Mount Thor, a startup that rents Mac hardware to people building AI agents, said on Nov. 12 that it has raised $45 million in equity funding, led by CRV and Conviction, across its Seed and Series A rounds.
The company did not say how the money divides between the two rounds, and it declined to disclose a valuation. It also announced separate non-dilutive financing, meaning capital that does not require giving up ownership, to pay for hardware purchases. No amount was given, according to the company's announcement.
Who put in the money
Other participants included CSC Leasing, Enea Capital, Long Journey Ventures and Upper90, the announcement said. Offline Ventures lists Mount Thor on its portfolio page, though it discloses neither an amount nor terms.
The company also named a group of strategic angel investors affiliated with Cognition, Core Automation, Daytona, Meta, OpenAI, Together AI and Vercel, without identifying the individuals. The announcement does not establish that those employers invested, and the affiliations should not be read as evidence of customer relationships or partnerships.
Public databases do not agree on how to classify the round. Fundz calls the full amount a Series A, while Raised500 calls it a Seed. Neither matches the company's own description of a combined raise.
One backer summed up the bet in the announcement: "We're thrilled to back Peter and the team as they make it easier to use Apple's platforms at scale for agentic AI." The announcement as reviewed does not tie the quote to a named investor.
What Mount Thor sells

On its website, the company markets complete macOS environments for training, testing and deploying agents. These include persistent environments and data generation for reinforcement learning in which an AI system learns to operate a computer. All of this is the company's own description; no independent deployment figures have been published.
The documentation is more concrete. Mount Thor offers bare-metal Macs, meaning customers get the physical machine rather than a shared slice of it, as well as macOS virtual machines. Managed compute is charged per Mac in 24-hour windows, and customer-managed fleets are billed under contract. Customers can connect by SSH or a graphical desktop. Releasing a bare-metal machine wipes it, and deleting a virtual machine permanently deletes its disk.
Mount Thor also plans to offer enterprise inference, the running of trained models for end users, using open-source models optimized for Apple silicon. Unite.AI reported that the announced inference lab is a development priority rather than a source of published benchmark results. Coverage reviewed by Citybiz named no paying customers and gave no contract values or customer counts.
The founder

Peter Voell is the founder and chief executive, Citybiz reported on Nov. 12. The Information described him in an Aug. 30 article as a former OpenAI employee who worked on compute infrastructure. On the day of the announcement he published the "Mount Thor Manifesto," which lays out the company's thesis on macOS agents and enterprise inference.
In it, Voell argues that running Macs at cloud scale takes fleet management, networking, scheduling and virtualization capabilities well beyond putting machines in racks. That is a statement of the demand he expects, not a measured market outcome.
Rivals and signs of demand
Mount Thor is entering a field with established players. MacStadium has published material on its Orka tooling for agents, and the EC2 documentation of Amazon Web Services lists several Apple-silicon configurations.
On demand, MacStadium's 2026 DevOps and AI survey found that 26% of respondents named procurement and lead times among their biggest infrastructure challenges. Aaron Tilley of The Information discussed large Mac purchases by OpenAI and other labs on the outlet's TITV program on Aug. 31, and described Mount Thor as a new "neocloud," the industry's shorthand for specialized GPU and compute rental firms, contributing to enterprise Mac demand.
What's next
Mount Thor has not said how it will spend the money, nor has it offered a timetable for additional facilities or a capacity target. Its website mentions Tier 1 colocation providers, the highest-grade data center operators, without naming them or their locations. The company maintains a careers page on Ashby, a recruiting platform, but its listings could not be retrieved for this report.
