Reuters - Meta begins testing its first in-house AI training chip
Facebook owner Meta is testing its first in-house chip for training artificial intelligence systems, a key milestone as it moves to design more of its own custom silicon and reduce reliance on external suppliers like Nvidia, two sources told Reuters.
Deployment and Production Plans
The world’s biggest social media company has begun a small deployment of the chip and plans to ramp up production for wide-scale use if the test goes well, the sources said. Meta's goal is to bring down its infrastructure costs as it heavily invests in AI tools for growth.
Meta's new training chip is a dedicated accelerator designed specifically for AI tasks, making it more power-efficient than integrated GPUs typically used for AI workloads. The company is collaborating with Taiwan-based chip manufacturer TSMC to produce the chip.
Development and Testing
The test deployment followed Meta's successful "tape-out" of the chip, a crucial step in silicon development involving sending an initial design through a chip factory. Meta considers this milestone a significant achievement in its chip development process.
Meta's chip, part of the Meta Training and Inference Accelerator (MTIA) series, aims to enhance the company's AI capabilities and efficiency. The company has faced challenges in the past but sees its chip development as a strategic long-term investment.
Future Plans
Meta executives envision using their own chips for training by 2026, starting with recommendation systems and later expanding to generative AI products. The company's strategy involves a gradual approach to chip development, focusing on different AI applications over time.
The decision to test in-house chips reflects Meta's commitment to innovation and self-sufficiency in AI technology development.
Stay updated on the latest news in the AI chips market at AI Chips News.
