Why Meta's Reshaped AI Labs Drive Demand for Telcos

It has been a turbulent year across the global AI sector, with telcos increasingly drawn into the conversation due to the infrastructure demands of new and emerging technologies. Few companies have been more prominent in this shifting AI sector than Meta.

Under the leadership of Mark Zuckerberg, Meta has pursued an aggressive recruitment strategy, bringing in leading AI engineers and computer scientists from rivals and disruptive startups.

The influx of expertise reflects Zuckerberg’s clear ambition: to develop what he has described as “artificial general intelligence” or “superintelligence”.

For telcos, the implications of these developments are significant. The sheer compute power and data throughput required to support advanced AI models will depend on next-generation networks, low-latency connectivity and expanded data centre capacity.

To accelerate progress towards its goal, Meta recently completed a major internal reorganisation of its AI division, announced on 19 August.

The move not only illustrates its long-term vision for AI but signals the growing interdependence between hyperscale AI development and the telecommunications ecosystem.

A four-pronged organisational model

The newly branded Meta Superintelligence Labs has been divided into four specialised groups, each tasked with distinct responsibilities.

• TBD Lab: Led directly by Meta’s new Chief AI Officer, Alexandr Wang, the division will drive the development of Meta’s large language models (LLMs), including the Llama tools that underpin the company’s AI assistant.
• Fundamental AI Research (FAIR): Continuing its long-term mandate since 2014, FAIR will remain focused on foundational AI research under co-founder Rob Fergus.
• Products and Applied Research: Headed by former GitHub CEO Nat Friedman, this unit will translate advanced AI research into consumer and enterprise-facing solutions.
• MSL Infra: With Aparna Ramani at the helm, this group will oversee Meta’s infrastructure requirements, including the data centres and computational backbone required to sustain large-scale AI operations.

As Alexandr noted in his internal memo: “Superintelligence is coming and to take it seriously, we need to organise around the key areas that will be critical to reach it — research, product and infra.”

Recruitment at record scale

The restructuring follows one of the most ambitious AI recruitment drives in the sector. Meta has invested heavily in talent acquisition, often luring senior researchers from direct rivals. High-profile appointments include Shengjia Zhao, an OpenAI researcher and co-creator of ChatGPT, who joined Meta as Chief AI Scientist.

Reports indicate that Meta has extended compensation packages valued in the hundreds of millions of dollars, with at least two offers reaching US$1bn over several years.

In total, it is believed to have recruited at least 18 researchers from OpenAI, alongside several key hires from Google.

At the same time, Meta has seen the departure of longstanding figures, such as Joelle Pineau, who moved to Cohere and Angela Fan, who joined OpenAI. Loredana Crisan, a VP of Gen AI, has also left to become Chief Design Officer at Figma.

A strategic reset

Beyond personnel changes, Meta is adjusting its AI development strategy. Alexandr’s team has shelved the company’s previous frontier model, known as Behemoth, after disappointing performance results. Work is now underway on an entirely new model, signalling a reset in direction.

In parallel, executives are considering making the next generation of Meta’s AI systems closed-source, diverging from its long-held philosophy of openness. The potential shift has created internal friction, with some established teams reportedly uneasy about both the new hires and the change in approach.

Despite the challenges, Alexandr has highlighted early signs of success: “Already in the past month, I’ve seen meaningful progress in each of these collaborations.”

Heavy investment in AI infrastructure

Vast financial commitments support Meta’s ambitions. Capital expenditure is projected to reach US$72bn in 2025, with the majority allocated to building new data centres and funding AI research.

Additionally, Meta has invested US$14.3bn in Scale AI, a move closely tied to Alexandr’s arrival as Chief AI Officer.

For the telecommunications sector, the level of investment highlights the scale of computational demand required to sustain AI research at a superintelligence level.

Hyperscale data centre build-outs, advanced networking capabilities and unprecedented compute requirements all highlight the close interdependence between AI and telecoms infrastructure.

During a recent investor call, Zuckerberg highlighted that superintelligence could usher in “a new era of individual empowerment”.

He further noted that AI has already strengthened Meta’s core advertising business, offering early proof of value while long-term ambitions remain under development.

The reorganisation is expected to remain in place for the foreseeable future as Meta looks to stabilise its AI operations and build momentum against competitors such as OpenAI and Google.

Whether these structural and financial commitments will translate into breakthroughs in artificial general intelligence remains an open question. Still, the telecommunications industry will be watching closely, given the fundamental role of connectivity, infrastructure and data transport in enabling such advances.