Meta AI Chip Meta to Begin Manufacturing In-House AI Processor in September
Meta plans to manufacture its first in-house AI chip as it expands computing infrastructure.
Meta AI Chip production is set to begin in September as Meta Platforms moves forward with its strategy to strengthen artificial intelligence infrastructure and reduce its dependence on third-party chip suppliers.
Mbappe France Beat Morocco to Reach World Cup Semi-Finals
According to an internal company memo reviewed by Reuters, Meta plans to manufacture its new AI processor, code-named “Iris,” as part of its long-term Meta Training and Inference Accelerators (MTIA) programme.
Meta Expands AI Hardware Strategy
The Iris chip is one of four generations of AI processors that Meta is developing in-house. The company aims to use custom-designed silicon to improve the artificial intelligence systems behind Facebook and Instagram.
Moreover, testing of the new chip took only six weeks and revealed no significant technical problems. As a result, Meta believes the project is progressing faster than previous in-house chip development efforts.
Reducing Dependence on Nvidia
Meta designed the Iris processor to meet its own AI computing needs. In addition, the company is working with Broadcom on the chip’s design, while Taiwan Semiconductor Manufacturing Company (TSMC) will manufacture it.
The custom processor will complement the graphics processing units (GPUs) that Meta continues to purchase from Nvidia and AMD. However, the company said integrating the latest commercial GPUs across its vast infrastructure has proved challenging and time-consuming.
Industry analysts say developing proprietary chips could help Meta lower computing costs and gain greater control over its AI infrastructure.
Massive Investment in AI Infrastructure
Meta plans to deploy seven gigawatts of computing capacity this year. So far, it has added one gigawatt and expects to install another 5.5 gigawatts before the end of the year.
Furthermore, the company intends to double its computing capacity to 14 gigawatts next year as demand for AI services continues to grow.
The internal memo also states that Meta could spend up to $145 billion on AI infrastructure this year, making it one of the largest investors in artificial intelligence globally.
Long-Term Supply Agreements
To support its expansion, Meta has secured multi-year agreements with several technology suppliers. For example, Samsung Electronics will provide memory chips, Sandisk will supply flash storage and Sumitomo Electric will deliver fibre-optic equipment.
Meanwhile, rising global demand for AI hardware has increased the cost of memory and semiconductor components. Analysts say the rapid increase in chip prices has become a growing concern across the technology industry.
Overall, Meta’s decision to manufacture its own AI chips marks a significant step in its long-term strategy to build a more efficient and independent AI ecosystem while supporting the rapid expansion of its data centre network.
