Meta is stepping into new territory with bold plans to revolutionize the AI industry. The company is gearing up to create the world’s most powerful AI cluster, an endeavor that could redefine what’s possible in the realm of artificial intelligence.
Under the leadership of Mark Zuckerberg, Meta is charging full steam ahead, aiming to establish dominance in the AI sector. The company has assembled a powerhouse team reminiscent of a superhero squad, and their latest project, dubbed Prometheus, promises to push boundaries further than ever before.
Prometheus is set to be the largest AI training cluster globally, and rather than conventional buildings, it’s housed in expansive tents. By the end of 2026, this innovative cluster aims to achieve unprecedented performance with 500,000 GPUs and a staggering IT power of 1,020 MW. Its computing power will exceed 3.17 trillion TFLOPS, outpacing even OpenAI’s capabilities.
This unique venture will harness natural gas as its primary power source, with on-site plants generating up to 200MW. Prometheus plans to leverage NVIDIA’s cutting-edge AI clusters, potentially valued in the trillions. However, strategic leasing agreements are likely in place to manage costs effectively.
Prometheus is poised to deliver groundbreaking advancements in AI computational power, surpassing even the formidable Stargate, valued at over $100 billion. Remarkably, while modern supercomputers like Frontier took nearly three years to become operational, Meta’s project is expected to take just 12 to 15 months, signaling a significant leap in computing evolution.
With each financial quarter, Meta’s investment grows, underscoring the company’s commitment to pushing beyond mediocrity. Looking ahead, Meta is also planning a “2GW-cluster” called Hyperion, indicating that their journey to lead the AI charge is just beginning.
