BotBeat
...
← Back

> ▌

AMDAMD
PRODUCT LAUNCHAMD2026-05-21

AMD Launches Ryzen AI Max Pro 400 with Support for Up to 192GB Memory

Key Takeaways

  • ▸Ryzen AI Max PRO 400 supports up to 192GB of LPDDR5X-8533 memory, a 50% increase from the 300 series
  • ▸Flagship Ryzen AI Max+ PRO 495 receives modest CPU/GPU/NPU clockspeed boosts reaching 5.2GHz CPU performance
  • ▸Can allocate 160GB to GPU, enabling 300B parameter AI model inference on a single SoC
Source:
Hacker Newshttps://www.servethehome.com/amd-reveals-ryzen-ai-max-pro-400-series-192gb-ram-for-ai-systems/↗

Summary

AMD has announced the Ryzen AI Max PRO 400 processor family, building on the success of its Strix Halo silicon. The new chips feature support for up to 192GB of LPDDR5X-8533 memory—a 50% increase from the previous 300 series—enabling systems to load larger AI models for local inference. The flagship Ryzen AI Max+ PRO 495 SKU receives modest CPU, GPU, and NPU clockspeed improvements, achieving 5.2GHz on the CPU and reaching the new Radeon 8065S GPU branding.

The major appeal of the 400 series is memory capacity rather than raw compute power. With up to 160GB allocable to the GPU, these chips can now support 300B parameter models on a single SoC—a first for non-Mac Studio systems. However, AMD is not significantly increasing compute throughput or memory bandwidth, meaning existing bottlenecks like pre-fill performance will persist. Systems are expected to ship in Q3 2026, with AMD focusing exclusively on PRO SKUs and positioning these primarily as AI inference platforms for enterprise customers.

  • Only PRO SKUs launching with full enterprise management features; non-PRO variants not confirmed
  • Systems begin shipping Q3 2026; memory bandwidth increased ~7% to 273GB/sec

Editorial Opinion

AMD's memory-first refresh is a pragmatic response to growing demands for local AI inference, positioning the 400 series as a capable platform for running larger models at the edge. However, without corresponding increases in compute throughput or memory bandwidth, this is fundamentally a capacity upgrade rather than a performance leap—users should expect existing bottlenecks to persist. The exclusive focus on PRO SKUs boosts margins but limits the addressable market for what could be a compelling consumer-focused AI platform.

Generative AIAI HardwareProduct Launch

More from AMD

AMDAMD
RESEARCH

AMD MI355X Proves Competitive for Frontier AI Inference at 2.75x Lower Cost Than Blackwell

2026-07-03
AMDAMD
RESEARCH

Stanford Researchers Develop Multi-Agent AI System to Improve HIP Kernel Generation for AMD GPUs

2026-07-02
AMDAMD
PRODUCT LAUNCH

AMD Launches ATOM: Inference Engine Optimized for Instinct GPU Production Workloads

2026-06-16

Comments

Suggested

CloudflareCloudflare
OPEN SOURCE

Cloudflare Launches Agentic Inbox: Self-Hosted Email Client with Built-In AI Agent

2026-07-05
Multiple CompaniesMultiple Companies
INDUSTRY REPORT

AI Infrastructure Boom Triggers Hardware Price Surge Across Consumer Devices

2026-07-05
Stanford UniversityStanford University
RESEARCH

Stanford Researchers Advance HIP Kernel Generation Using Multi-Agent AI and Reinforcement Learning

2026-07-05
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us