reinsurers price concentration risk, meaning how much value sits inside a radius small enough that one event, a storm, a fire, a grid failure, could hit several facilities that a single insurer or reinsurer has underwritten at once.
This is not a security product. It is closer to an actuarial map. But the reason it exists is worth sitting with, because it describes a shift that eventually reaches physical security too.
The number that explains why
The release cites global data center insurance premiums potentially growing from roughly $10 billion in 2026 to roughly $23 billion by 2030. That is not a modest forecast. Money moving at that pace into a category means insurers expect to be pricing a lot more of this risk in the near future than they have been pricing so far, and they are building the data infrastructure now to do it with more precision than "we insured a building like this before and nothing happened."
Rob Newbold, President of Verisk Catastrophe and Risk Solutions, put the framing directly: "AI is often discussed as a digital transformation story, but it is increasingly a physical infrastructure story." That is the sentence that matters here. Compute demand does not stay abstract. It turns into buildings, power draw, cooling systems, and physical access points, and all of those are things an insurer has always known how to think about. What is new is the scale, and the fact that so much of it is now concentrated in a small number of facility types. The database sits within Verisk's research organization, led by Jay Guin, the company's EVP and Chief Research Officer.
Why this is not really a story about data centers
Insurance underwriting has a pattern: it starts by mapping where value is concentrated, then it starts asking what condition that value is in, not as of the last inspection, but continuously. Sprinkler systems, fire suppression, structural inspections: none of those exist because a building owner found them exciting to install. They exist because an insurer eventually made continuous risk management cheaper than the alternative, either through a lower premium for having it or a much higher premium for not having it.
A database like Verisk's is the first step in that pattern for data centers: know where the concentration is. The natural next step, once that map exists, is to ask what the condition of each node in it looks like on an ongoing basis rather than at the time of the last site visit. That is a question physical monitoring can answer and a static inspection report cannot.
A regulator asking the same question, in a different industry
You do not need to guess whether this pattern shows up elsewhere. It is happening in pipeline safety right now. On September 1, 2026, the US Pipeline and Hazardous Materials Safety Administration published a notice under docket PHMSA-2026-1156 reopening comment on pending rulemakings. Separately, a PHMSA proposed rule published on July 8, 2026 would replace prescriptive, fixed-interval pipeline repair timelines with risk-based criteria driven by inspection technology. The projected saving to operators is roughly $390 million a year.
Read that proposal plainly: PHMSA is proposing to stop telling operators "inspect on this fixed schedule regardless of condition" and start allowing "inspect based on what continuous, technology-driven evidence says the actual risk is." That is a regulator making the same argument this piece is making about insurance: fixed-interval inspection is being replaced by continuous, evidence-based risk assessment, because the fixed-interval approach is both more expensive and less accurate than the alternative once the technology to do better exists.
Why underwriters are harder to argue with
A procurement committee evaluating a camera system can be argued with. It can be told the detection rate is good enough, that the budget does not stretch this year, that the last vendor was fine. An underwriter pricing a policy is doing something different: it is putting a number on risk it is choosing to hold, and it moves that number based on evidence, not on a relationship. Once an underwriter starts asking a facility operator to demonstrate continuous physical monitoring as a condition of a premium, or a lower premium, that question does not go away because a vendor's product roadmap slipped a quarter.
That is the mechanism this piece is describing: not that insurers demand a specific product tomorrow, but that the actuarial logic driving a database like Verisk's, mapping concentration first, asking about continuous condition second, is the same logic that eventually asks whether a facility has any ongoing visibility into what is happening on its floor, not just what happened during the last scheduled walk-through.
What we are not claiming
We want to be exact about what this piece is and is not. Nobody at Verisk has called us. No underwriter has asked a DHI customer to install anything as a condition of coverage. We have no case study here, because there is not one yet. This is a prediction built on a pattern: concentration mapping precedes continuous condition monitoring in every insurance line that has gone through this before, and data center risk looks like it is entering that cycle now, with a regulator making a parallel argument in pipelines at the same time. If we are right, the question "can you show continuous monitoring" arrives at facility operators from a direction most of them are not currently watching. If we are wrong, this ages as a piece that guessed early and missed. Either way, it is a guess, stated as one.