Cerebras: fastest tokens will decide the next phase of AI

Cerebras says its Wafer-Scale Engine is more than 50 times larger than any prior chip and delivers 10–15x faster inference than other platforms. Chief strategy officer Andy Hock argues that in agentic workflows a single query spawns a family of daughter agents, making token speed the deciding factor.

• Cerebras Wafer-Scale Engine is over 50x larger than any prior chip
• Company claims 10–15x faster inference than other platforms
• Cerebras raised $5.5 billion in its May 2026 IPO
• Customers shift from token maxing to maximizing value per token