Meta Eyes In-House AI Training Chips
Synopsis
Meta Platforms is developing custom AI chips designed to train artificial intelligence models, expanding its internal semiconductor program. The initiative builds on the company’s MTIA accelerator chips already used to run AI workloads across its platforms. The move comes as Meta increases spending on data centers and computing infrastructure. Technology companies worldwide are investing heavily in semiconductor development to support growing demand for artificial intelligence computing.
Meta Platforms is developing custom AI chips designed to train artificial intelligence models as it expands its internal semiconductor program. The initiative builds on the MTIA accelerator chips already used for AI workloads in Meta’s data centers. The move comes as global technology companies increase investment in AI computing infrastructure.
Key Highlights
- Meta is developing custom AI chips designed to train artificial intelligence models across its platforms.
- The initiative expands Meta’s MTIA chip program currently used for recommendation and inference workloads.
- Meta reported $200.97 billion revenue in 2025 with major spending on AI infrastructure and data centers.
- Global technology companies and governments are increasing investment in semiconductor development for artificial intelligence.
Meta Platforms is moving forward with plans to develop custom AI chips designed to train artificial intelligence models, expanding its internal semiconductor program as demand for computing power continues to rise across the technology industry.
The company’s effort builds on its Meta Training and Inference Accelerator (MTIA) initiative, a chip program already used in the company’s data centres to run certain AI workloads such as recommendation systems across its social media platforms. By expanding this program, Meta aims to design processors capable of training future AI systems used in its products.
Expanding the Meta AI chips program
Training AI models involves processing large datasets to allow systems to learn patterns and improve predictions. This process requires powerful hardware, typically graphics processing units (GPUs) supplied by companies such as Nvidia and AMD.
Meta has previously deployed MTIA chips for AI inference, which refers to running trained models to deliver results like content recommendations. The new effort focuses on chips capable of handling AI training, a more computing-intensive stage.
The custom processors are expected to complement, rather than replace, hardware purchased from existing chip suppliers as Meta continues expanding its AI infrastructure.
Rising AI infrastructure spending
The push to build Meta AI chips comes as the company increases investment in data centres and computing capacity needed for artificial intelligence systems.
Meta reported $200.97 billion in revenue for 2025, with total costs and expenses of $117.69 billion, according to the company’s latest financial results. Capital expenditure reached $72.22 billion, reflecting significant spending on data centres, servers and AI hardware.
The company has indicated that infrastructure investment tied to artificial intelligence will remain a major component of future spending.
Global competition in AI semiconductor development
Meta’s move reflects a wider trend in the technology sector, where companies are developing proprietary processors to support artificial intelligence workloads.
Major technology firms in the United States, including Google, Amazon and Microsoft, have launched their own AI chips for cloud computing and data-centre operations. Taiwan remains a key global hub for advanced semiconductor manufacturing, while South Korea hosts major memory-chip producers.
Governments are also investing heavily in domestic chip industries. The European Union has introduced semiconductor funding programs, while India and other countries are promoting local manufacturing to reduce reliance on imported chips.
Industry estimates suggest the global AI chip market could exceed $500 billion by the early 2030s, driven by demand for computing power needed to develop and deploy artificial intelligence systems.
Quick FAQs
Q1. Why is Meta developing custom AI chips?
Meta is building custom AI chips to train artificial intelligence models more efficiently and support growing computing demand across its platforms.
Q2. What is Meta’s MTIA chip program?
MTIA, or Meta Training and Inference Accelerator, is Meta’s in-house chip designed to run AI workloads such as recommendation systems.
Q3. Will Meta stop using Nvidia and AMD chips?
No. Meta’s custom AI chips are expected to complement existing GPUs from suppliers like Nvidia and AMD.
Q4. How much is Meta spending on AI infrastructure?
Meta reported capital expenditure of about $72 billion in 2025, largely driven by investments in AI data centers and computing hardware.
Follow Inspirepreneur Magazine for the business news.
Pooja Malik is a business journalist with over six years of experience covering startups, entrepreneurship, and emerging trends. She has previously worked with leading media platforms such as YourStory Media and BW BusinessWorld, where she reported on business, policy, and market developments. Currently, she serves as Editor at The Inspirepreneur Magazine, where she writes and edits stories across business, lifestyle, and travel, with a focus on clarity, accuracy, and reader relevance.
