Ghost Raises $11M to Ship $3,499 AI Computer Core
Ghost emerged from stealth on October 5, 2026 after a year of building, announcing an $11 million seed round led by Andreessen Horowitz to ship Core, a $3,499 personal computer built for AI agents. TechCrunch
The company was founded by Zain Javaid, 19, who serves as founder and CEO. The founding team includes Nicholas Chua, 19, Yifei Chen, 24, and Gautam Sharda, 23. Other investors in the round include Abstract, Audacious Ventures, SV Angel, and Nova.
Core is priced at $3,499. Pre-orders for the first batch opened October 5, 2026. Shipping is scheduled for the last week of October.
The hardware choice is explicit. Core bundles an Nvidia RTX Pro 4000 SFF Blackwell GPU into the price. The box has no screen. It is accessed through a phone or app. Ghost builds the units at its San Francisco facility.
Software follows the same local-first logic. Core ships with three models preinstalled: Qwen-3.8-Next, Qwen-3.8-27B, and Gemma-4-31B. Users can install other models from Hugging Face or load their own trained models.
The broader context here is about where agent compute lives. Cloud APIs optimize for elasticity and shared capacity. A dedicated local machine optimizes for persistence, data locality, and direct control over the model stack. For engineers running long-lived agents that touch private files, credentials, and internal tools, that distinction matters.
In my view, Ghost is testing whether developers and power users will pay workstation money for sovereignty. The $3,499 price bundles compute that would otherwise be rented incrementally. The tradeoff is familiar from the early PC and workstation era. Upfront capital expense in exchange for predictable access and freedom to modify the stack without permission or metering.
Looking at what this means for builders, the open model policy is the part to watch. Preinstalled weights lower setup friction. Hugging Face installs and support for custom trained models keep the system open to iteration. That matters more than any single model checkpoint, since agent scaffolding changes quickly and teams need to swap models without swapping hardware.
Worth flagging for enterprise readers, a screenless, phone-managed form factor simplifies one problem and creates another. It removes peripheral management and fits an agent model where interaction is asynchronous and delegated. It also puts more weight on remote management, access control, and update hygiene, all standard concerns for always-on infrastructure, now moved into a home or office.
There is a longer arc worth keeping in mind. Each computing shift has moved from shared to personal when costs allowed, from timeshare to PC, from hosted servers to on-device mobile compute. Local agent hardware fits that pattern if model quality per watt keeps improving and if users value keeping context close. Ghost will not settle that question with a first batch. It gives engineers a concrete artifact to test it with, and that is often how useful categories start.
