Mount Thor Raises $45 Million to Put Macs in the Data Center for AI Agents

Mount Thor has raised $45 million in equity funding to run Apple hardware as cloud infrastructure for AI research and agent deployment. The San Francisco startup sells what computer-use agents increasingly need: full macOS environments where they can open applications, learn, and carry out work over extended periods.

The financing spans Seed and Series A rounds. CRV and Conviction co-led the Series A; Chad Byers and Offline Ventures co-led the Seed. CSC Leasing, Enea Capital, Long Journey Ventures and Upper90 also participated, along with angel investors affiliated with Cognition, Core Automation, Daytona, Meta, OpenAI, Together AI and Vercel. Mount Thor separately raised non-dilutive capital for hardware purchases but has not disclosed how the $45 million splits between the rounds, or the size of that additional financing.

Why do AI agents need a Mac in the cloud?

An agent that navigates software needs more than a model endpoint. It needs somewhere to launch applications, inspect their state, manipulate files and resume a task when the next instruction arrives. For workloads built around Apple applications or development tools, that environment has to be macOS.

Mount Thor's infrastructure serves three target segments:

• research labs building reinforcement learning environments for computer-use models;
• providers offering macOS sandboxes;
• agent products that need persistent machines.

Apple software development and enterprise inference round out the target workloads. Each use case carries different demands: a training experiment might reset an environment repeatedly, a deployed assistant might need one that persists for weeks, and a development platform might provision machines in parallel to build and test code.

"Making Apple's operating system and silicon available at datacenter-scale is a massive untapped opportunity in AI infrastructure," said Peter Voell, Mount Thor's founder and CEO. Voell points to Apple's investment in accessibility and human-computer interaction as part of the platform's appeal for agents and the people supervising them.

What does the product actually look like?

Mount Thor's compute documentation describes two deployment options. Bare metal allocates a physical Mac exclusively to one customer. A virtual machine provides an isolated macOS environment on a Mac reserved for that customer, with two VM slots per host.

The split affects permissions too. The documented bare-metal account is a permissive non-root user; VM users get administrator access, making VMs the option for workloads that change system settings. Dedicated hardware matters when performance measurements need an entire machine or software won't run virtualized.

The VM service is currently in preview. Customers pick an image, create an instance and connect via SSH or remote desktop through private authenticated tunnels, with guest addresses kept off the public internet. For agent developers, that turns a Mac into a provisioned execution environment rather than a workstation someone sets up by hand.

How does management work at scale?

Mount Thor offers a portal, its mthr command-line tool and APIs. Its Kubernetes interface exposes VMs and bare-metal Macs as custom resources within a customer's tenant namespace, with actions governed by assigned permissions. Kubernetes here is a management layer for allocating Apple compute — it doesn't turn macOS into a Linux container.

The company also splits its offering into Mount Thor-managed fleets, where it prepares, monitors and wipes machines, and customer-managed deployments, where organizations run their own Macs at a Mount Thor site under their own tools and security model.

Where does the money go?

Mount Thor says it deployed its first data center in under six months. Its team includes people who previously built infrastructure at Cerebras, CoreWeave, Fluidstack, Nebius, OpenAI and xAI. The funding will back capacity expansion and hiring, with colocation relationships already in major U.S. metropolitan areas and Europe and Asia planned.

The company also plans an inference lab around open-source models optimized for Apple silicon, though the announcement offers no serving benchmarks or prices that would establish an advantage over other accelerators. The larger bet is simple: autonomous software will demand operating environments as well as model compute, and Mount Thor wants Apple infrastructure to be something developers allocate at scale instead of building the physical fleet themselves.