Microsoft, in collaboration with OpenAI, is reportedly planning to build a $100 billion supercomputer named “Stargate”. This supercomputer is intended to power OpenAI’s next generation of artificial intelligence systems. Stargate is the fifth and final phase of Microsoft and OpenAI’s plan to construct several supercomputers across the United States.
The supercomputer is rumored to be one of the largest and most advanced data centers globally, potentially occupying several hundred acres of land and requiring up to 5 gigawatts of power. It is considered crucial for training and operating more advanced AI models than the current ChatGPT-4. The launch of Stargate could be as early as 2028.
The success of Stargate largely depends on OpenAI’s ability to deliver its next major AI upgrade, rumored to be GPT-5, which is expected in early 2025. OpenAI reportedly failed to ship a new AI project named “Arrakis” in 2023, largely blaming the limitations of current supercomputers.
The creation of Stargate could explain why Sam Altman, OpenAI’s CEO, was rumored to be creating a very expensive AI chip factory. The biggest challenge to building an AI supercomputer of this scale will be securing Nvidia GPUs to power it. Nvidia’s chips have been in short supply throughout the AI hype cycle, and a competitor from Altman and Microsoft could give them an unprecedented advantage.
read more > gizmodo.com