Nokia’s Bell Labs Cuts: The AI-RAN Gamble That Could Redefine Open RAN’s Future

In the fast-evolving world of telecommunications, few stories carry the weight of history and the urgency of the future quite like the one unfolding at Nokia. Recent reports of sweeping cuts at Bell Labs, the legendary research arm that has birthed everything from the transistor to the laser, have sent shockwaves through the industry. Combined with Nokia’s deepening reliance on Nvidia for its AI-RAN ambitions, the moves raise profound questions about the role of fundamental research, the perils of strategic concentration, and what it all means for the Open RAN movement that has long promised to disaggregate and democratize the radio access network.

As an analyst covering Open RAN, I’ve watched the convergence of AI and radio networks evolve from theoretical promise to commercial reality. But the story of Bell Labs’ cuts is not just about cost-cutting or even about Nokia’s immediate future. It’s a cautionary tale about how the industry’s shift toward AI-native networks could reshape the very institutions that made modern telecommunications possible—and what that means for the open ecosystem that many believe is the industry’s best hope for innovation.

The News That Broke the Camel’s Back

On September 8, 2026, Light Reading reported that Nokia plans sweeping cuts at Bell Labs, a move that has alarmed even its former president, Marcus Weldon. Weldon, who led Bell Labs from 2014 to 2020, publicly criticized the decision, arguing that Nokia is putting “all its eggs in one basket” by betting heavily on Nvidia’s GPU-based AI-RAN technology. According to Weldon, the AI-RAN work that Nokia’s CTO, Pallavi Mahajan, champions has “almost no Bell Labs involvement,” despite Bell Labs having invented AI-RAN and virtual RAN concepts years ago. He sees an “over-dependence on Nvidia” that fails to leverage Bell Labs’ deep reservoir of innovation.

This news comes on the heels of Nokia’s broader restructuring. The company, which employed over 103,000 people in 2018, finished last year with 74,100 employees (excluding Infinera). Under former CEO Pekka Lundmark’s plans, another 4,000 jobs were slated for cuts by the end of 2026. Now, with additional cuts at Bell Labs, the workforce could shrink further. The move is part of Nokia’s strategy to focus on AI-native networks, but it raises a critical question: can a company innovate its way to the future if it dismantles the very institution that has historically been its innovation engine?

Bell Labs: A Legacy Under Threat

To understand the gravity of this moment, we must look back at Bell Labs’ storied history. Founded in 1925, Bell Labs has been the birthplace of some of the most transformative technologies of the 20th and 21st centuries. The transistor, the laser, the charge-coupled device (CCD), the Unix operating system, and the C programming language all emerged from its halls. Its researchers have won nine Nobel Prizes. In the telecom world, Bell Labs laid the groundwork for cellular communications, fiber optics, and, more recently, software-defined networking and virtualized RAN.

When Nokia acquired Alcatel-Lucent in 2016, it inherited Bell Labs, which had been part of Alcatel-Lucent. At the time, Nokia touted Bell Labs as a crown jewel, a source of future breakthroughs that would drive its technology leadership. Indeed, under Weldon’s leadership, Bell Labs made significant strides in areas like virtual RAN and even the AI-RAN concept that is now associated with Nvidia. But as Weldon notes, there’s a corporate “amnesia” that often causes Nokia to forget where its best ideas came from.

The proposed cuts threaten to sever the link between Bell Labs’ research and Nokia’s product development. Weldon’s warning echoes a familiar theme in the tech industry: when companies prioritize short-term profits over long-term research, they risk losing the innovative edge that made them successful in the first place. For Open RAN, which depends on a vibrant ecosystem of contributors, the diminishment of a key research powerhouse could have ripple effects.

The Nvidia Bet: A Double-Edged Sword

Nokia’s strategy under CEO Justin Hotard (who took over from Lundmark in 2025) has been to embrace AI-RAN as a key growth area. In June 2026, Nokia and Nvidia announced a partnership to bring GPU-accelerated AI-RAN to market, with Nvidia’s hardware powering both connectivity and AI inferencing from the same equipment. This partnership has been widely covered on this blog, including in our pieces on Nokia’s commercial AI-RAN platform and its implications for Open RAN.

The appeal is obvious: GPUs can handle both the compute-intensive tasks of running a RAN and the demands of AI applications, potentially reducing the need for dedicated hardware and opening new revenue streams for operators. But Weldon’s critique highlights a significant risk: by tying its AI-RAN future to Nvidia’s proprietary GPUs, Nokia may be repeating a mistake it made after acquiring Alcatel-Lucent, when it relied heavily on Intel for its silicon needs. That strategy failed when Intel’s 10-nanometer process node slipped, leaving Nokia without a competitive ASIC strategy.

Weldon argues that Nokia needs a companion ASIC strategy to mitigate the risk of relying on a single vendor. Without it, the company could find itself at the mercy of Nvidia’s pricing and roadmap, unable to differentiate its offerings. This is not just a Nokia problem; it’s a broader industry concern. As AI-RAN gains traction, many operators are wary of becoming dependent on a single chip vendor, especially one with as much market power as Nvidia.

For Open RAN, which champions multi-vendor interoperability, the reliance on Nvidia’s proprietary GPUs could be seen as a step backward. Open RAN’s promise lies in its ability to allow operators to mix and match components from different vendors, fostering innovation and reducing costs. If the AI-RAN layer becomes dominated by Nvidia, that promise could be undermined.

Historical Context: The Evolution of RAN and AI

To fully appreciate the significance of Nokia’s current strategy, it’s helpful to trace the evolution of radio access networks and the role of AI within them. Traditional RANs have been built on proprietary, purpose-built hardware, with baseband units and radio units tightly coupled. The introduction of software-defined networking and network functions virtualization (NFV) in the 2010s began to disaggregate these functions, allowing them to run on commercial off-the-shelf servers. This paved the way for Virtualized RAN (vRAN) and, eventually, Open RAN, which standardized interfaces between components.

AI’s role in the RAN has also been evolving. Initially, AI was used for network optimization, such as traffic prediction and energy management, running in the network management layer. But as compute power increased and AI algorithms became more sophisticated, the idea of embedding AI directly into the RAN’s real-time processing—like scheduling and beamforming—gained traction. This is the essence of AI-RAN: using AI to make the RAN itself more efficient and capable.

Nokia has been at the forefront of this trend, launching what it called the industry’s first commercial AI-RAN platform in early 2026. The platform leverages Nvidia’s accelerated computing to run AI workloads alongside traditional RAN functions. Similarly, Ericsson launched its AI-in-RAN as a commercial software subscription in June 2026, running AI models directly on baseband hardware already installed at cell sites. These moves signal a shift from AI as an add-on to AI as an integral part of the RAN.

But the path has not been without its challenges. As our blog has noted, AI-RAN hype has collided with flat RAN economics, and operators are demanding to see clear returns on their investments. The market for RAN equipment has been flat, with some analysts projecting a 29% decline in mobile network spending from 2026 to 2031. In such an environment, cost-cutting and efficiency are paramount, but so is innovation that can drive new revenue.

The Open RAN Connection

For Open RAN, the stakes are particularly high. Open RAN’s value proposition is its ability to reduce vendor lock-in, foster innovation, and lower costs through multi-vendor interoperability. The movement has gained momentum, with major operators like Vodafone, Telefonica, and Orange committing to Open RAN deployments. However, the integration of AI into the RAN introduces new complexities, and the role of proprietary silicon from companies like Nvidia could become a point of contention.

Nokia’s strategy with Nvidia is not necessarily antithetical to Open RAN. After all, Open RAN defines interfaces and standards, not the underlying hardware. A GPU-accelerated baseband unit can still conform to Open RAN specifications, allowing it to interoperate with other vendors’ components. But the risk is that the AI-RAN layer becomes a de facto standard, with Nvidia’s CUDA ecosystem becoming as entrenched as Intel’s x86 architecture was in the server market.

This is where Bell Labs’ research could have provided a counterbalance. By developing alternative approaches to AI-RAN, such as specialized ASICs or novel algorithms, Bell Labs could have helped ensure that the ecosystem remains diverse and competitive. With its potential diminishment, that counterweight may be lost.

The Broader Industry Implications

Nokia’s moves are not occurring in a vacuum. The telecom industry is undergoing a massive transformation, driven by the rise of AI, the rollout of 5G-Advanced and 6G, and the need for more efficient, flexible networks. In this context, research and development are more critical than ever. Yet, we are seeing a troubling trend: many telecom vendors are cutting back on long-term research to focus on near-term product development.

Ericsson, for example, has also been restructuring its R&D, though it has maintained a strong focus on AI and cloud-native technology. Huawei, despite sanctions, continues to invest heavily in R&D, as evidenced by its recent SingleRAN 22.1 software update, which includes AI-powered features like iBeam 3.0 and “0 Bit 0 Watt 0 Loss” energy-saving technology. Huawei’s approach is to embed AI into its software, rather than relying on external GPU vendors, which gives it more control over its roadmap.

In contrast, Nokia’s reliance on Nvidia could be seen as a strategic vulnerability. If Nvidia’s GPUs fail to meet performance or cost expectations, or if Nvidia decides to prioritize other markets, Nokia could be left without a viable AI-RAN solution. This is the same risk that Weldon identified with Intel.

Moreover, the cuts at Bell Labs could have a chilling effect on innovation across the industry. Bell Labs has been a source of fundamental research that has benefited not just Nokia but the entire telecom ecosystem. Its work on information theory, coding, and networking has laid the groundwork for many of the technologies we take for granted today. Diminishing that resource is a loss for the industry as a whole.

A Look Ahead: What This Means for Open RAN Operators

For operators investing in Open RAN, the developments at Nokia should serve as a wake-up call. They need to be vigilant about vendor dependencies, not just at the RAN level but also at the chip level. The promise of Open RAN is that it enables a multi-vendor approach, but if the underlying silicon is dominated by a single player, the benefits of openness may be eroded.

Operators should also consider the importance of research and development in their own organizations and in the broader ecosystem. The challenges of AI-native networks, 6G, and sustainability will require sustained innovation. If the industry’s research institutions are weakened, the pace of innovation could slow, ultimately affecting operators’ ability to meet future demands.

One potential positive outcome of this situation is that it could spur new entrants to fill the gap left by Bell Labs. Startups and academic institutions are already making significant contributions to AI-RAN research. For example, the Open RAN ecosystem has seen a proliferation of small companies developing specialized components, from radio units to AI-powered optimization tools. The open-source community is also playing a role, with projects like OAIBOX (OpenAirInterface) providing platforms for experimentation.

But these efforts are unlikely to replace the deep, long-term research that Bell Labs has historically provided. The industry needs both incremental innovation, which comes from startups and product teams, and fundamental research, which comes from institutions like Bell Labs. Cuts to the latter could have consequences that are not immediately apparent but will be felt in the years to come.

Conclusion: The Innovation Imperative

As I reflect on the news of Bell Labs cuts and Nokia’s Nvidia bet, I’m reminded of a quote often attributed to the great innovator Alexander Graham Bell: “The most successful men in the end are those who succeed in the long run, not those who succeed in the short run.” Nokia’s strategy may yield short-term gains in efficiency and cost savings, but it risks sacrificing the long-term innovation that has been its lifeblood.

The Open RAN community has always championed openness and innovation. We should be concerned when one of the industry’s most storied research institutions is diminished. We should also be wary of strategies that concentrate power in the hands of a few dominant players.

The future of telecom, and of Open RAN in particular, depends on a vibrant ecosystem of players—vendors, operators, researchers, and startups—each contributing their unique strengths. As we navigate the AI-native era, we must ensure that the ecosystem remains open, diverse, and innovative. That means supporting research, encouraging competition, and being mindful of the dependencies we create.

Nokia’s decisions will have far-reaching implications, not just for the company itself but for the entire industry. As analysts, we must watch closely and ask the hard questions. Is the AI-RAN future a single-vendor affair, or can it be an open, multi-vendor landscape? Can the industry afford to lose the kind of fundamental research that Bell Labs has provided for a century? And what will the telecom landscape look like in 2030 if we prioritize short-term gains over long-term innovation?

These are the questions that will shape the next decade of telecommunications. And as the story continues to unfold, we at Open RAN will be here to provide analysis and context.

Sources

• Nokia plans sweeping Bell Labs cuts to alarm of ex-president - Light Reading
• Nokia’s AI-RAN Push Can Operators Really Move at Software Speed - Open RAN
• Nokia and NVIDIA’s Commercial AI-Native RAN Tracing the Path from Disaggregated Open RAN to GPU-Powered Intelligence - Open RAN
• Huawei’s SingleRAN rebuilds radio software for AI - Telecoms.com
• Infrastructure Upgrade: Global trends in rebuilding digital infrastructure for AI - tele.net.in
• Nokia Opens Mobile Core Early Access Globally as Operators Prepare for AI-Native 5G Networks - TelecomLead