One of the first things Börje Ekholm did on becoming Ericsson's boss in early 2017 was to prioritize investment in mobile, a field Hans Vestberg, his predecessor, had oddly allowed to lie fallow as big Chinese rival Huawei was busily planting crops. Other assets were sold as Ericsson, in investor parlance, "doubled down" on the core business. Spurred on by a US and European backlash against Huawei, Ericsson rediscovered its mobile mojo and reclaimed lost ground. But today it looks less diversified than ever, a one-trick 5G pony. That's not necessarily a good look in a stagnant 5G market.
Nokia, meanwhile, had gone the other way, expanding into fixed, Internet Protocol (IP) and optical technologies with its €15.6 billion (US$16.1 billion, at today's exchange rate) takeover of Alcatel-Lucent in 2016. The difficulty of integrating its former rival's mobile assets with Nokia's was partly blamed for a subsequent stumble in 5G, one that seemingly cost the Finnish vendor a lucrative contract with Verizon and some global market share. Diversification also didn't work out as then-CEO Rajeev Suri had planned. Customers don't want to buy "end-to-end" bundles of core, fixed and mobile technologies, said Pekka Lundmark shortly after replacing Suri.
But Nokia's more diverse portfolio has given the company multiple potential fields to harvest when other soil is barren. While Ericsson is bogged down in 5G, Nokia is out threshing in data centers and artificial intelligence, a connectivity market worth about €20 billion ($20.7 billion) a year to suppliers of IP and optical technologies, estimates Lundmark. At the network infrastructure unit housing those assets, sales rose nearly a fifth year-over-year for the final quarter of 2024, to more than €2 billion ($2.1 billion). The hiring of Justin Hotard, who currently runs Intel's data center and AI business, to succeed Lundmark in April is the latest sign of Nokia's keen interest in this market.
Could it, then, do an Ericsson, quitting mobile to "double down" on AI data centers? Ever since Nokia lost yet another big contract in the US, where vendors are thought to generate a healthy share of their profits, there has been industry chatter about the possibility of a mobile exit. That recent setback came in late 2023, when AT&T stunned observers with its decision to replace Nokia, then serving about a third of its sites, with Ericsson, already active across the other two thirds. Hurt by the loss of such a big customer, Nokia suffered a 21% drop in mobile revenues last year, to around €7.7 billion ($8 billion). Nokia's organizational cuts have since then torn into mobile.
The Greenland of network connectivity
Several ideas have been floated. In August, Samsung was rumored to have held talks about an acquisition of Nokia's mobile assets. Separately, a well-connected industry source who spoke on condition of anonymity thinks the US government will engineer a takeover, viewing Nokia, like Greenland, as a strategic asset it currently lacks. Attempts to cultivate sustainable American alternatives through the concept of open radio access network (RAN) technology, which allows suppliers of specific parts to be joined together at the same mobile site, have failed. In the age of Trump, anything is possible. But a deal would in many ways be counterintuitive and inadvisable.
From a geopolitical perspective, it would, for a start, rob Europe of a big chunk of one of its few technology successes. In the current environment of protectionism and nationalistic chest-thumping, it could work out nastily for European jobs and threaten Europe's role in future mobile research. There would be inevitable consequences for academia and startups in the sector. It would mark a continuation of Europe's economic and technological decline. Further US encroachment is unlikely to meet the approval of Europe's telcos, either. One goal of open RAN was to produce more local alternatives. If Europe cedes control of a main existing vendor, telcos will have even less choice within Europe.
Is this to overstate the importance of Nokia in mobile? Absolutely not. Research by Omdia, a Light Reading sister company, gave it a 19.5% share of the RAN market in 2023, up from 17.8% in 2022. Only Huawei and Ericsson are bigger, and the Chinese company is now politically unacceptable in many countries. No other vendor outside China comes close. Samsung is the next biggest, and its share of the market in 2023 was just 6.1%, down from 7.6% the year before.
Despite the damage caused by AT&T, Nokia's mobile business is also much healthier now than it was at the time of the Verizon loss. It has improved its software and expanded its base of chip suppliers, adding Marvell Technology and Broadcom alongside Intel, from which it has slowly retreated in the RAN. Indeed, it is apparently now far less dependent than Ericsson is on the troubled x86 chipmaker.
Finns can be fans of fins and fans
Why, then, was it ditched by AT&T? One theory is that AT&T was unhappy with Nokia's reliance on fans and active cooling – as opposed to fins (not Finns) and passive cooling – in its radio units. Yet Nokia insists it caters to both varieties, fans and fins, in its latest 5G products. If this were really the issue, why have telcos in other parts of the world not similarly decamped? Orange, a cost-conscious stickler for quality, has just renewed a 5G deal with Nokia covering parts of France.
No, the likelier explanation – lent weight by the statements of AT&T, Ericsson and Nokia – is that AT&T, for efficiency reasons, wanted a single provider of service management and orchestration. It also probably wanted a single provider of RAN compute (baseband). Ericsson, besides being highly regarded for its 5G technologies, had more of the AT&T footprint. That made it the natural choice.
Analysts have bought AT&T's pitch about this being an open RAN move. Yet Ericsson admits it is not fully compliant with specifications (its purpose-built baseband will not support open RAN specs for massive MIMO, an advanced 5G technology). That may restrict the opportunity for others to slot in. And some application developers are already blocked by Ericsson's hostility toward the concept of xApps, which would allow third-party software to run in near real time on an Ericsson platform. Fujitsu and Mavenir have been named as potential contributors of radio units, but they will probably be limited to small cells and less traffic-heavy sites.
Perversely, the moves by AT&T and Verizon, which profess enthusiasm for open RAN and diversification, have made the US look more like an Ericsson empire where a few vassals are permitted to roam. Although Samsung was the main beneficiary of Nokia's loss at Verizon, the South Korean company has made no other inroads with the country's main network operators. In the absence of viable new 5G suppliers, and with Chinese vendors increasingly unwelcome, a weakened Nokia looks bad for everyone.
No country for old RAN
It is hard to imagine what the US, whose big telcos have seemed determined to squeeze Nokia out of mobile, would do any differently from the Finnish company as the owner of its mobile assets. There is no US company with which Nokia could be merged to produce a 5G titan, its margins boosted by the combination of portfolios and pruning of duplicate resources. The aggregate share of all RAN vendors outside the top seven, none of which is American, was just 2.7% in 2023, according to Omdia. Meaningful consolidation has already happened. Nokia itself is a product of mergers and acquisitions involving Motorola, Nortel and Panasonic, besides Siemens (the former Nokia Siemens Networks) and Alcatel-Lucent (previously separate companies).
A US owner, government or otherwise, could always invest more in research and development (R&D) than Nokia has done. The more 5G-focused Ericsson outspent Nokia last year by roughly $200 million, pumping $4.9 billion into R&D, up from just $2.9 billion during Vestberg's final year as CEO. Annual R&D spending by Nokia has actually fallen by $200 million over this same period.
But is this really the problem, and what would the US owner gain from an increase? RAN product sales globally shrank $5 billion in 2023, to about $40 billion, according to Omdia. They are likely to have fallen by the same amount last year, based on the midpoint of an Omdia forecast issued in late 2024. Telcos have been slashing 5G budgets in response to higher costs and their inability to grow sales. Any spending is likelier to go on AI these days than it is on 5G.
What's more, swaps are unusual, especially on the scale of AT&T's. The sheer expense of a rollout means operators tend to stick with a RAN vendor until its products have fully depreciated. Open RAN does not change this because most of the cost is in one specific part – the radio unit. Outspending Nokia in 2023, when Ericsson had a budget of $4.65 billion compared with Nokia's $4.5 billion, did not translate into market share gains for the Swedish vendor. In fact, Ericsson's market share fell from 25.7% the year before to 24.3%, according to Omdia.
The RAN industry, meanwhile, can expect no growth over the next five years, says Dell'Oro, another market research company. Why this outlook would attract the US, other than for geopolitical reasons, is unclear. But if Nokia's shareholders were tempted by a generous offer to walk away from what appears to be a moribund market, they should think very carefully.
Hotard in the hot seat
The first thing to note is that Nokia remains profitable – even apparently at its shrinking mobile business. At group level, it beat Ericsson on core profitability measures last year, finished 2024 with €4.9 billion ($5.1 billion) in net cash and proposed an increase in dividends. Thanks to network infrastructure, its sales also grew a tenth year-over-year for the final quarter, to just less than €6 billion ($6.2 billion), while Ericsson's ticked up just 1%. In the last year, Nokia's share price has risen almost 40%. This is not a company that looks endangered by circumstances.
The RAN market, judging by Dell'Oro's forecasts, is also now past the inventory corrections and spending cuts that have hammered 5G vendors in the last couple of years. With Trump back in the White House, the backlash against Chinese vendors may gather momentum. For operators under pressure to switch, Nokia is one of the few viable alternatives.
Most importantly of all, the gloominess that surrounds today's RAN market does not mean it will always be stagnant. Techniques such as network slicing, the exposure of new 5G features to software developers through application programming interfaces (APIs) and the further extension of 5G connectivity into the business market could all spur mobile growth in the latter half of the decade.
There is even a possibility the AI boom will lift 5G. In what giant chipmaker Nvidia has dubbed an AI RAN, a telco would run network software on AI chips hosted in edge data centers and sell inference-as-a-service on the spare capacity. Nvidia reckons any telco could earn $5 over a five-year period for every $1 it invests. That lofty estimate has been ridiculed by some analysts. But many industry participants will take encouragement from any talk of growth.
Hotard's background in data centers and AI has obviously fueled the speculation about a mobile exit for Nokia. But when he was introduced on stage to analysts in Helsinki this week, one of his most interesting albeit overlooked comments alluded to this possible intersection between AI and mobile. "We can focus on core network infrastructure," he said. "But if we believe that we will have multiple ways to AI in the future, I think there will be opportunities in mobile networks as well."
Selling its mobile arm would certainly make Nokia look as undiversified as Ericsson, leaving the Nordic countries with network infrastructure and 5G specialists that rarely meet in tenders. Strengthened by its imminent $2.1 billion takeover of Infinera, a US optical equipment specialist, Nokia would be Europe's sole Tier 1 vendor of AI data center connectivity products at a time when the market is booming. As Ericsson can undoubtedly appreciate, that's all very well until the excitement shifts elsewhere.