close
close

tub-blois

Breaking: Beyond Headlines!

Meta’s Upcoming Llama AI Models Train on a ‘Bigger Than Anything’ GPU Cluster
aecifo

Meta’s Upcoming Llama AI Models Train on a ‘Bigger Than Anything’ GPU Cluster

Managing such a gargantuan array of chips to develop Llama 4 will likely present unique engineering challenges and require large amounts of energy. Meta executives evaded an analyst question on Wednesday about energy access constraints in parts of the United States, which have hampered companies’ efforts to develop more powerful AI.

According to an estimatea cluster of 100,000 H100 chips would require 150 megawatts of power. The largest national laboratory supercomputer in the United States, El Capitanon the other hand, requires 30 megawatts of power. Meta plans to spend up to $40 billion in capital this year to provide data centers and other infrastructure, an increase of more than 42% from 2023. The company expects even more growth torrid of these expenses next year.

Meta’s total operating costs increased about 9 percent this year. But overall sales — largely thanks to advertising — jumped more than 22 percent, leaving the company with bigger margins and bigger profits even as it invested billions of dollars in Llama’s efforts.

Meanwhile, OpenAI, considered the current leader in cutting-edge AI development, is burning through cash despite charging developers for access to its models. What for now, this remains a non-profit enterprise said it formed GPT-5, a successor to the model that currently powers ChatGPT. OpenAI said GPT-5 would be larger than its predecessor, but it said nothing about the computing cluster it uses for training. OpenAI also said that in addition to scale, GPT-5 will incorporate other innovations, including a recently developed tool. approach to reasoning.

Sam Altman, CEO said that GPT-5 will constitute “a significant leap forward” compared to its predecessor. Last week, Altman responded to a report indicating that OpenAI’s next frontier model would be released by December. while writing on X, “fake news out of control”.

On Tuesday, Google CEO Sundar Pichai said that the latest version of the Gemini family of generative AI models is in development.

Meta’s open approach to AI has sometimes proven controversial. Some AI experts worry that making much more powerful AI models freely available could be dangerous, because it could help criminals launch cyberattacks or automate the design of chemical or biological weapons. Although Llama is fine-tuned before release to limit bad behavior, removing these restrictions is relatively simple.

Zuckerberg remains optimistic about the open source strategy, even as Google and OpenAI push proprietary systems. “It seems pretty clear to me that open source will be the most cost-effective, customizable, reliable, high-performance and easiest-to-use option available to developers,” he said Wednesday. “And I’m proud that Llama is leading the way in this area.”

Zuckerberg added that Llama 4’s new capabilities should be able to power a wider range of features in Meta services. Today, Llama’s signature model-based offering is the ChatGPT-like chatbot known as Meta AI, available on Facebook, Instagram, WhatsApp and other apps.

More than 500 million people use Meta AI every month, Zuckerberg said. Over time, Meta hopes to generate revenue from ads from this feature. “There will be a growing set of queries that people will use it for, and monetization opportunities will exist over time as we get there,” Susan Li, Meta’s chief financial officer, said on the call. from Wednesday. With the potential for revenue from advertising, Meta might well be able to subsidize Llama for everyone.