Prometheus: Behind Meta’s Quest for Superintelligence
More and more technology companies are experiencing losses as Meta grows, poaching some of the best AI experts in the world to build its superintelligence team that is focused on AGI development.
Now the team is solidifying, Meta is betting hundreds of billions of dollars on infrastructure to build this artificial superintelligence.
The social media company’s CEO Mark Zuckerberg outlines the plan, describing his vision for AI systems that can outperform human intelligence.
The strategy hinges on two enormous data centre projects. Prometheus, a 1 gigawatt facility due to start operations in 2026, marks the first phase.
Then Hyperion, designed to scale up to 5 gigawatts across multiple phases over several years.
At an estimated US$30bn per gigawatt for this type of infrastructure, according to semiconductor research firm SemiAnalysis, Meta is making one of the largest single investments in AI computing power.
The company will become the first lab to launch infrastructure exceeding 1 gigawatt of capacity.
Meta’s Superintelligence team of top AI researchers so far
These facilities will power Meta’s growing collection of AI models, including future versions of Llama, the company’s answer to OpenAI’s ChatGPT and Google’s Gemini.
The company also plans additional “titan” clusters to boost total computational capacity even further.
This is where Meta’s new team comes in, aiming to run this operation, bringing together the company’s AI efforts under leadership that includes Alexandr Wang from Scale AI, Daniel Gross from Safe Superintelligence, former GitHub boss Nat Friedman, Ruoming Pang, a distinguished engineer who managed Apple’s foundation models team and more.
The division got a boost from Meta’s US$14.3bn purchase of Scale AI, a company that specialises in labelling data for machine learning (ML) and created the HLE benchmark.
This deal was partly aimed at fixing quality problems that plagued Llama 4.
Meta’s hiring spree continues, poaching talent from Google DeepMind, OpenAI and Anthropic.
The new recruits include some big names: Hongyu Ren and Jiahui Yu, who both worked on GPT-4o, Jack Rae from Google’s Gemini team, Pei Sun, another Gemini expert and Shengjia Zhao, who helped create ChatGPT.
The company has thrown serious money at recruitment, with packages reportedly reaching US$300m over four years.
The Meta CEO has also personally courted candidates at his homes in Lake Tahoe and Palo Alto, showing just how important this push is to Meta’s future.
“Meta Superintelligence Labs will have industry-leading levels of compute and by far the greatest compute per researcher,” he says.
Prometheus and Hyperion targeting 2026 for launch
Meta’s infrastructure plans put it in direct competition with Microsoft, which backs OpenAI, and Amazon, which provides cloud services to AI developers while building its own AI capabilities.
Both have poured money into data centres to handle AI workloads.
But Meta reckons it has an advantage through independence and deep pockets: “We have the capital from our business to do this,” the CEO says, pointing to the steady revenue stream from advertising on Facebook, Instagram and WhatsApp.
This means that the move reflects how the AI race has become as much about hardware as software.
Training the most advanced AI models requires enormous amounts of processing power and memory, typically using specialised chips from companies like Nvidia.
The Meta CEO has been talking up Meta’s superintelligence ambitions internally for months.
In a memo to staff, he wrote: “As the pace of AI progress accelerates, developing superintelligence is coming into sight. I believe this will be the beginning of a new era for humanity and I am fully committed to doing what it takes for Meta to lead the way.”
His latest announcement spells out how he plans to get there: “For our superintelligence effort, I’m focused on building the most elite and talent-dense team in the industry.”
The scale of Meta’s commitment has also caught the attention of industry watchers.
Karl Freund, Principal Analyst at Cambrian AI Research, tells the BBC: “Clearly, Zuckerberg intends to spend his way to the top of the AI heap. The talent he is hiring will have access to some of the best AI Hardware in the world.”
