The networking and AI match is a good one. AI is advancing rapidly (DNNs, GNNs, LLMs, etc.), and telecom networks produce vast quantities of telemetry data and generate complexity, pushing the limits of classic networking expertise. If, as Jensen Huang from Nvidia says, AI fundamentally changes computing, telecom must adopt this technology.
In the medium term, two primary opportunities exist: AI in customer and business operations (chatbots, yield management, etc.) and AI in network operations (optimization, assurance, etc.). In a Goldilocks scenario, the two link customer service to an improved network experience. In each case, AI/ML is applied to existing architecture or workflow to generate a small gain on the previous process – let’s say, for argument, a 10% gain. On their own, these gains are helpful, if not game-changing. Like in professional sports, however, marginal gains add up to a competitive advantage.
The Race is Forever Changed
In this sense, we can liken AI for telco to the Alphafly running shoe that enabled athletes to knock serious amounts of time out of long-distance records. A marathon is still a marathon, but the race has changed forever.
Once everyone gets the same shoes, the field levels out again. The corollary in telco is that every operator will have to work with AI simply to stay competitive. In this analysis, there’s plenty of work to keep operators and their vendors busy – most of the gains, operators will bank in efficiency improvements. Still, hopefully, customers should get better service as well. The French have an appropriate saying for this outcome: “Plus ça change, plus c’est la même chose.”[Author’s note: too high brow for Light Reading?] But this will only be the start of a journey. The longer-term opportunity for AI in the service provider sector is much greater.