Skip to content

Business

Mount Thor raises $45 million to expand its Mac cloud for AI

Edited and reviewed by AI Ping Editorial Team

3-line summary

  1. Mount Thor raised $45 million across its seed and Series A rounds.
  2. Approved teams can rent a dedicated Mac or an isolated macOS virtual machine remotely.
  3. The startup plans to expand its data center capacity and study running open models on Apple silicon.
PracticalDevelopers
Fictional compact silver desktop computers on server racks
AI-generated conceptual image: Racks of compact computers represent cloud infrastructure built on Macs. · AI Ping

Mount Thor said on November 12 that it had raised $45 million across its seed and Series A rounds to expand a cloud built on Apple silicon. The startup rents remote Macs to teams that develop AI agents or test software in macOS. CRV and Conviction co-led the Series A, while Chad Byers and Offline Ventures co-led the seed. The company also raised separate financing to buy hardware without diluting shareholders. It did not break out the amounts for the two equity rounds or the size of the equipment financing.

Developers can request a complete physical Mac or create an isolated macOS virtual machine on rented hardware. That gives teams doing Mac-specific work a way to run tests in parallel without buying and setting up a fleet themselves. Mount Thor’s quickstart guide says it reviews access requests individually and requires a prepaid balance before customers can create managed machines. The funding announcement describes an expansion of existing operations.

What a rented Mac gives developers

A physical Mac is reserved for one customer for the duration of the lease. Mount Thor’s documentation lists performance measurements that need dedicated hardware and software that cannot run inside a virtual machine among the reasons to choose it. The account can run tools and workloads. System settings are outside that account’s permissions, and releasing the machine wipes its contents.

A virtual machine provides an isolated installation of macOS with administrator access inside the guest. That makes it suitable for tests that change system settings or run background services. The company reserves a host Mac for the customer, with room for two virtual machines. A second guest can run in the open slot, sharing the physical Mac’s resources. Deleting a guest removes its disk as well.

A compact silver desktop computer beside two clear blocks
AI-generated conceptual image: A desktop computer and two clear blocks represent two virtual machine slots on one Mac. · AI Ping

Mount Thor bills dedicated Macs in 24-hour windows. Returning a computer early ends access, with charges for the window already underway still applying. Its virtual-machine documentation says the second guest is included in the charge for the host Mac. For both products, the company places a hold on the customer’s prepaid balance when it provisions a machine. The funding announcement gives no rates for individual configurations.

Organizations can also arrange for their own Macs to be housed at a Mount Thor site. In that arrangement, the provider runs the facility and tracks inventory while the customer manages the operating system and software. Documentation for the service lists serial-number inventory, problem reports and notifications among capabilities available under a customer’s agreement. The customer-managed service does not give teams the on-demand virtual machines available under the managed service.

Where the new funding will go

Mount Thor describes training computer-use models and testing software for Apple platforms as separate uses for the same kind of infrastructure. A training environment gives a model practice operating windows, menus and other parts of a desktop. A software testing environment lets developers run the same feature in multiple setups and examine how it behaves. Both can require more Mac environments than a team keeps available on its own desks.

A person using a keyboard in front of two blurred screen panes
AI-generated conceptual image: Two panes on a monitor represent software testing in remote macOS environments. · AI Ping

The company also plans to put some of the new funding toward research on running already-trained, openly available models on Apple chips. Founder Peter Voell pointed to the M5 Ultra as hardware for that work. Apple lists up to 512GB of unified memory for the M5 Ultra Mac Studio. That specification describes Apple’s hardware. Mount Thor disclosed no speed measurements or number of machines for the planned inference work in its funding announcement.

Mount Thor says its first data center took less than six months to deploy. The announcement gives no location or opening date for the next phase of expansion, and no date for the inference lab.

Sources