What HP’s RTX Spark PCs Are And Why They Matter
HP’s RTX Spark PCs are a new line of Windows laptops, desktops, and workstations that combine NVIDIA’s Grace Blackwell-based RTX Spark superchip with pre-configured software environments so developers can run advanced AI workloads locally instead of relying on cloud infrastructure. Built around a 20‑core Arm CPU and Blackwell RTX graphics with 6,144 CUDA cores, RTX Spark delivers up to 1 petaflop of local AI computing performance and supports up to 128 GB of unified, high‑bandwidth memory, bringing data center‑style throughput into portable and deskside systems. HP is aiming these machines at developers building agentic applications, local AI agents, and hybrid workflows where models and tools must stay close to the data. For many teams, this could reduce latency, improve privacy, and simplify experimentation with large models and complex pipelines.
Inside the RTX Spark Platform: Petaflop AI for Local Computing
At the heart of HP’s new AI developer workstations is NVIDIA’s RTX Spark platform, which is effectively a Windows-ready version of the NVIDIA GB10 Grace Blackwell superchip found in the DGX Spark mini AI workstation. The chip pairs a 20‑core Arm CPU with Blackwell-based RTX graphics, providing up to 1 petaflop of local AI processing performance and unified memory up to 128 GB for demanding transformer and diffusion workloads. This design targets local AI computing where inference, fine‑tuning, and multi‑agent pipelines run directly on the device. According to Liliputing, major PC makers including Asus, Dell, HP, Lenovo, Microsoft, and MSI plan to ship RTX Spark PCs starting this fall, positioning the platform as a common foundation for high‑end AI developer systems. For HP, this means aligning its portfolio to an ecosystem that spans consumer notebooks through to compact desktops.
OmniBook Ultra 16, OmniBook X 14, And Compact Desktops For Builders
On the client side, HP is bringing NVIDIA RTX Spark to the OmniBook Ultra 16 and OmniBook X 14, two thin Windows notebooks aimed at creators, gamers, and AI developers who want strong local AI performance without losing battery life or mobility. HP and partner reports describe these as among the thinnest RTX Spark PCs, offering a mix of high-performance graphics, petaflop‑class AI acceleration, and all‑day usage for coding, training small models, and running local agents. HP is also preparing a compact RTX Spark‑powered desktop (reported as an OmniDesk Mini or similar deskside configuration) that mirrors the laptop hardware for creators and AI enthusiasts who prefer a stationary setup. These systems are meant to be plug‑and‑play AI developer workstations, with HP highlighting pre‑configured environments and open‑source toolchains to cut down initial setup time.
ZGX Fury GB300 And ZGX Nano: Enterprise-Grade AI Developer Workstations
Beyond mobile RTX Spark PCs, HP is extending its AI developer workstations with deskside systems built around NVIDIA Grace Blackwell GB300 technology. The coming HP ZGX Fury GB300 will use a GB300 Grace Blackwell Ultra Desktop Superchip and add Windows support later this year so enterprise teams can build and run always‑on frontier AI agents inside existing Windows workflows. HP executives note that over 70 percent of enterprise PCs run Windows, so aligning GB300 systems with that environment is key for adoption. For highly regulated environments, HP’s ZGX Nano focuses on secure, local AI computing with a zero‑trust approach. It combines tightly integrated hardware and software, physically restricts wireless and some external interfaces, and is intended for sectors where AI workloads must run locally with minimal attack surface.
What Developers Can Expect This Fall And Beyond
For AI developers, HP’s RTX Spark PCs signal a clear move toward petaflop AI performance on the desk and in a backpack. Windows laptops and desktops using RTX Spark are expected this fall from HP and other OEMs, with HP’s OmniBook Ultra 16 and OmniBook X 14 slated for release later in 2026 and a compact RTX Spark desktop planned for August 2026. HP positions these machines as open, flexible tools: pre‑configured environments, open‑source toolchains, and compatibility with NVIDIA’s full‑stack AI software should shorten the path from prototype to production. Meanwhile, Grace Blackwell GB300‑based ZGX Fury and secure ZGX Nano systems give larger organizations a local alternative to cloud for frontier models and sensitive workloads. Together, the lineup is designed to let developers build, test, and deploy AI agents and hybrid pipelines without leaving the local machine.
