Meta is rolling out custom-designed AI processors to reduce infrastructure expenses. The first chip launches in data centers next year, with a second model arriving by end of 2027.
Meta's latest processors are engineered to lower the company's substantial spending on AI infrastructure. By developing chips in-house rather than relying on third-party vendors, the company aims to improve cost efficiency while scaling its AI operations.
The first processor enters production next year as Meta deploys it across its data centers. A follow-up model is scheduled for completion by the end of 2027, suggesting a multi-year development roadmap.
Building proprietary chips aligns with Meta's broader strategy of controlling its computing infrastructure. The move mirrors efforts by other major tech companies to reduce dependence on external chipmakers and optimize performance for their specific workloads.
Meta has heavily invested in AI development, particularly for recommendation systems and generative AI applications. Custom silicon could give the company technical advantages while trimming operational expenses tied to AI model training and deployment.
U.S. data centers are projected to consume more natural gas than Germany and Japan combined by 2035, driven by surging demand from artificial intelligence infrastructure.
Google has announced a celebration event for October 5, with preorders opening September 21. The event details remain limited pending the official announcement.
Privacy-focused X post viewers Nitter and XCancel have gone offline following escalated legal action from X. The services allowed users to browse posts without creating accounts.
Amazon Web Services data centers in Abu Dhabi and Bahrain remain largely offline over six months after Iranian drone attacks damaged the facilities, disrupting the Gulf region's push into artificial intelligence infrastructure.