Phaze's $8.75M Seed Turns Idle Workstations Into Shared Desktops for People and Computer-Use Agents
On November 4, 2026 Utah-based Phaze announced an $8.75 million seed led by a16z, with Indie, NextView, Makers, and Lerer Hippeau. The Parsec founders' remote-desktop stack now exposes an MCP server so agents can operate company-managed GPUs under existing IAM.
Phaze, a Utah remote-access company founded by the Parsec team, has raised $8.75 million in seed funding led by Andreessen Horowitz to treat high-end workstations as shared infrastructure for both employees and AI agents. Indie, NextView, Makers, and Lerer Hippeau participated, according to the November 4, 2026 company release. a16z general partner Martin Casado, who invested in Parsec and sat on its board through the Unity acquisition, published a same-day note explaining the bet: agents get more done when they receive full desktops and apps rather than thin APIs, and many of those apps are OS- or GPU-specific enough that remote access is the practical path.
Phaze’s product story has two faces on one pipe. For people doing 3D, design, and software work, it is a latency-sensitive remote desktop meant to make a distant workstation feel local. For agents, the same session model grants control of the remote machine’s graphical interface and installed applications without installing separate computer-use software on that box. The company recently shipped an MCP server so computer-use agents can connect through the Model Context Protocol instead of only through a custom integration. Agents inherit the user’s identity and access policies; enterprise authentication, directory sync, and audit logging are listed as first-class controls.
Why the desk PC becomes optional
CEO and co-founder Benjy Boxer’s company blog argues that IT still buys peak-demand workstations per employee even when those machines sit idle much of the day—especially for CAD, video, engineering, and game pipelines where specialized software needs serious GPUs. One manufacturer Phaze says it has spoken with runs about 30,000 workstations that can cost more than $10,000 each, implying more than $300 million of hardware sized for peaks rather than shared pools, before power and refresh. Agents multiply the problem: one employee may run several agents at once, each needing a computer of its own. Buying another dedicated box per agent does not scale. Phaze’s pitch is to assign and reclaim machines from a pool—on a desk, in a rack, or in a data center—while users and agents connect remotely.
Founders Benjy Boxer and Chris Dickson previously built Parsec, sold to Unity in 2021. The Phaze name nods to Piers Anthony’s dual-world Apprentice Adept series: people and agents working across machines at once. Casado’s note stresses interactive, graphics-heavy workloads—Blender, Unreal, and similar—where “local-like” remote performance matters, and says the stack was designed from the ground up for agent administration rather than bolted on later.
What agents can do on a managed machine
The release lists concrete patterns. An agent can work on a separate computer in the background while a person keeps their own laptop. Developers can point coding agents at machines that already have the right test images and IDEs. IT can let support agents reach employee machines under policy. People and agents can share one remote session, passing control back and forth: Phaze describes an early IT support agent that connects, diagnoses, attempts a fix, then hands the session to an authorized human and takes it back to finish. Funding will expand engineering, product, and customer growth; no valuation or ARR figures appear in the seed materials.
Independent latency percentiles, multi-tenant isolation benches, and third-party red-team results for the MCP path are not in the announcement. Competitive positioning against other remote-desktop and browser-sandbox agent hosts is implied by the Parsec lineage rather than tabulated. For buyers already drowning in idle GPU workstations and starting to let agents touch real apps, Phaze’s claim is narrower and operational: one access layer, existing IAM, shared hardware, humans and agents on the same glass.
- #Phaze
- #a16z
- #Remote Desktop
- #MCP
- #Computer Use
Author
Raj M
Contributor
AI Systems Architect is a seasoned technology leader with over 15 years of experience in the IT industry working with Fortune 500 companies. With a solid foundation in multi-agent systems, open-source LLM infrastructure, and enterprise deployment, he excels at building scalable production-grade AI platforms.