- Telcos may be shifting AI costs rather than cutting them
- AI token costs could pressure telco budgets as higher usage offsets cheaper models and automation gains, Bain warned
- Telcos need workflow redesign to make AI economics work, measuring cost per outcome — not just cost per token
Telcos have been touting AI as a cost saver, using it to justify thousands of layoffs. But while the humans are disappearing, the costs aren’t always following suit. New research from Bain & Company warned that telcos risk falling into an opex trap as AI costs grow and could end up facing total costs that are higher – not lower – than where they started.
“The risk is a dangerous cost-creep scenario: higher operating expenses without proportional gains in productivity, customer experience, or growth,” Bain’s team wrote.
The firm outlined two potential outcomes for telcos. In the first scenario, telcos are able to replace 20-30% of their traditional costs with AI-related expenses around agents, tokens and data. In the second (dubbed the “cost creep” model, telcos fail to cut traditional costs and those AI expenditures simply add 20-30% to their overall spending.
It’s worth noting that neither scenario shows an overall reduction in costs: they either stay the same or go up.
The scenarios outlined are particularly interesting given Gartner recently predicted 30% of employees laid off due AI will need to be rehired by 2030, likely at a significantly higher cost.
“Business and IT executives who use AI primarily as a tool for cost cutting risk making reductions that are too deep and too soon, affecting their ability to innovate their business model and compete in new markets as AI continues to mature,” Gartner VP Analyst Tori Paulman said.
Telco opex savings under scrutiny
Bain’s warning is consistent with one issued by GSMA Director of AI Technologies Louis Powell in a recent interview with Fierce.
“We are very concerned with people or operators adopting AI to try to reduce opex, but all they're doing is just growing their cloud costs 10x, and they feel they're doing a good job, but they're just really moving their budgets around,” he said. “That is really what we should be doing, is making sure the economics works, the scale, and not trying to solve the fanciest use case that sounds very cool with the most advanced models.”
Indeed, some operators are trying to avoid the cost trap. For instance, AT&T recently detailed its efforts to use open models as part of a “tokenomics” strategy designed to allow it to keep costs under control.
But Bain noted that even cheaper models can come with big bills. “Employees discover new uses, power users consume tokens at scale (especially in network operations and customer care, given high-volume workflows), tasks become more complex, and teams shift to newer models—whose more complex reasoning chains consume more tokens per request—instead of picking the most cost-efficient model for the task,” they wrote.
Bain argued a mindset shift is required to avoid falling into AI opex quicksand.
“For AI, the meaningful unit of economics isn’t cost per token. It’s cost per resolved customer issue, network incident, proposal generated, or software release,” they added.
The worst thing operators can do, Bain concluded, is treat AI as a bolt-on tool. This approach creates a scenario where people continue executing legacy workflows while AI tools handle isolated tasks around the edges. And that’s a recipe for cost creep.
Read more about telco AI and token costs on Fierce Network
Open models are driving AT&T’s AI ‘tokenomics’ strategy