In a bold move, Mark Zuckerberg, CEO of Meta, has announced the company’s plan to acquire an astounding 350,000 Nvidia H100 GPUs. This significant investment is part of Meta’s strategic initiative to develop an artificial general intelligence (AGI) – an AI capable of learning and performing a wide range of tasks, much like human intelligence.
The aim is to harness this technology in powering a new wave of innovative services and devices. Among the envisioned applications are advanced digital assistants and augmented reality glasses. Zuckerberg’s vision is to push the boundaries in every area of AI, creating powerful tools for creators, businesses, and consumers alike.
Meta’s choice of the Nvidia H100, an enterprise-grade GPU renowned for its efficiency in training large language models, is a testament to the company’s commitment to this endeavor. Zuckerberg revealed that by the end of the year, Meta plans to have around 350,000 H100 units. This number also includes the equivalent computing power from other GPUs, bringing the total to approximately 600,000 H100 equivalents.
The financial commitment for this project is substantial. With each H100 unit priced at around $30,000, the total cost for the GPUs alone is estimated to be about $10.5 billion. This figure does not include the additional expenses for electricity and other operational costs.
This move by Meta is indicative of the lengths leading tech companies are willing to go to in developing new AI models. The trend is not unique to Meta, as other major players in the tech industry have also made significant investments in similar technologies.
Beyond the company’s internal use, Zuckerberg has expressed Meta’s intention to open-source their AI models. This decision aligns with the company’s previous actions, such as the open-sourcing of Llama 2, and reflects their commitment to making this groundbreaking technology as accessible and beneficial as possible. Through this, Meta aims to foster innovation and development in the wider tech community.
