Facebook’s parent company Meta has embarked on an ambitious venture into artificial intelligence (AI). On this note, CEO Mark Zuckerberg recently shared the organization’s next AI model, Llama 4. This new generation of technology will require more computing power than any of its predecessors.
Key points
According to Zuckerberg, training Llama 4 will need about ten times the computing power used to train Llama3 during a recent earnings call. It shows how intricately designed and sophisticated these AI models have become.
However, Meta is not a newcomer in the field of AI. There is Llama 3 which had eight billion parameters for example – think of them as bricks that make up the AI. And it has been upgraded to a version with 405 billion parameters -the biggest open-source AI model built by Meta.
But why would Meta be so involved in this? Zuckerberg believes it is better to be prepared for the future. He prefers having extra computer system capacity at hand when needed than lack enough of it because creating new Artificial Intelligence projects can take long time and he does not want Google or Apple to outpace his firm.
To do this, they are planning huge investments in data centers, servers and network infrastructure. In other words, they spent $8.5 billion on these segments in second quarter of 2024 which increased by one third from last year’s level at the same time period.
The Chief Financial Officer (CFO) of Meta Inc., Susan Li mentioned that there are various data center undertakings that are being contemplated to ensure future readiness for generating sizable computational powers required by forthcoming AI agents like Llama. What such means is that spending will continue rising throughout coming year too due to investment made here?
It should also be noted that Facebook currently doesn’t expect profitability out of their AI initiatives. They are playing a long game and constructing infrastructure which can be utilized in flexible manner for various purposes. This encompasses training AI models and also using them to do things like rank content or give recommendations on Facebook or Instagram.
Meta perceives some benefits that can accrue despite the high costs. The company has already observed increased user engagement on Facebook Reels, its short video feature, as a result of AI-based recommendations. This is critical since Meta competes with TikTok platform for younger users’ attention.
Meta’s greater bet on computing power underscores its determination to remain at the forefront of this exciting and competitive industry as AI technology continues advancing rapidly.
