Frames and Fans
Live
Handhelds & consoles

Microsoft AI PC Roadmap Defines Four Memory Tiers

Microsoft's AI PC roadmap categorizes hardware into four tiers based on memory, from 8GB entry-level machines to 128GB+ workstations and personal supercomputers, setting new standards for on-device AI capabilities.

Microsoft's AI PC roadmap categorizes hardware into four tiers based on memory, from 8GB entry-level machines to 128GB+...

Microsoft has defined a four-tier AI PC roadmap based on memory capacity and compute power. The plan establishes hardware requirements from entry-level devices to frontier systems, positioning Windows as the central platform for on-device AI.

Tier 0 Entry-Level Devices

Entry-level AI PCs start at 8GB of memory. These systems use basic processors like Intel Core Series 3 or 4, AMD Ryzen 100 or 200 series, and Snapdragon C, X, and X Plus systems-on-a-chip. They are priced around $500. Their AI capabilities are limited to basic tasks within browser tabs, email, and Microsoft Teams. Microsoft sees these 8GB machines as volume drivers for emerging markets, education, and commercial segments.

Tier 1 Mainstream AI PCs

Mainstream AI PCs require 16GB to 64GB of memory. This tier is defined by Copilot+ compliance. The 16GB threshold is specifically required for a device to qualify for the Copilot+ designation. Processors in this category include Intel Core Ultra Series 3 and 4, AMD Ryzen AI 400 and 500 series, and Snapdragon X2 Plus and Elite chips. High-end subsegments feature AMD Ryzen AI MAX+ 400, NVIDIA RTX Spark N1X, and Intel Core Ultra Next chips. These devices handle local cloud work for extended AI capabilities, including answering local questions and managing long-running tasks.

TierMemory RangeKey Processor ExamplesPrimary Use Case
Tier 08 GBIntel Core 3/4, AMD Ryzen 100/200, Snapdragon C/X/X PlusBasic tasks, emerging markets
Tier 116 GB - 64 GBIntel Core Ultra 3/4, AMD Ryzen AI 400/500, Snapdragon X2Copilot+ compliant mainstream AI
Tier 2128 GB+AMD Ryzen AI MAX+ 400, NVIDIA RTX SparkLocal model building and serving
Tier 3160 GB - 512 GB+Intel Crescent Island, NVIDIA GB300, AMD Threadripper HaloFrontier model development

Tier 2 Workstation Platforms

Workstation-class devices require 128GB or more of memory. This capacity enables local AI model building and external serving. Current hardware for this tier is represented by AMD's Ryzen AI MAX+ 400 and NVIDIA's RTX Spark. These platforms open new shipment channels for upcoming hardware like Intel's Crescent Island and NVIDIA's GB300. Most next-generation Tier 2 and Tier 3 platforms are still in planning or about to launch and are not yet in mass production.

Tier 3 Frontier Systems

The final tier consists of personal supercomputers for frontier AI model development. They use server-grade chips with massive memory allocations. Intel's upcoming Crescent Island GPUs scale from 160GB to 512GB of memory per GPU. NVIDIA's GB300 appears in its DGX Station workstation platform. AMD's contribution is the Threadripper Halo Station. Microsoft has not disclosed shipment targets for any tier, and market acceptance by OEMs and consumers remains untested. The company states that PC memory specifications are shifting from being 'good enough' to being 'priced by AI capability.' Microsoft has rolled back on some aspects of the Copilot+ ecosystem but maintains its emphasis on AI PC hardware development. The company said: "Windows is the home for hybrid intelligence."

Related coverage

More from Handhelds & consoles