AMD’s new Ryzen AI Max and Ryzen AI Max PRO 400 Series processors enables businesses to run artificial intelligence (AI) agents on their own computers, with less need to send work and company data to cloud services.
Systems using the processors are now available from AMD partners. The company is pitching them to businesses developing AI agents that can handle tasks across several applications.
Unlike chatbots that mainly respond to questions, AI agents can look up information, review documents, and take authorized actions. A single assignment may involve several rounds of processing, adding to the bill when a business pays a cloud provider for each use.
Running some of that work on a company computer could help manage those expenses. It also allows businesses to process internal files on their own equipment.
Whether that makes financial sense depends on how much AI a business uses, what the tasks require, and the cost of buying and operating the computers.
The Ryzen AI Max 400 Series supports up to 192GB of memory, up from 128GB in the previous generation. Of that total, up to 160GB can be assigned to graphics processing, which also handles AI workloads.
The additional memory gives AI agents more room to work with lengthy documents, several AI models, and other applications running at the same time.
AMD says the Ryzen AI Max PRO 400 Series can run models with more than 300 billion parameters in a compressed format without using the cloud. Parameters are the values an AI model uses to process information.
The processors support Windows and Linux, as well as tools developers use to build and test AI applications.
HP is using the AMD Ryzen AI Max+ PRO 495 processor with Perplexity Portable Computer in its HP ZBook Ultra G3a 16. The computer can run AI agents alongside professional applications and call on cloud models for steps that need more advanced capabilities.
That setup allows businesses to choose where each task runs. Work involving internal documents, engineering files, or source code can stay on the computer when the local model can handle it. Other tasks can go to a cloud service.
AMD has also introduced a Tokenomics Calculator to estimate the costs of running AI locally, in the cloud, or through a combination of both.
The calculator considers usage, hardware requirements, and cloud pricing. Its comparisons use simplified assumptions; actual costs will vary with equipment, electricity rates, and workloads.