ain Javaid, the 19-year-old founder and CEO of Ghost, recently secured $11 million in venture funding to build a hardware wedge against the entire cloud-based artificial intelligence industry. Ghost's maiden product, Core, is a dedicated desktop computer designed to run local AI agents entirely on your desk, bypassing external subscriptions and corporate data harvesting. According to the company, it connects directly to your apps, files, and local devices, builds context about your life over time, and operates continuously rather than waiting for you to prompt it.
At its core—pun intended—the machine is an unassuming black box built around an Nvidia RTX Pro 4000 graphics card with 24GB of video RAM, paired with 64GB of system RAM. With a rated performance of 770 TOPS, the hardware is explicitly over-engineered to handle complex local workloads that would choke a standard consumer laptop. Out of the box, the PC comes pre-loaded with three open-weight models: Qwen-3.8-Next, Qwen-3.8-27B, and Gemma-4-31B. You can swap these out for your own models if you are so inclined, but the entire pitch rests on avoiding the agonizing headache of manual configuration.
Why Local AI Commands a Luxury Price Tag
The fundamental appeal here is simple: escaping the monthly subscription treadmill and keeping your digital life out of corporate server farms. Ghost is targeting users who want intelligent agents managing their schedules, files, and smart home gadgets without broadcasting every personal query to the cloud. You interact with the PC using an app on your phone or laptop, issuing voice commands whenever you prefer. This means the computer quietly builds context about your daily routines while locked down behind local encryption that even Ghost claims it cannot access.
Core updates its models automatically as new upgrades emerge, unless you don't want that to happen, and the idea is that it's a seamless system and similar to how Tesla works.
Who Is Actually Buying a Garage-Built AI Rig?
If you already own a maxed-out Mac mini or a fully spec'd Studio, spending nearly three and a half grand on a dedicated AI box might feel wildly redundant. But for enthusiasts who refuse to mess around setting up complex open-source weights and terminal environments by hand, this turnkey approach solves a very real friction point. As Zain Javaid points out, while you certainly can run large language models on existing consumer hardware, the out-of-box experience is a miserable multi-step chore.
Whether a niche garage startup can sustain hardware manufacturing at scale remains an open and expensive question, but the early market signal is deafening. Pre-orders for the first batch of PCs sold out on the very first day they went live. Priced at $3,499 in the US, Core is an aggressive, unapologetic gamble on the future of private, unmetered personal intelligence for those with deep pockets and a healthy distrust of the cloud.