Semiconductor Architecture for Extreme Scale AI in Manufacturing Operations

Scaling and operationalizing AI for a wide variety of applications, such as manufacturing, requires massive computing resources. Large manufacturing processes often involve hundreds or thousands of data sources and billions of data points. For successful implementation, organizations need solutions that are powerful and can quickly train models, but that are also simple to operate and scale. Current infrastructures using clusters of GPUs to train large AI models can take weeks or months, inhibiting innovation, time to deployment, and delayed inference. Andy Hock, VP of Product Management at Cerebras Systems, will discuss the critical need for dedicated AI compute architecture and the impact it will have on innovation, enabling organizations to rapidly iterate on large AI models and drive positive business outcomes, including predictive yield, optimized supply chains, and automated quality assessments. He will also share how Cerebras has addressed sustainability in its approach to developing AI technology, creating a system that is both power and space-efficient, and enabling data centers to run large models in less time, at less power.