Meta has unveiled details about its latest AI semiconductor chip, Artemis, which will power AI services on Facebook, Instagram, and WhatsApp.
Nvidia is a dominant force in the semiconductor chips market. Major Artificial Intelligence technology developers and big tech companies are hoarding Nvidia’s latest semiconductor chips at soaring prices. Meta is one of the biggest buyers of Nvidia’s H100 and H200 chips to power its AI services. Even though the tech giant will keep buying more chips from Nvidia, it has taken a significant leap towards self-reliance.
Meta has unveiled details about its in-house semiconductor chip, Artemis, which will reduce its dependency on Nvidia for AI-powering hardware. Mark Zuckerberg revealed intentions to procure approximately 3,50,000 H100 flagship chips from Nvidia and an additional 6,00,000 chips for other vendors this year. Simultaneously, Artemis production will gradually increase to shift the dependency from external chips while meeting AI computational needs.
Artemis will be manufactured by an AI chip-making company, Taiwan Semiconductor Manufacturing Co (TSMC), using an advanced 5nm process. The Meta Training and Inference Accelerator chip boasts three times higher performance power than its predecessor. The MTIA chip is deployed across all data centers to serve AI applications on all Meta platforms.
Meta’s effort to develop in-house chips underscores its effort to leverage cutting-edge semiconductor technology and expand Meta’s capabilities to support generative AI workloads. Meta plans to develop more AI hardware to minimize reliance on external vendors. Due to their high demand, semiconductor chips are heavily commoditized, and getting quick replacements during emergencies may become a daunting task for Meta’s data centers. To prevent revenue loss, in-house chips will serve as a quick backup solution in case of unwanted incidents.
With Artemis coming to plow the semiconductor chips ground, Meta will significantly expand its AI offerings across all platforms. The tech giant has already enabled multiple features on Facebook, Instagram, and WhatsApp that leverage Artificial Intelligence technology. Meta continues to invest in custom semiconductor solutions to gain a competitive advantage over its rivals’ AI expansion efforts. The development of indigenous AI infrastructure will significantly ramp up its research and development, unlocking new opportunities to sustain its leadership position in the social media industry.
