admin Facebook giant Meta Platforms is excited about Nvidia’s new artificial intelligence giant coming later this year. The unveilede B200 “Blackwell” is expected to massively improve the capabilities of clients focusing on a 30-fold gain in speed for chatbot-style dialogs.
Speed increments or reductions for data-intensive training are yet unknown, but Nvidia has intimated plans to support chips by the end of the year later; and in 2025, it will materialize it in an extended-scale internet. Meta is dependent on Nvidia because Meta uses several hundred thousand Nvidia chips to upgrade content recommendation systems and generative AI assignments.”
Last October, CEO Mark Zuckerberg announced their plans to accumulate close to 350,000 last-generation H100 chips by year-end. To date, the total equivalent adds up to 600,000 H100 chips when taking other GPUs into account. By adopting the Blackwell chip, Meta makes an important stride in its commitment to advance the boundaries of AI expertise.
For now, the development company uses two GPU clusters with roughly 24,000 H100 GPUs in each to train their third-generation Llama model. Subsequent to the third generation, the fourth and further ones will use the Blackwell chip which makes the chip an important part of Meta’s AI future. Meta’s chimerical leap forward in the onward with its progressive embrace of the latest cutting-edge AI technologies permits the transformation of the frontiers for the content recommendation systems besides that of generative AI products .
Furthermore, the inclusion of the Blackwell chip among Meta’s collection facilitates extra substantial productivity and prepares for the discovery of modern indices for innovation fuelled by AI. Meta’s vanguard role in the forthcoming era of intellectual computer usage makes its relationship with Nvidia as an outlier and merit that acts as a light pillar orientating the business into additional capabilities of AI.
Magazine
