Advertisement
  1. News
  2. Technology
  3. Meta tests in-house AI chip to reduce dependence on Nvidia: A major step in AI development

Meta tests in-house AI chip to reduce dependence on Nvidia: A major step in AI development

Facebook owner Meta is testing its first custom AI training chip, aiming to cut infrastructure costs and lessen reliance on Nvidia. With its new AI chip, Meta is positioning itself to take greater control of its AI infrastructure, a move that could reshape the competitive landscape.

Meta
Meta Image Source : Reuters
Written By: Saumya Nigam @snigam04
Published: , Updated:
New Delhi:

Meta, the parent company of Facebook, Instagram, and WhatsApp has started testing its first-ever in-house chip designed for training artificial intelligence (AI) systems. The move marks a significant milestone in Meta’s journey toward custom silicon development, which will potentially reduce its dependence on external suppliers like Nvidia. If successful, Meta plans to scale up the production of this AI chip for broader implementation.

Meta’s push for custom chips to lower AI costs

According to a report by Reuters, the development of this custom AI chip is part of Meta’s long-term strategy to cut down infrastructure costs as it heavily invests in AI-driven technologies. The report further states that Meta has projected expenses between USD 114 billion and USD 119 billion for 2025, with up to USD 65 billion allocated to AI infrastructure. 

The new chip, described as a dedicated accelerator, is expected to improve efficiency in AI training tasks compared to traditional GPUs.

Collaboration with TSMC and challenges in chip development

Meta is reportedly working with Taiwan Semiconductor Manufacturing Company (TSMC) to produce this AI training chip. The testing phase has already started after the company completed its first ‘tape-out’, which is an important step in chip design where an initial prototype is sent for manufacturing. However, developing custom chips comes with risks—Meta previously scrapped a similar project after encountering failures during small-scale testing.

Despite the past setbacks, Meta successfully introduced an AI inference chip last year (2024), which was used to optimize content recommendations on Facebook and Instagram. The company aims to extend its chip capabilities to support generative AI applications, such as its chatbot, Meta AI, by 2026.

Meta’s AI strategy and market impact

Meta remains one of Nvidia’s largest customers, having spent billions on its GPUs for AI training and inference tasks. However, as AI researchers debate the limits of scaling large language models, companies like Meta are increasingly looking for alternative approaches. 

 

ALSO READ: Apple’s iOS 19, iPadOS 19 and macOS 16 to get biggest redesign: Major UI overhaul and more

 

ALSO READ: iQOO 15 and iQOO Neo 11 series tipped to feature 2K display, 7,000mAh battery and more

Read all the Breaking News Live on indiatvnews.com and Get Latest English News & Updates from Technology
Advertisement
Advertisement
Advertisement
Advertisement
 
\