This New Local AI Mini PC Is Invisible (HP EliteBoard G2a)
HP's EliteBoard G2a packs a 50-TOPS AI PC into a keyboard for $3,028, but 16GB single-channel memory makes it poor for local LLMs compared to mini PCs. HP's EliteBoard G2a is a full Windows computer built into a keyboard, starting at $3,028 with an AMD Ryzen AI 5 PRO 435, 16 GB of memory and a 50-TOPS NPU. By AMD's own sizing guidance for its Ryzen AI laptops, 16 GB pairs with Gemma 3 4B: a small model that is fine as an offline assistant, but far from the reasoning powerhouse that price implies. Every HP configuration ships its memory on a single stick in a two-slot board, so it runs single-channel, and 64 GB is the ceiling. AMD's FAQ ties memory bandwidth to output speed, and the modules are customer-replaceable, so a second matching stick gives back the dual-channel bandwidth the system was designed for. The 50-TOPS NPU helps read your prompt in Lemonade's hybrid mode on Windows; it adds no memory and does not write the answer faster. For about $470 more, GMKtec's EVO-X2 with 128 GB and the Ryzen AI Max+ 395 holds models like gpt-oss-120b, a 59 GiB file no EliteBoard configuration can fit. HP's own store sells the first-generation G1a with the same 32 GB single stick, the same Radeon 840M graphics and the same 50 TOPS for $1,349, against $3,846 for the 32 GB G2a. Nobody has published benchmarks or thermals for the G2a yet: the film reads official specifications, AMD's guidance and store listings. Prices are store listings read on 16 October 2026. The EliteBoard's real strength is the form factor: a 670-gram computer under your fingertips, one USB-C cable to a compatible monitor, an optional internal battery, and customer-replaceable memory, storage and battery for hot desks and corporate fleets. For local AI, buy a big-memory mini PC instead, and if you already own an EliteBoard, add a second matching memory stick. Chapters: 0:00 Intro 1:49 What Fits 5:10 The NPU 6:47 The Price 8:24 Mini PCs 11:31 Who It's For 13:00 Verdict Tools & resources mentioned: - LM Studio - Ollama - Lemonade: https://github.com/lemonade-sdk/lemonade - llama.cpp About The Stack The Stack helps you build with AI. Each video takes one tool, model, or workflow and shows how it works in a few focused minutes, with the real benchmarks and real costs. We go deep on Claude Code and Cursor for AI coding, AI agents and MCP servers, the open-source AI tools and GitHub repos most people miss, RAG and vector search, fine-tuning, and running local LLMs on your own machine with Ollama and LM Studio. We compare models like ChatGPT and Claude, test AI automation with Zapier, Make, and n8n, and flag the tools that actually ship. Subscribe for new breakdowns: https://www.youtube.com/@the-stack-ai?sub_confirmation=1 #localai #minipc #ryzenai #llm #eliteboard
