What is behind the dispute between former Bell Labs president and Nokia CTO Marcus Weldon and Nokia over the research institution’s future—including Weldon’s claim that staffing has fallen from more than 1,200 researchers to about 600, his criticism of the removal of the “Preside

The dispute is fundamentally about whether Nokia is hollowing out Bell Labs’ distinctive, long-horizon research culture—or modernizing it into a smaller, commercially focused AI-and-networking research unit. Weldon sees the organizational and physical changes as evidence of decline; Nokia portrays them as the start of a new chapter. 34

• Weldon’s case: The former Bell Labs president and Nokia CTO says the research workforce has fallen from more than 1,200 to roughly 600. He objected to dropping the century-old “President” title for the laboratory’s leader and to removing plaques recognizing former leaders, treating these as institutional signals that Nokia is minimizing—or “erasing”—Bell Labs’ history. Those are Weldon’s allegations, rather than figures Nokia has publicly confirmed in the material reviewed. 164

• Why the symbols matter: For Weldon, Bell Labs was valuable not merely as a corporate R&D department but as a protected, multidisciplinary institution where physicists, mathematicians, software researchers and network engineers could pursue foundational problems over long time horizons. Its historic work spans the transistor, laser, information theory, Unix and C/C++—the kind of cross-field research he argues is especially necessary for AI, rather than obsolete because of it. 915

• Nokia’s answer: Nokia rejects the claim that it is devaluing the lab, saying Bell Labs remains an important part of the company whose role is to turn world-class research into commercially valuable, industry-shaping technology. The appointment of Guru Parulkar signals a more explicitly AI-oriented direction rather than an abandonment of research. 37

• The Murray Hill move intensifies the disagreement: The proposed relocation from the historic Murray Hill campus to a smaller New Brunswick site by 2028 is easy for critics to read as a contraction in both scale and identity. Nokia can reasonably frame it as a facilities and cost decision; Weldon reads it alongside staffing, titles and commemorative removals as cumulative evidence of institutional retreat. Insufficient evidence is available to establish that the move itself will reduce research capability.

• The corporate strategy behind it: Nokia is redirecting technical emphasis toward programmable, GPU-based AI-RAN: Nvidia’s $1 billion investment was paired with Nokia plans to build future RAN software around CUDA and make GPUs central to 5G/6G products, initially alongside custom hardware. 5 This is a meaningful contrast with Ericsson’s continued reliance on internally designed custom ASICs for its RAN strategy. 12

• Why this creates tension with the Bell Labs ideal: A GPU/CUDA-led strategy can make networks more adaptable, accelerate AI deployment and connect Nokia to Nvidia’s developer ecosystem. But it can also shift the company toward product-platform integration and external technology dependence—precisely the sort of near-to-medium-term commercial orientation Weldon fears could displace independent basic research. Nokia’s view is that AI-RAN and 6G are themselves frontier research problems worthy of Bell Labs.

• Restructuring context: Reports of a Hangzhou radio-R&D closure and roughly 1,600 affected jobs by the end of 2026, alongside rising restructuring costs, reinforce the impression of geographic and organizational consolidation. 69 The evidence supports a major reduction of Nokia’s China R&D footprint, but “full China exit” should be treated cautiously unless Nokia formally confirms it, because reports vary in scope. 613

In short, both sides accept that Bell Labs must be relevant to AI. They differ over the model: Weldon argues relevance requires restoring the lab’s scale, autonomy, historical continuity and broad scientific range; Nokia argues relevance requires concentrating research on AI-enabled networks and converting it faster into deployable products.