Back from holiday, first day of Fall, and what do we have: a couple of stories, either side of the weekend, say something about the contradictory state of the mobile market – that telecoms has become too important to fail, but is not easy enough or valuable enough to pay for itself. Or something like that; that existing regulation and economics are tough, anyway, and that future R&D is being neglected, leaving it ripe for disruption by ‘big tech’. Andrea Donà, networks director at VodafoneThree, has said the UK will be undermined in the AI race because its mobile networks – ranking a woeful 55th in the world for performance, apparently – are creaking under the strain already, and hobbled for future upgrades by bureaucratic red tape. And the revolution will not be tele-phoned in; AI needs new-fangled 5G SA and 6G data networks, which require problematic planning permissions even just for software patches. Donà’s firm has committed £11bn to its 5G network, but is having a hard time just to spend it – it seems.

Meanwhile, telecoms hasn’t been a ‘growth industry’ for a decade. Even if (various earnings-before) EBITDA scores look better lately – Deutsche Telekom, Orange, and Telefónica have posted 7.5%, 5%, and 3.8% rises (over different accounting periods) – it is hardly the double-digit stuff of corporate AI dreams, and the revenue numbers are less exuberant: 4.4%, 3.5% and 1.7%, respectively. It is growth, but not brilliant growth. And the capex story doesn’t show an industry rushing to spend (yet): Deutsche Telekom’s cap-ex fell 6.2% in the quarter; Orange’s fell 2.4% outside Africa and the Middle East; Telefónica’s rose 1.4%. These are mature enterprises getting better at extracting returns from existing assets – increasingly by using software (AI) – rather than a sector that has clearly perceived an AI gold rush, and money to throw at it. The adjacent telco tale about fiber and optical networking looks very different, of course, on the grounds it is already being wired into the AI build-out – as Sean explains here and here.

AT&T’s ‘advanced connectivity’ revenues – a mixture of fiber, fixed wireless, and mobile – grew 5.1% in Q2, with EBITDA up 8%. Verizon has just booked a $1bn dark-fiber DCI deal with Google Cloud, and expects its metro-edge data center refit to ramp in 2027. Meanwhile Lumen, as RCR has reported at some length, has seen its ‘strategic revenue’ rise 14%, NaaS base double to 3,000-plus, and PCF roster expand with $13bn of deals with hyperscalers and frontier firms; it is building 40-odd million new fiber-optic strand miles, and 400G to 90 percent of its cloud on-ramps. Which explains why the telco picture is skewed – between fiber firms connecting clouds and offices, and mobile firms connecting people and robots. There are good lessons here, but mobile operators, particularly outside the US, still have work to do – and time to pass, until AI workloads start to shift out of data centres and fiber routes. Which remains, to an extent, the domain of R&D projects – and brings us to the other story doing the rounds.

Marcus Weldon, former president of Nokia Bell Labs, was vocal last week on LinkedIn about the contraction at Nokia Bell Labs, where the headcount has fallen by “about half” from more than 1,200 researchers at the end of his tenure in 2021. The story was picked up by the mobile trade press, and defended by Nokia. The conversation was also extended on social media – where Sebastian Barros suggested the $1.3tn telecom industry has a $200bn R&D gap, between the $50–80bn (he calculates) it spends and the $260bn (he thinks) it should spend. Weldon pushed back: the problem is not simply that the industry cannot afford the research. During his tenure, Bell Labs’ research spend was 1% of its parent’s revenue, while patent revenue was about 5%, with more than 75% of those patents created by Bell Labs. “The real issue is corporate myopia and lack of ambition or commitment to disruptive innovation in an industry that has become demoralized and devalued by many factors (unfavorable regulation, unfair competition, and unwitting and unending concession of value to hyperscalers).”

Ouch. “The real question then is whether it is too late to reverse this.” Nokia is a funny case – as RCR has written lots. Its strategy illustrates both sides of the argument: the total near-term financial logic to put focus on optical, IP and other AI networking infrastructure, and the weird counter-logic to reduce R&D headcount at Nokia Bell Labs and also (!!!) sell its private (campus) 5G business – which are both about where AI goes next, with robotics and whatever-else. None of which means Nokia has stopped innovating, or that Bell Labs is somehow finished. This industry is good at agreeing on the future, and then building it. It is one of its great strengths – as Barros pointed out. But SpaceX doesn’t need 6G to reinvent connectivity, Nvidia does not need to be a telco to reshape networks, and AI infrastructure firms have very different structures and incentives. All of which puts telcos, critical as they are, at odds with the market evolving on top of them.