HP Z2 Mini G1a

The AMD Ryzen AI Max+ PRO 395 platform brings CPU, GPU, and NPU together around a single, unusually large unified memory pool — and the HP Z2 Mini G1a packages it into a workstation-class mini PC built for continuous business use. For SMBs looking to run document processing, coding assistants, or internal chat tools on substantial open-weight AI models without routing every query through a metered cloud API, the Z2 Mini G1a offers on-site inference at a fixed hardware cost, backed by HP's workstation validation and remote-management tooling.

Hardware Specifications

At its core sits the AMD Ryzen AI Max+ PRO 395: 16 Zen 5 CPU cores (32 threads, boosting to 5.1 GHz, with 64 MB of L3 cache) alongside an integrated AMD Radeon 8060S GPU and an XDNA 2 NPU rated at 50 TOPS. All three engines draw from a shared 128 GB LPDDR5x-8533 unified memory pool across a 256-bit bus, delivering roughly 256 GB/s of bandwidth without the CPU-to-GPU copy overhead of a discrete graphics card. Storage is handled by up to 4 TB of PCIe Gen4 NVMe SSD across dual M.2 2280 slots (plus a dedicated M.2 2230 slot for WLAN), and the entire system fits into an 8.55 × 16.8 × 20 cm, 2.3 kg chassis — small enough for a desk, dense enough for a rack.

128 GB Unified Memory: Run 100B+ Class Models Without Per-Token Billing

Because CPU, GPU, and NPU address the same memory pool, up to 96 GB can be assigned to the GPU for model weights and inference, while the remainder stays reserved for the operating system and background applications — adjustable without a reboot via AMD's Variable Graphics Memory setting. This is enough headroom to host 100B+ parameter class open models — including Meta Llama 3, OpenAI GPT-OSS, and Mixtral — locally, for as many employees and queries as the business needs.

For everyday interactive use — support chatbots, document Q&A, code assistance — mid-sized models run comfortably at conversational speed. The same 128 GB pool also accommodates a single 200B-class model for heavier reasoning tasks, run in batch or overnight rather than turn-by-turn, without ever leaving the building or incurring accumulating per-query charges.

Built for Continuous Business Use

A 300 W internal power supply lets up to five HP Z2 Mini G1a units fit into a single 4U rack space with no external power bricks — a practical path for SMBs standardizing on one dedicated local AI workstation per employee, or consolidating several into a small server closet. Four selectable performance profiles — High Performance, Performance, Quiet, and Rack Mode — let the same hardware suit a shared office desk or a rack deployment. Every unit is validated through over 360,000 hours of HP testing and MIL-STD 810H durability standards.

HP ZCentral Remote Boost lets a single Z2 Mini G1a serve as a remote workstation for several part-time users.

AMD XDNA 2 NPU: On-Device Business AI Acceleration

Independent of the CPU and GPU, the integrated XDNA 2 NPU delivers 50 TOPS of dedicated AI acceleration — enough to meet Microsoft's Copilot+ PC certification requirements.

Software, Security & Management

The Z2 Mini G1a ships with Windows 11 Pro and is Linux-ready (Ubuntu 24.04 LTS, WSL2). Business-grade protection comes from HP Wolf Pro Security and HP Sure Click, with HP BIOSphere Gen6 firmware protection and TPM 2.0 included as standard, while HP Performance Advisor, HP Image Assistant, and HP Manageability Kit 10 simplify fleet-wide deployment and monitoring.

Summary

For SMBs evaluating the AMD Ryzen AI Max+ 395 category for local AI hosting, the HP Z2 Mini G1a combines the platform's defining strength — a large, flexible unified memory pool — with workstation-grade durability, rack-ready density, and enterprise management tooling, at an affordable price.