FULL STORY

Perplexity Portable Computer Expands from Linux to Windows with NVIDIA

Perplexity's fully local AI agent, Portable Computer, first landed on Linux and DGX Spark, then launched on Windows in partnership with NVIDIA to run locally on RTX PCs.

2026-09-04 ~ 2026-09-15 · 2 episodes · 10 posts

Episode 1 · Perplexity Portable Computer Lands on Linux and DGX Spark (2026-09-04, 3 posts)

Perplexity's fully local agent stack, Portable Computer, is now available on Linux with support for NVIDIA RTX GPUs with 24GB+ VRAM, and has also landed on NVIDIA's DGX Spark.

Episode 2 · Perplexity Brings Portable Computer to Windows RTX PCs for Fully Local AI Agents (2026-09-14, 7 posts)

On September 14, Perplexity officially announced that its local agent product, Portable Computer, is now available on Windows, jointly promoted with NVIDIA. The product runs harnesses, agents, and models locally on PCs equipped with NVIDIA GeForce RTX GPUs featuring 24GB or more of VRAM, as well as RTX PRO workstations. When handling local files and connected apps, tasks never need to be sent to the cloud—sensitive data stays on-device—while cloud resources can still be tapped when needed. This marks a major step in Perplexity's push to bring AI Agents from the cloud to local deployment, underscoring the value of data privacy and cost control.

Confirmed

  • Portable Computer is now available as a Windows app and can run inference on the user's own machine.
  • The hardware requirement is an NVIDIA RTX GPU with 24GB+ VRAM, covering GeForce RTX and RTX PRO workstations.
  • The product previously supported platforms such as DGX Spark; Windows is a newly added platform.
  • NVIDIA's official account confirmed the partnership; a Microsoft executive also shared the news, with @jeffgrimes9 relaying that as AI workload capabilities grow, Windows is becoming a platform for running AI locally—helping users get work done while keeping control of their data and cutting costs.

Why it matters

  • Local execution means sensitive files and connected-app data never have to be uploaded to the cloud, directly addressing privacy and compliance concerns for both enterprises and individual users.
  • The 24GB+ VRAM requirement sets a clear user threshold, reflecting on-device agents' reliance on high-end consumer GPUs—and could fuel demand for premium RTX cards and workstations.
  • With NVIDIA and Microsoft ecosystem players joining the push, local agent execution is shaping up as the next battleground for PC platforms.