Why AI-RAN is essential here: sensor fusion is real-time AI inference, and a drone won’t wait for data to travel to a distant cloud. This enables low latency as well as data sovereignty and cybersecurity. AI-RAN brings GPU computing to the radio access network, allowing the fusion to happen right where the data originates. In defense applications, GPU power consumption and unit cost matter far less than performance and keeping data local, which eliminates the main objection operators have against AI-RAN. And the programmability of GPUs is valuable here as well: improvements can be delivered via software updates without the need for new hardware.
The size of the contract is, of course, practically negligible for Nokia, but funding from the DoW opens the door for AI-RAN into defense budgets rather than relying solely on operator capex.