Nvidia RTX Spark PCs Bring Local AI to Lenovo and Acer Windows Systems

Nvidia says Lenovo and Acer will introduce the first Windows PCs using its RTX Spark chip in October 2026. The announcement is part of Nvidia’s effort to move more AI inference and agent workloads from remote data centers onto laptops and compact desktop PCs.

What Nvidia announced

The October systems are expected to use RTX Spark in slim laptops and compact desktops. Nvidia positions the platform for AI enthusiasts, developers and creators who want to run capable agents locally and securely, while also supporting graphics-intensive creation and gaming workloads. 15

Lenovo and Acer join a broader group of PC makers tied to the RTX Spark rollout. Nvidia had previously identified systems from ASUS, Dell, HP, Lenovo, Microsoft Surface and MSI, with Acer and GIGABYTE models to follow. 11

What is inside RTX Spark?

RTX Spark combines an Arm-based Nvidia Grace CPU with Nvidia Blackwell RTX graphics in a single processor platform. Nvidia specifies configurations with up to 128GB of unified memory and claims up to 1 petaflop of FP4 AI performance. 6

That combination matters for local AI because model inference needs both compute and memory. Nvidia’s stated aim is to put enough of both into thinner Windows laptops and efficient small desktops for personal AI agents, content creation and gaming. 56

Why Nvidia is pushing AI onto the PC

Nvidia frames RTX Spark as a PC built around personal agents rather than conventional app-by-app computing. Running workloads locally can keep prompts, files and agent context on the device; Nvidia’s PAIR software is designed to route AI work across local RTX and DGX Spark systems while preserving that local context. 9

The practical benefits will depend on the specific software and model. A system’s useful local-AI capability is shaped by the model’s size and quantization, available memory, framework support, and whether the task is inference, training, graphics work or agent orchestration. Nvidia’s petaflop rating is therefore a vendor performance specification, not a like-for-like result against every competing computer.

The Microsoft and MediaTek connection

RTX Spark is the outcome of Nvidia’s roughly three-year collaboration with Microsoft to “reinvent” the Windows PC for the AI era, according to Nvidia and Reuters. The partnership is intended to give developers a Windows path to build, tune and run agents across local and cloud environments. 14

MediaTek is Nvidia’s development partner for the processor platform. MediaTek says the joint work combines its CPU, connectivity and power-efficiency expertise with Nvidia’s AI platform and RTX technologies for slim laptops and small desktops.

Together, these relationships explain the product’s strategy: Nvidia is not launching a single branded PC line, but enabling a Windows ecosystem of OEM hardware built around local AI workloads.

How RTX Spark compares with Apple’s new Mac mini and Mac Studio

Apple is pursuing the same broad local-AI opportunity through its own integrated desktop lineup. Its latest Mac mini is offered with M6 or M5 Pro chips, while the refreshed Mac Studio uses M5 Max or M5 Ultra; Apple says its new chips and software frameworks let developers run and fine-tune large AI models locally on a Mac. 17

The difference is principally one of platform and go-to-market. RTX Spark is designed for Windows hardware from multiple manufacturers and pairs Nvidia’s Grace CPU with Blackwell RTX graphics. Apple controls the hardware and macOS stack for the Mac mini and Mac Studio. Both approaches emphasize local models and agents, but their real-world results cannot be inferred from headline specifications alone. 617

Bottom line

The Lenovo and Acer launch gives Nvidia a concrete October 2026 entry point into Windows PCs designed around local AI agents. RTX Spark’s Grace-and-Blackwell design, up to 128GB of unified memory and Nvidia’s claimed 1-petaflop FP4 capability are meant to bring substantial AI processing closer to the user. Whether it becomes a meaningful alternative to Apple’s AI-focused Macs will depend on shipping configurations, pricing, software compatibility and independent performance testing. 6