Every so often a development in an adjacent field quietly redraws the boundaries of your own. Auto insurance may be approaching one of those moments, and it is coming not from InsurTech, but from wireless engineering.
Over the weekend I read the May 2026 IEEE Spectrum cover story, "Teaching Radio Waves to Compute," by Ana I. Pérez-Neira and Alphan Sahin. The subject is technical, but the implication is strategic.
The idea is Over-the-Air Computation (OAC). Today, connected devices take turns transmitting, each waiting for its own slot. OAC inverts that, allowing many devices to transmit at once and using the physical superposition of radio signals to compute an aggregate result directly at the receiver. In plain terms, the network stops merely carrying data and begins summarizing it. The research is early, and performance still depends heavily on synchronization and network design. But the direction of travel is what matters.
Set that against three shifts already underway in our industry.
First, risk is migrating from the driver to the system. As automated systems take control under defined conditions, some manufacturers have signaled willingness to accept responsibility, though liability frameworks remain dependent on jurisdiction and product design. Personal auto, one of the largest segments in U.S. Property & Casualty (P&C), could gradually shift from pricing how a person drives toward how a product, software stack, or fleet performs. That is a fundamentally different underwriting discipline, and it calls for different data.
Second, the data is outgrowing the way we use it. Connected and autonomous vehicles (CAVs) generate far more telemetry than today's usage-based programs. Yet underwriting models rarely need every sensor event; they need patterns: how a fleet behaves at a difficult intersection, how often automation disengages, where braking anomalies cluster. Anyone who has run a telematics program knows the gap: vast data moving through pipelines, while only a fraction ever shapes a pricing or claims decision.
Third, privacy is becoming a source of advantage, not just obligation. The more raw data an insurer moves and stores, the larger its regulatory and reputational exposure.
This is where OAC becomes interesting. If aggregate risk signals can reach insurers while far less raw data leaves the vehicle, the economics of telematics change: lower data movement, lower latency, a stronger privacy posture. Combined with common standards and governance, it could also help compare aggregate system performance across fleets and platforms: a capability the market lacks today, and one that grows in value as we begin underwriting manufacturers and operators rather than individuals.
If these trends mature, new product concepts become easier to imagine: zone-rated policies priced on real-time conditions, parametric coverage triggered by fleet-level thresholds, vehicle-to-everything (V2X) handoff coverage for the Level 3 gray zone, and shared-liability structures spanning owners, operators, OEMs, and software providers.
None of this is a 2026 story. OAC still needs advances in synchronization, standards, OEM participation, and regulation before it scales. The value in raising it now is not to predict a timeline, but to ask a better question: will our industry keep moving ever-larger volumes of raw data around, or will the networks themselves grow smart enough to deliver only the signals underwriting and claims actually use?
My instinct is that the firms thinking about this now will be better positioned when the technology is ready. Curious where others believe it lands first.
