Removed from being a divisive barrier, the presidents’ wall at Bell Labs’ headquarters in Murray Hill commemorated the 15 leaders of the long-lasting US analysis group till earlier this yr, when the position was quietly scrapped.
Having celebrated its centenary in 2025, Bell Labs stays a colossus within the historical past of telecom and tech, a pioneer of the primary telephony techniques, inventor of the transistor and developer of laptop programming languages nonetheless used now. However the removing of president as a job title, and the low-key appointment of Guru Parulkar, a former Intel govt, as self-described Bell Labs “head” a couple of weeks in the past, appear to replicate a current dimming of its prospects below as we speak’s administration.
Named after well-known phone inventor Alexander Graham Bell, Bell Labs began life as a part of the Bell System monopoly that was in the end damaged up by American trustbusters. Between 1984 and 1996, it operated as a subsidiary of AT&T earlier than most of it was spun off into Lucent Applied sciences. After that firm’s $13.4 billion mega merger with France’s Alcatel, Bell Labs turned a unit of Alcatel-Lucent in 2006. A decade later, additional consolidation and one other €15.6 billion (US$18.1 billion, at as we speak’s alternate price) deal introduced it below the possession of Finland’s Nokia, the place it’s now falling sufferer to cuts.
These look unprecedented this century and lift questions on Nokia’s assurance that Bell Labs types a vital a part of its analysis actions. After Nokia’s takeover, Bell Labs was understood to have employed about 750 individuals in analysis together with between 250 and 300 in requirements. By 2024, the quantity in analysis had dropped to precisely 681, based on a number of sources with data of the matter. Earlier than the top of subsequent yr, Nokia goals to scale back the analysis headcount at Bell Labs to simply 408, stated these sources.
This is able to clearly equate to a discount of about 46% in contrast with the determine lower than six years in the past. It would shrink the variety of devoted analysis facilities from 19 to 6, based on an inside slide obtained by Gentle Studying and verified individually. Bell Labs’ community techniques and safety analysis (NSSR) lab, which had operated in Europe and the US, is one exercise that has already been discontinued, stated sources.
Nokia responded with the next electronic mail assertion in full: “The AI supercycle presents a defining alternative to rethink the position of networks in an AI-driven world. With this ambition in thoughts, Nokia is bringing its analysis groups again collectively as one unified Bell Labs, with a sharper concentrate on the deep applied sciences that can form the way forward for networks and AI. By concentrating world-class experience on the highest-impact priorities, Nokia will strengthen its skill to translate pioneering analysis into applied sciences with real-world functions and industrial influence.”
It added: “This strategy builds on the targeted, related analysis mannequin that enabled Bell Labs to ship among the most consequential breakthroughs in trendy know-how, from the transistor to the UNIX working system. It aligns with the technique the corporate set out at Capital Markets Day and reinforces Bell Labs’ central position in Nokia’s know-how management and long-term innovation agenda.”
Included beneath this paragraph (the marginally blurred picture was a characteristic of the unique picture), Nokia’s inside slide presents the most recent restructure as a streamlining and “renewal” that can connect sources to what’s described as a “targeted strategic portfolio.” However certainly one of Bell Labs’ former presidents is each unimpressed and deeply nervous in regards to the implications.
Company amnesia
Marcus Weldon ran Bell Labs between 2013 and 2021 and was additionally Nokia’s chief know-how officer for many of that interval. In late August, he made a few of his emotions in regards to the newest strikes clear in a LinkedIn submit that was picked up by a number of titles, together with this one. Weldon, who amongst different issues at present advises our father or mother firm Informa Group on its AI technique, has now shared his considerations about the way forward for Bell Labs with Gentle Studying.
They embrace an unshakeable view that Nokia undervalues the contribution of Bell Labs and doesn’t recognize its significance to the larger enterprise. “I can record tens of improvements that remodel the Nokia companies, so there are two levels of folly,” stated Weldon. “One will not be recognizing the patents produced as materials, worthwhile worth from build-up. Secondly, this concept that nothing of worth has been executed in 15 years is simply ridiculous.”
He blames a company “amnesia” for Nokia’s incapacity to remember the origins of know-how worth, declaring that Bell Labs’ improvements can take a few years to percolate by means of the enterprise. Achievements throughout his personal time as a Bell Labs president included work on digital radio entry community (RAN) know-how and even the AI-RAN idea that’s now related to Nvidia, stated Weldon. “By the point it turns into a part of the enterprise, nobody remembers the place it got here from.”
That type of amnesia will not be new and meant Weldon needed to frequently struggle for Bell Labs and champion its significance to senior administration at Nokia, he stated. Due to these efforts, he was in a position to defend an annual spending funds of about $250 million, equaling roughly 1% of Nokia revenues. However the newest cuts, which he describes as a “top-down dismissal of the worth that Bell Labs gives,” would cut back that determine to simply 0.4%, by his calculation.
“That’s surprising as a result of that has by no means, ever been the case within the 100-year historical past of Bell Labs,” he stated. “Even when among the constituent corporations like Lucent and Alcatel had been struggling, they didn’t reduce Bell Labs funding beneath that $200 million, $250 million stage.”
Apart from constructing merchandise primarily based on Bell Labs’ analysis, Nokia additionally monetizes it immediately by means of the licensing of patents by its Applied sciences division, not too long ago subsumed into a brand new Cellular Infrastructure (MI) enterprise group. Throughout Weldon’s time, he stated, Applied sciences would generate annual revenues of about $1 billion on that $250 million in prices – then booked below “Group Widespread” bills – making it a extremely worthwhile enterprise primarily based on patents alone.
The restructuring that created MI makes any comparability between then and now a lot tougher. Earlier than MI got here into existence this yr, Applied sciences booked full yr gross sales for 2025 of about €1.5 billion ($1.7 billion) with an working margin of greater than 70%. Inside MI, revenues from the licensing of know-how requirements for the primary half of 2026 got here to €792 million ($920 million), massively exceeding the €532 million ($618 million) that MI reported in working revenue. “I feel the thought is to deliver it again in in order that it may well bolster the cellular enterprise,” stated Weldon on the inclusion of Applied sciences in MI.
The ramifications are troubling. For the primary half, your complete Nokia enterprise managed an working revenue of €735 million ($854 million) on a “comparable” foundation, Nokia’s most popular metric. If the margins for Applied sciences had been roughly the identical as in 2025, Nokia would have owed three quarters of its working revenue to licensing. And if the worth of most patents originates with Bell Labs, Weldon’s obvious alarm is comprehensible.
Cuts at Bell Labs might even have a a lot larger influence on Nokia’s involvement in cellular than on its fixed-line actions, judging by Weldon’s evaluation. That’s partly as a result of requirements growth is a much more vital effort in cellular, he defined. He additionally dismisses the suggestion that Nokia would have the ability to offset the cuts at Bell Labs by persevering with to develop the funding in R&D that occurs elsewhere within the group.
Throughout Nokia, R&D expenditure rose 6% year-over-year for the primary half, to about €2.3 billion ($2.7 billion), after rising 7% in 2025, to €4.6 billion ($5.3 billion). Since February final yr, it has included the investments made by Infinera, the optical tools specialist that Nokia acquired for $2.3 billion. Infinera’s R&D invoice got here to about $300 million in 2024, earlier than the takeover.
Whereas Nokia will surely have groups of individuals concerned in requirements growth, they’re very totally different from the Bell Labs workers liable for core analysis, based on Weldon. “They’re very technical individuals, however they are not doing breakthrough analysis. They’re type of representing the analysis to the requirements group,” he stated. “I might say that quantity mixed with the analysis quantity is a vital barometer for profitable management in future wi-fi applied sciences.”
All of the eggs in a single basket
Weldon additionally believes that Nokia’s current tie-up with Nvidia looms giant in its shapeshifting. Final yr, the Finnish firm notably accepted a $1 billion funding from the large chipmaker and introduced plans to develop future 5G and 6G community merchandise utilizing Nvidia’s graphics processing models (GPUs) quite than conventional application-specific built-in circuits (ASICs).
“The AI-RAN that Pallavi talks about has virtually no Bell Labs involvement based on my sources, regardless of Bell Labs inventing AI-RAN and digital RAN,” stated Weldon, referring to Pallavi Mahajan, Nokia’s present chief know-how officer. “I feel there’s an over-dependence on Nvidia, and advocacy for Nvidia as being the place the place that shall be developed, not leveraging Bell Labs’ improvements.”
He’s certainly one of quite a few specialists to precise concern about that relationship. “I’m not arguing towards AI-RAN – I am saying the danger of not having a companion ASIC technique is identical danger that Nokia took after it acquired Alcatel-Lucent,” he stated. “Nokia didn’t have an ASIC technique. It had an Intel-centric technique, and Intel failed them famously as a result of 10-nanometer failure.”
That entire affair was subsequently coated in some element by Gentle Studying. The chance now, based on Weldon, is clearly not that GPUs will not exist. It’s, in his opinion, that Nokia is placing “all its eggs in a single basket” for Nvidia’s commercially unproven, GPU-based AI-RAN know-how.
Information of cuts at Bell Labs sits within the context of an ongoing and far larger headcount discount at Nokia. Having had as many as 103,000 staff in 2018, it completed final yr with 74,100, excluding Infinera, which employed 3,418 individuals in December 2024, two months earlier than it was acquired. Underneath restructuring plans initiated by Pekka Lundmark, Nokia’s earlier CEO, one other 4,000 jobs had been as a consequence of be reduce earlier than the top of 2026. However current selections might make the workforce even smaller.
In July, when Nokia reported first-half outcomes, present CEO Justin Hotard upped this yr’s restructuring prices from €250 million ($290 million) to €800 million ($929 million), with the rise linked to further deliberate layoffs in China and Europe. A couple of weeks later, Nokia confirmed it could shut a analysis facility in Hangzhou, at the price of 1,600 jobs.
Messaging exchanges between staff in China, obtained by Gentle Studying, point out websites in Beijing, Chengdu, Qingdao and Shanghai may additionally be shuttered. That would have a significant influence on headcount, with Nokia nonetheless using round 7,200 individuals in China, Hong Kong and Taiwan final yr.
“Because the market share has diminished, I feel that is really a defensible resolution,” stated Weldon on studies of cuts in China. When he talks about these at certainly one of Nokia’s oldest US belongings, defensible appears to be the final phrase on his thoughts.
