AMD chips target enterprise workloads
AMD managing director for Asia-Pacific Sales Peter Chambers highlights the company’s push toward AI-powered business computing as its new Ryzen AI Max and Ryzen AI Max PRO 400 Series processors bring advanced AI capabilities directly to personal computers.
Photograph courtesy of AMD
AMD is expanding its push into enterprise artificial intelligence with new Ryzen AI Max and Ryzen AI Max PRO 400 Series processors designed to run advanced AI models and automated tasks directly on business computers.
The processors allow companies to develop and operate AI agents locally, reducing the need to send every request or corporate document to cloud-based services.
Unlike conventional AI chatbots, which primarily respond to questions, AI agents can perform multistep tasks, analyze information and interact with applications with varying degrees of autonomy.
AMD said its latest processors combine up to 16 Zen 5 CPU cores, 40 GPU compute units and a neural processing unit capable of 55 trillion operations per second.
The Ryzen AI Max 400 Series also supports up to 192GB of unified memory, with as much as 160GB available to the GPU, allowing compatible systems to handle larger AI workloads.
According to AMD, the Ryzen AI Max PRO 400 Series can run certain AI models exceeding 300 billion parameters at 4-bit quantization without relying on cloud processing.
The company said local execution could help businesses retain greater control over sensitive documents, source code and proprietary information.
However, cloud services remain useful for tasks requiring more advanced models or computational resources.
AMD advocates a hybrid approach, allowing companies to process suitable workloads locally while using cloud-based models when necessary.
HP is adopting this approach with its ZBook Ultra G3a 16, which combines AMD’s Ryzen AI Max+ PRO 495 processor with Perplexity Portable Computer to support local and cloud-based AI operations.
Cost is another consideration.
AMD said companies running intensive AI workloads could reduce recurring cloud inference expenses by processing some tasks locally.
Its Tokenomics Calculator allows organizations to compare estimated expenses for local, cloud and hybrid deployments based on hardware costs, electricity consumption and anticipated usage.