Wireless

Ex-Bell Labs chief says 6G will need more than AI ‘magic dust’

By Monica Alleven Sep 11, 2026 12:11pm

Former Bell Labs president Marcus Weldon says 5G’s biggest miss wasn’t the technology itself, but operators’ failure to invest in and commercialize the use cases
6G risks repeating the same mistake if the industry rebrands old 5G promises with “AI” instead of solving real business and infrastructure problems
Weldon explained some of his criticism behind Nokia’s treatment of Bell Labs, saying funding has been sharply reduced under current leadership and warning that the research institution is being diminished

The wireless industry might be doing a good job selling itself on 6G, but Marcus Weldon isn’t buying the AI magic dust.

Weldon is a familiar academic and research leader in the 5G era. When 5G was just a glimmer in the industry’s eye, he advised everyone to figure out the problems that 5G needed to solve before trying to set the standards. In 6G, he’s advocating much of the same – with the benefit of hindsight.

That might not sound too radical. Where he stands out from everybody else is … well, having worn the title of Bell Labs president from 2013 to 2021 puts him in a unique position. And he’s making waves talking about the severe cuts he’s seeing at Bell Labs – warning that Nokia risks gutting the very institution that helped define the future of communications for 100 years.

In an interview with Fierce, Weldon didn’t disparage the 3GPP process, which is where wireless standards are developed and ratified before they’re adopted by wireless carriers.

In his view, what works well is the fact that innovative researchers are able to create dream scenarios that they pursue in the labs and test in the real world, then standardize around “very interesting use cases.”

Those use cases drive the evolution of new technology toward standards. “They do a good job of marketing those to operators to get support for that work because operators participate in standards as well and want to see things standardized that have material value,” he said.

That’s all great for creating early excitement within the industry.

However, “where I think it falls apart is operators, in the end, don't materialize those use cases. They don't invest enough. They don't create new teams that are specialized in those areas,” he said.

Startups – at least, before the AI frenzy – may take five and even 10 years to develop the genesis of an idea and see it through to a working prototype before bringing it to market. But mobile operators don’t have that kind of time.

“It’s almost like you've got this massive wealth of innovative tech that is developed and standardized and then built by vendors, but the operators at the end of the day only care about mobile subscribers and delivering service to mobile subscribers in a profitable way,” he said.

What did 5G get wrong?

That gets to Weldon’s bigger point about 5G. Not everyone believes 5G was a disaster, but those who do blame the industry for what it often does: too much hype.

Much of that hype centered around three big technical promises: low latency, high reliability and high bandwidth. But those capabilities needed more than the industry could deliver on its own.

Take autonomous vehicles. Even with all the bells and whistles that 5G could offer, the ecosystem still required entire road infrastructure to be upgraded to allow vehicle sensing – and then computation at the edge to manage those vehicles not as individual pieces but as part of a collective ecosystem.

Private 5G networking was another promising area and while it took off to some extent, it was nowhere near as large as envisioned. Again, there was insufficient advocacy, Weldon said.

“People have just stuck with largely Wi-Fi-based environments and not deployed cellular because again, there wasn’t the imperative for these automated use cases, sensing and automation and optimization,” he said.

History as lesson for future

Some view 4G LTE as the last glory days for wireless operators. It truly solved a problem that everyone was interested in: consumer bandwidth.

The argument could be made that LTE won because of the iPhone. Apple did the marketing – but it was all about the handset and some app store services. It wasn’t about the network.

Since then, hyperscalers, for the most part, have led and telecom has followed. One you’re following, “all the momentum and the magic and mystery and the potential has been taken by someone else. I think that’s telecom’s disease,” he said.

“Standards are good for global access and interoperability, but terrible for innovation,” he said, pointing to lost revenues in SMS and location-based services. Even Nokia had a mapping service, and “they gave that up.”

Uber was a great service supported by 4G LTE, but it’s not as if wireless carriers are getting a windfall from that. They support the service, but they’re not making big bucks from it.

Addressing 5 challenges that start with “un”

Is it possible for telecom to recover and get into the driver’s seat?

Yes – if they solve five problems – and they all start with the letters “un,” he said.

They consist of the following: unfavorable regulation, unfair competition, unsupportive shareholders, unimaginative leadership and underrepresented technical strategies.

The question then becomes: Are we going to solve those in the 6G era?

“I see no sign of it,” he said. “The use cases are not any different really than the 5G use cases. We now just say ‘AI,’” he said. “There’s some technical changes about spectral efficiency and bandwidth. But sprinkling AI magic dust on a network isn’t going to manifestly change” anything.

AI RAN and GPUs

Speaking of AI, AI RAN is all the rage in technical circles these days. Bell Labs was an early pioneer here. Seven or eight years ago, Bell Labs built what at the time they called “GPU RAN,” he said.

He acknowledged that he was part of the hype around Massive MIMO – which, by the way, was a concept led by inventor Tom Marzetta when he was at Bell Labs. The esteemed research unit – named for Alexander Graham Bell – kickstarted a lot of notable tech during its heyday, but according to Weldon, its future under Nokia’s leadership currently looks bleak.

It just so happens Nokia’s future is closely tied to GPU behemoth Nvidia, which last year invested $1 billion in the Finnish vendor.

The GPU gets a lot of hype over its capabilities, but Weldon said “it’s not as if the GPU has magic.” The ASIC has been used for decades and it does much of the same work, but it’s not as reprogrammable, he said.

He noted Nokia’s previous history putting all its eggs into the Intel basket for ASIC and “that fell apart,” crippling Nokia for years. It just so happens Nokia’s current CEO Justin Hotard came from Intel, as did several of its senior-level executives since then.

But Hotard came from Intel’s data center division, not the telecom side, he noted. “If they were on the telecom side, they would have remembered this,” he said.

What's the trouble at Bell Labs?

Part of the problem stems from the lack of financial support to continue innovating at Bell Labs.

Weldon sounded the alarm in a LinkedIn post last month about the staff cuts happening at Nokia Bell Labs. Nokia told Fierce that Bell Labs “remains a deeply important part of Nokia, with a long track record of turning world-class research into technologies that deliver commercial impact and move our industry forward.”

But it also acknowledged that Bell Labs is “entering a new chapter,” declining to comment on cuts to specific business units.

Bell Labs employed about 750 people after Nokia’s takeover in 2016. By 2024, that number dropped to 681 and before the end of next year, Nokia aims to reduce the research headcount at Bell Labs to 408, according to Light Reading.

Weldon said that even when Bell Labs was under the ownership of AT&T, it was intent on finding solutions that would bring lasting value to the business.

“This idea that Bell Labs doesn’t do things of value to the business division is folly,” he said.

The consequence? In the near term, “Bell Labs is being decimated under that logic,” he said.

According to Weldon, the telecom industry is losing R&D firepower when it needs it the most. That could be the hardest lesson to learn from the 5G-going-into-6G era.