Cerebras launched a new chip designed to accelerate AI training and inference. This hardware release coincides with the GLM-5.3 API launch and reports of cybersecurity slowdowns at OpenAI. Engineers can now leverage higher throughput for large-scale models. These updates highlight a tightening race between custom silicon providers and established model labs.