AMI 2.0 is a network decision, not a meter decision

Second-generation metering programs are usually framed as meter replacements, because the meters are the visible asset and the largest line item. The framing is wrong, and it is wrong in a way that costs utilities for decades. The durable decision in an AMI 2.0 program is the communications network — the field-area network that connects the meters and, increasingly, everything else at the edge of the grid. The meter is a device on that network. Devices come and go. Networks persist.

The asset lives don't match

A meter has a service life of roughly fifteen years, and utilities are now living through exactly that cycle: the first-generation fleets deployed in the late 2000s and early 2010s are aging out on schedule. The communications architecture underneath them does not turn over on the same clock. Radio technologies evolve, head-end systems get replaced, but the architectural choice — how devices in the field reach the utility, who owns the intermediate infrastructure, what the coverage model is — spans vendor generations. A utility choosing a field-area network today is choosing the environment in which it will buy meters, line sensors, fault indicators, and automated switches for thirty years or more. Put the two lifetimes side by side and the arithmetic is blunt: a well-chosen communications architecture outlives two full meter generations. The meter being installed this year, and its replacement fifteen years from now, will both live on the network being specified today.

That asymmetry should drive the procurement. Instead, most programs are structured the other way: the meter vendor is selected on meter merits, and the network arrives as whatever that vendor happens to ship. The consequential decision gets made by default, as a side effect of the incidental one.

Three architectures

Three architectures dominate the field, and each locks the buyer in through a different door.

Mesh networks put the intelligence in the endpoints: each meter is a router, and density creates coverage. The economics are attractive in urban and suburban territory, where devices are close enough to relay for one another, and the utility owns the infrastructure outright. The lock-in is protocol-level. Mesh implementations remain substantially proprietary despite years of standards work, which means the network vendor and the device vendor are, in practice, the same decision. Every future edge device either speaks that vendor's mesh or needs its own path home.

Public cellular inverts the model: no field infrastructure to own, coverage wherever the carriers have built, per-device economics that suit sparse rural territory where mesh density never materializes. The lock-in here is the roadmap. The carrier decides when 4G sunsets, what IoT service tiers exist, and what connectivity costs in year twelve. Utilities that lived through earlier network sunsets — forced to replace working communications modules because a carrier retired a technology — understand this is not hypothetical. The device fleet outlives the carrier's interest in the technology it was built on.

Private LTE offers carrier-grade radio under utility control: licensed spectrum, owned infrastructure, one network for AMI, distribution automation, and mobile workforce. The latency and reliability characteristics are the best of the three, and the roadmap belongs to the utility. The lock-in is capital and competence. Spectrum, towers, and core infrastructure demand a scale of investment — and a telecommunications operations capability — that only larger utilities can justify alone, which is why spectrum consortia and shared-network arrangements keep appearing.

None of these is the answer. The answer is knowing which constraints you are accepting, and accepting them on purpose.

What second-generation actually buys

The reason the network deserves this scrutiny is that the second-generation business case does not rest on meter reading. Meter-to-cash was the first generation's case, and it was consumed by the first generation. What justifies reinvestment is what the network enables beyond billing.

Outage sensing at the edge: meters that report loss of voltage in seconds, turning outage management from inference into observation, with restoration verified at the premise rather than assumed at the breaker. DER measurement: interval visibility of export at the connection point, which is the raw material for the feeder-level operational picture distributors now need. Interval flexibility: the measurement and communication substrate for time-varying rates, demand response verification, and eventually settlement of dispatched flexibility.

Each of these makes demands the old meter-reading duty cycle never made — latency, upstream capacity, event-driven traffic, device diversity. Whether the network can meet them is determined by the architecture chosen at procurement, not by anything decided later. A network bought for daily reads will deliver daily reads, and the business case built on everything else will quietly fail in the field.

The procurement implication

The practical conclusion is unglamorous: specify and evaluate the network as its own decision, with its own criteria, even when commercial reality bundles it into a single RFP with the meters. That means network requirements written against the full device roadmap, not the meter count. It means coverage, latency, capacity, and device-onboarding criteria that are scored separately from meter accuracy and installation logistics. It means asking every bidder what happens when a device that is not their meter needs to join the network in year eight — and weighting the answer.

Bundled procurement is often unavoidable. Bundled evaluation is a choice, and it is the choice that decides whether the utility is buying a network with meters on it, or meters with a network problem attached.