Its spindly legs a metallic blur, a robot athlete explodes across the finish line faster than Usain Bolt, crashes into a padded blue wall and promptly falls on its back in what looks like a mock celebration. Others competing at China’s Robot Olympics simply explode. One has some kind of C-3PO-style meltdown after attempting a shoulder press with the lightest barbell ever.
For all the “what’s the point?” questions and mechanical silliness, the recent Beijing event for nerds and humanoid enthusiasts may have got a few telcos thinking about how they can make money from new connected things. That is starting to look more important than ever now that significant parts of the industry appear to have given up on an earlier money-making scheme.
“Given up” was exactly the language used by Timotheus Höttges, the CEO of Deutsche Telekom, back in May to describe the status of a campaign that came to be known as “fair share” or “fair contribution.” It had pitted some of Europe’s biggest telcos against the giant US Internet companies, derided as freeloading “large traffic generators” that clogged up telecom networks with bandwidth-hungry services and yet paid zilch toward their maintenance. A campaign that always seemed like a massive folly now appears dead.
“Telekom has given up on that fair share debate,” Höttges told reporters in May on a call to discuss recent financial results. “We’re not making any headway, and we’re not getting political support either.” Instead, the operator would look to “team up with big Internet companies based on partnership,” he added. “And then we’ll have to make sure that we can offset costs in another way here.”
Customers have all they need
The entire premise was always deeply flawed. For one thing, while it is true that Internet traffic carried on telecom networks has continued to escalate (albeit at a slower rate in the last few years), there is no evident correlation between this growth and telco costs. Telefónica, one of the few big telcos that discloses traffic levels (or that used to until last year), watched the volume of petabytes on its networks soar from 86,591 in 2020 to 159,436 in 2024. But over the same period, its operating expenditure fell from €30.2 billion (US$35 billion) to around €24.8 billion ($28.7 billion). Telefónica also pumped about €5.9 billion ($6.8 billion) into capital expenditure in 2020. In 2024, it spent around €600 million ($695 million) less.
Data from Analysys Mason also shows that telco capital intensity, or capital expenditure as a percentage of sales, is on the slide. By the early 2030s, the consulting and analyst company expects it to have fallen from historical levels of 20% to “just over” 10% in high-income markets. Telcos currently enjoy an abundance of capacity, according to Tom Rebbeck, a research director at Analysys Mason. “We have the capacity with the fiber network and with the mobile network,” he said during a presentation earlier this year. “We’re giving customers all that they need. That’s a huge shift in the market.”
(Source: Telefónica, Light Reading)
The idea that networks are straining and cracking as Internet companies thoughtlessly pile on the petabytes would therefore seem to be very mistaken – even if that story suited the telcos’ fair share agenda. The once-in-a-lifetime job of climbing poles and digging up roads to lay fiber across entire countries has accounted for the bulk of spending on infrastructure rollout. Capacity can subsequently be added through the less costly installation of new electronics, or even a software upgrade.
In mobile, similarly, most investment would be consumed by the building of new sites. But little of this densification is happening. To hit coverage targets, operators instead rely on more advanced network technologies or lowband spectrum freed up by the retirement of older generations.
Thanks to new network software marketed under the letters AI, Ericsson and especially Nokia are also now promising substantial improvements in spectral efficiency on much of the same infrastructure. Spending on the radio access network (RAN) has already fallen from $45 billion in 2022 to $35 billion last year, according to Omdia, a Light Reading sister company. Dell’Oro, another analyst business, expects cumulative RAN expenditure to be 10% to 20% lower in 6G than it was in 5G for the first six years after launch.
Can you spare any change?
Even if all that were different, fair share looked like a telco attempt to charge twice for the same usage. The real large traffic generators are not the Internet companies but operators’ own gigabyte-guzzling customers. If its service charges do not allow a telco to profit, that is a problem the telco industry has created. But few big operators report losses.
Without Amazon, Google, Microsoft and other such content companies, millions of people would see no reason to buy expensive broadband and smartphone services. In the same content vacuum, operators would not have had to invest so much in 5G and fiber rollout. But would a status as providers of basic voice connectivity have suited them better?
The Internet companies could have made a similar case, arguing they should enjoy a share of connectivity revenues as the real drivers of demand. Today, their combined capital expenditure massively exceeds that of telcos across North America and Western Europe. Most goes on the infrastructure for cloud computing and AI services, used by telcos and their customers.
Viewing a big Internet company as a greedy, freeloading user of networks is like saying a successful Hollywood filmmaker is exploiting cinemas. Blockbusters sell more tickets, but they also necessitate investment in seats and new facilities. If a cinema chain in a competitive market could not juggle those factors to make a profit, it would hardly consider demanding payment from the studio.
The difference is that growth in petabytes, unlike growing cinema attendance, has not brought a concomitant increase in sales. But even if telcos would now struggle to charge per gigabyte, nobody forced them to adopt the pricing and business models they did.
Höttges’ gloomy assessment of fair share’s prospects has been accompanied by silence in other parts of the industry. Asked what it thought of his comments, telco lobby group Connect Europe – formerly known as ETNO (European Telecommunications’ Network Operators) – had still not even responded several days later, when this article was published.
As positive as the demise of fair share looks, it leaves the industry short of its revenue-growth story. Instead of charging customers per gigabyte, some telcos are now trying to sell them a supposedly guaranteed higher-quality connection based on network slicing, which reserves capacity for a segment of users. But it is problematic, not least because it implies customers outside the slice are on relatively poor-quality connections.
Could AI be an opportunity? Current data suggests AI services account for a tiny fraction of traffic on access networks. Even if it grew, telcos would have to find a way of charging for additional traffic, something they have previously failed to accomplish. But if AI meant providing connectivity for new objects, there might be an upside.
Last year, Morgan Stanley predicted the number of humanoid robots would reach 1 billion by 2050. At Beijing’s Robot Olympics, Huawei said it had deployed its 5G Advanced “GigaUplink” system to boost connectivity for robots sending data to the network. It would, said the Chinese vendor, provide a “connectivity foundation for humanoid robots to move beyond arenas such as the ‘Ice Ribbon’ sports complex and into everyday life.”
Outside China, many will remain skeptical. But an army of robots cooking dinners, serving drinks and showing their bipedal creators how running is done could be just what telecom needs.
