LongCRBSJune 30, 2026

Baron Fifth Avenue Growth Fund on Cerebras Systems Inc.

Thesis

The Fund initiated Cerebras because it believes fast AI inference will become increasingly valuable as AI shifts toward reasoning and agentic workflows. Cerebras' wafer-scale architecture keeps more calculations on-chip and can generate tokens substantially faster than leading GPU-based approaches across several open-source models. Commercial validation includes a multi-year 750MW OpenAI agreement valued at more than $20 billion and a partnership with Amazon to bring Cerebras inference to AWS. The Fund believes wafer-scale computing represents a durable moat because Cerebras spent nearly a decade solving difficult architecture, packaging, yield, and software challenges that competitors have not commercialized at comparable scale.

We believe Cerebras' moat is durable because wafer-scale computing is difficult to replicate
Baron Fifth Avenue Growth Fund — Baron Fifth Avenue Growth Fund Quarterly Letter

Key risks

  • The thesis depends on agentic AI adoption continuing to grow and customers placing increasing value on inference latency.
  • The company must continue executing on a technically difficult wafer-scale computing architecture.