For Ericsson boss, 'evolutionary' 6G puts end to Gs cycleFor Ericsson boss, 'evolutionary' 6G puts end to Gs cycle

The next mobile standard, likely to appear in 2030, will not be accompanied by the capital expenditure spike of previous generations, says Ericsson's CEO.

Iain Morris, International Editor

January 27, 2025

8 Min Read
Ericsson CEO Börje Ekholm
Börje Ekholm, Ericsson's CEO, says 6G will be an 'evolution of 5G' rather than 'a new type of generation.'(Source: Ericsson)

Almost as old as the mobile phone, the Gs cycle is now midway through its fifth generation. Roughly once a decade, with the encouragement of its suppliers, the global community of telcos changes network technology, effectively replacing its engines. At first, the upgrades were obvious to customers and easily marketable. A shift from 2G to 3G brought Internet services (belatedly) to what had previously been sophisticated walkie talkies. With 4G came a speedier Internet experience and video. But if 5G's attraction remains less discernible to users, its largest vendor outside China has now surprisingly suggested it may be the last of the big changes.

In comments that might have struck some observers as counterintuitive, Ericsson has now joined a chorus of voices downplaying the likelihood 6G will bring anything radically different to mobile networks. "6G, if you think about it as a technology, will probably get introduced around 2030," said Börje Ekholm, Ericsson's CEO, on his quarterly call with analysts last week. "What is more important is that 6G is actually an evolution of 5G and we should think of 6G not as a new type of generation where you upgraded from 3G to 4G or 4G to 5G and you needed also to upgrade the whole network."

Big kit vendors have previously earned most of their sales when a new G is first introduced and telco money flows into it. Revenues at Ericsson's main networks business rose a tenth on a like-for-like basis in 2020, as operators spent on 5G rollout, and by another 7% the following year. But the telco appetite for investment quickly faded before the new generation had been as widely deployed as earlier standards, largely because operators did not witness similar sales growth to Ericsson after initial rollouts.

Ericsson, of course, is not about to give up on 5G some five years before its successor is due to appear. Yet Ekholm acknowledged a recent slowdown in the rate of mobile data traffic growth, and Ericsson is conscious that many of its customers show little enthusiasm for expensive future hardware upgrades. In response, it is putting far more emphasis on software and what it calls the "programmable" network.

Push the button

Instead of selling a complete set of mobile network gear for a particular site, Ericsson would activate and charge for new features on top of equipment that has already been deployed. In principle, this could even come from a different supplier. A component of a newish $14 billion contract with AT&T is supposedly a cloud radio access network (RAN) that replaces Ericsson's 5G-dedicated appliances with Dell servers and general-purpose Intel chips. Some of the radio units powered by Ericsson's software are to come from Fujitsu and Mavenir, says AT&T. Ericsson talks about shifting from a "vertical" model of selling a full product set for a site to a "horizontal" approach where its software is one layer running through the whole network.

"5G and 5G Advanced even more so will be cloud-based and will have new types of principles and be structured in a horizontal way and 6G will be an evolution of that," explained Ekholm. "If you are going to move into 6G, you actually need to have the 5G network and the architecture of 5G built out already and then you can benefit from adding the new frequencies and new capabilities that 6G will give." Far from being a "normal new generation," 6G will end the peaks and troughs of investment, he said. "That means for us we are not going to see those cycles in the future."

The evident hope is that traffic growth will come to matter less than service differentiation facilitated by Ericsson's network smarts. A telco might, for instance, want to offer a high-capacity service for someone video-streaming from a music concert, or a low-latency guarantee for a factory owner. New artificial intelligence (AI) applications could make these features even more valuable, according to Ekholm. Software developers will be able to access them via application programming interfaces, standardized across the industry. A ballyhooed 5G technology called network slicing would reserve specific advanced features for particular services or customers.

Ekholm is now guiding for a change in the financial profile of Ericsson with this shift. "We are entering a new phase here where our customers will look different and we'll look different, but that is a positive with less hardware components and more software content in our sales," he said. The fattening of profit margins that accompanied last year's revenue drop of 6% was attributed partly to a higher share of software sales. Within the networks business, where revenues fell 8% in 2024, Ericsson's gross margin rose seven percentage points, to 46.6%.

Not giving up on hardware

None of this, however, means Ericsson will suddenly quit the hardware game. For all Ekholm's talk of cloud-based 5G, the "virtualization" of the RAN that entails moving from application-specific integrated circuits (ASICs) to general-purpose hardware has not rapidly advanced. In 2023, it accounted for about a tenth of all baseband, the compute side of RAN, according to Light Reading sister company Omdia, which expects its share to double by 2028. Purpose-built RAN is forecast to remain the default three or four years from now.

"Hardware will continue to be an important part of what we do," said Fredrik Jejdling, the head of Ericsson's networks business group, on a call with Light Reading after earnings were published last week. "We just launched our 6672 baseband now with close to 100 billion transistors across two ASICs in there and it is a very high-performing entity. Customers choose to do different things across different technologies and that's why, in essence, we offer optionality." ASICs developed by Ericsson for its radio units, moreover, are becoming "even more specialized," said Jejdling.

On the hardware side, the other big uncertainty is whether Nvidia might succeed the currently beleaguered Intel as the main general-purpose platform of the future. Since last year, it has been pitching a concept it calls AI RAN that would substitute an Nvidia superchip – formed of a graphical processing unit (GPU) and Arm-based central processing unit (CPU) – for one of Intel's x86-based CPUs in virtual RAN or the purpose-built option of an Ericsson ASIC. One prominent telco executive, speaking on condition of anonymity, thinks the capacity limits of foundries, increasingly dedicated to GPU production, could force RAN chipmakers to pivot away from customized chips.

Jejdling downplays that concern and insists the industry is in a stronger supply chain position than it was during the pandemic. "We are still working with safeguarding and securing that we can get demand under difficult circumstances as well," he said. Nevertheless, in the interests of efficiency, Ericsson is determined to "virtualize" software to the fullest extent possible, allowing it to run the same code on a multitude of different hardware platforms. "We have made the decision internally that everything will become virtualized, and we'll give the customer the option whether that is deployed on our own compute portfolio or another compute portfolio," said Jejdling.

Is this achievable? At Layer 1, the category of RAN software that is hungriest for IT resources, there tends to be a lot of hardware dependency, according to various experts. Ericsson still appears to have no commercial hardware alternatives to Intel for its virtual RAN software. And the compute unified device architecture (CUDA) platform used by Nvidia would necessitate some rewriting of RAN software by developers, the chipmaker itself has admitted. "They do have to reprofile into CUDA," said Soma Velayutham, the general manager of Nvidia's AI and telecom business, during an interview with Light Reading last month. "That is an effort."

The death of globalization

Another cost challenge for Ericsson ahead of 6G's arrival is the fragmentation of what has previously been a globalized industry. "I think the whole world is moving from a cost-optimized supply chain to resilience," said Ekholm on the quarterly earnings call. "You need to factor in resilience in the supply chain and that is why we built a US factory, and we are investing to increase capacity in the US as well."

The natural implication is that any move from a global supply chain, with production heavily concentrated in Asia, to one dispersed across regions and countries would drive up costs. Yet Ericsson attributed the fattening of its gross margin partly to "supply chain efficiency actions" in its earnings report.

"What we're talking about now is a model where we work much more proactively with our customers on demand, sensing and shaping, talking about our medium- and long-term portfolio," said Jejdling. The goal, he explained, involves "working on stabilizing their demand and our ability to supply over a shorter period, thereby taking away a lot of logistics costs, taking away a lot of inefficiencies in the system, and allowing ourselves to build a lot less inventory in the system as well."

A full retreat from globalization could "potentially" hurt the industry's economies of scale, Jejdling admits. But he agrees that automation is helping Ericsson to mitigate some of the impact. A state-of-the-art factory in Texas runs with a fraction of the people Ericsson would previously have needed and a facility in Estonia is "largely" the same, according to the Ericsson executive.

All that has undoubtedly supported cost reduction and a shrinking of the Ericsson workforce. Some 9,400 internal and external (contractor) jobs were cut last year, and internal headcount has fallen from more than 111,000 in late 2016, just before Ekholm took charge, to about 94,000 now, despite the $6.2 billion takeover of Vonage and $1.1 billion acquisition of Cradlepoint during his tenure. Annual revenues per employee, at today's exchange rate, have risen from $182,304 to $240,141 over this period.

What has never been cut under Ekholm is the annual amount dedicated to expenditure on research and development (R&D). It rose 6% last year, to 53.5 billion Swedish kronor (US$4.9 billion) and has grown from as little as SEK31.6 billion ($2.9 billion) in 2016, after Hans Vestberg, Ekholm's predecessor in the CEO job, prioritized diversification outside mobile networks over core R&D. Whatever else changes, Ericsson seems unlikely to repeat that mistake.

Read more about:

AIHeadcountOmdia

About the Author

Iain Morris

International Editor, Light Reading

Iain Morris joined Light Reading as News Editor at the start of 2015 -- and we mean, right at the start. His friends and family were still singing Auld Lang Syne as Iain started sourcing New Year's Eve UK mobile network congestion statistics. Prior to boosting Light Reading's UK-based editorial team numbers (he is based in London, south of the river), Iain was a successful freelance writer and editor who had been covering the telecoms sector for the past 15 years. His work has appeared in publications including The Economist (classy!) and The Observer, besides a variety of trade and business journals. He was previously the lead telecoms analyst for the Economist Intelligence Unit, and before that worked as a features editor at Telecommunications magazine. Iain started out in telecoms as an editor at consulting and market-research company Analysys (now Analysys Mason).

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like