Meta Platforms is preparing to launch a new business unit named Meta Compute, which will offer its excess GPU and AI processing power to external clients. This move, reported on July 1, 2026, has led to a 9% jump in Meta's stock, effectively introducing a new revenue stream for the company.
What’s truly significant is Meta's planned capital expenditures for 2026, estimated to be between $115 billion and $135 billion. This staggering amount rivals the total value locked across all decentralized finance (DeFi) platforms at various points during the past year, signaling Meta’s serious commitment to AI infrastructure.
On March 11, 2026, Meta announced four new generations of its MTIA (Meta Training and Inference Accelerator) chips, with the MTIA 300 already in production. These chips are designed to handle various workloads, including training for ranking and recommendations.
Meanwhile, Meta is venturing into competition with established players like AWS, Google Cloud, and Microsoft Azure by providing raw infrastructure and hosted AI models. This competition could have repercussions for decentralized AI projects, which offer alternatives that are often cheaper and less censorable than centralized solutions. Projects like Render Network and Akash Network highlight the demand for distributed computing, emphasizing benefits such as permissionless access and privacy-sensitive workloads.
This material is for informational purposes only and does not constitute financial advice.
