Broadcom and Google have signed long-term agreements covering custom AI processors and rack-level infrastructure, extending their collaboration on hyperscale AI systems through 2031. The deals formalize Broadcom’s role in both TPU development and the supporting networking stack used in Google’s next-generation AI deployments.
According to an SEC filing, Broadcom. will develop and supply custom Tensor Processing Units (TPUs) for Google LLC across multiple future generations. A separate supply assurance agreement commits Broadcom to provide networking and related components for Google’s AI racks over the same period, reflecting the increasing integration of compute and interconnect technologies in large-scale AI clusters.
The filing also details an expanded collaboration with Anthropic. Beginning in 2027, Anthropic is expected to access approximately 3.5 gigawatts of TPU-based AI compute capacity through Broadcom as part of a broader multi-gigawatt deployment. The level of consumption will depend on Anthropic’s commercial performance, and the companies are in discussions with additional operational and financial partners to support the infrastructure rollout.
- Broadcom to develop and supply custom TPUs for Google’s future AI systems
- Networking and AI rack component supply agreement extends through 2031
- Expanded collaboration includes Anthropic accessing ~3.5 GW of AI compute starting 2027
- Deployment linked to Anthropic’s commercial growth and partner participation
- Reflects tighter coupling of silicon, networking, and rack-scale design in AI infrastructure
“Broadcom and Google have entered into a Long Term Agreement for Broadcom to develop and supply custom Tensor Processing Units (‘TPUs’) for Google’s future generations of TPUs and a Supply Assurance Agreement for Broadcom to supply networking and other components to be used in Google’s next-generation AI racks through up to 2031.”
🌐 Analysis
The agreements underscore how hyperscale AI infrastructure increasingly depends on long-term partnerships that span custom silicon and system-level integration. Broadcom’s involvement across both TPU development and networking positions it within the core design loop for Google’s AI platforms.
The inclusion of a multi-gigawatt capacity framework tied to Anthropic reflects the scale at which AI labs are now planning infrastructure, where power, supply chain commitments, and integrated system design are coordinated in advance. Similar multi-year capacity agreements across the sector point to sustained demand for tightly coupled compute and interconnect architectures.
🌐 We’re tracking the latest developments in networking silicon. Follow our ongoing coverage at: https://convergedigest.com/category/semiconductors/