Ghost AI's $11M Bet on Personal AI Hardware: A 19-Year-Old's Vision for Privacy-First Computing

Ghost AI, an artificial intelligence hardware company, announced it has closed an $11 million funding round to develop what the startup describes as the first personal computer built specifically for running personal AI agents. Andreessen Horowitz led the investment, with participation from Abstract, Audacious Ventures, Nova and SV Angel.

The startup was founded by a group of young entrepreneurs, with 19-year-old Zain Javaid serving as Chief Executive. Over the past year, the team has been developing a high-powered device called Core that operates without requiring a monitor or peripheral devices. The machine functions as what Ghost describes as a "brain in a box," engineered exclusively to run AI agents that operate on behalf of the user.

Ghost's founding team contends that the prevailing approach to AI agents—deployed on cloud servers managed by large corporations—presents fundamental problems. Since AI agents require access to substantial volumes of user data, including sensitive personal information, they demand considerable trust from users. In a post on X, Ghost argues that an AI agent with deep knowledge of your life should "not live on infrastructure someone else owns."

Javaid explained to TechCrunch that he and his co-founders identified a market opportunity for a robust, purpose-built, locally-hosted device that people could deploy at home to power the AI agents managing their digital activities. With this approach, AI agents can continuously gather information about users, process their private data and execute tasks entirely offline, never transmitting any information to cloud services.

https://x.com/ghostai/status/2107162749675721173?ref_src=twsrc%5Etfw

Core's Technical Specifications

Core is equipped with an Nvidia RTX Pro 4000 SFF Blackwell graphics processing unit, providing substantial computational resources for personal agents. The device includes its own software layer, file system and web browser.

At launch, Core ships with several open-source models, including Alibaba Group Holding's Qwen-3.8, Qwen-3.8-2.7B and Google LLC's Gemma-4-31B. Users can also install additional models downloaded from Hugging Face and similar repositories.

All of the source code, logic, and model weights live on the device, not the services. We build this device to not be dependent on having an outside entity maintaining the service.

Zain Javaid

Since Core lacks a display or conventional operating system like Windows or Ubuntu, users interact with it through a dedicated smartphone application and web interface. This control center allows users to communicate with their personal agents via text or voice, issue instructions, delegate tasks and pose questions. The device supports automatic updates that Ghost delivers wirelessly when connected to the internet, similar to how Tesla Inc. manages software updates for its vehicles.

Privacy and Security as Core Differentiators

Ghost emphasizes security and privacy as the primary advantages of Core. Because all models run locally and user data resides on the device, Ghost implemented full encryption. The user maintains the encryption keys, preventing Ghost and any other third party from accessing the machine's contents. The device incorporates a firewall designed to monitor all outgoing requests from the user's agents, preventing unauthorized data transmission and malicious activities.

Javaid highlighted that the device was engineered for durability. Since all source code, logic and model weights operate on the hardware itself, Core would continue working indefinitely even if Ghost ceased operations. This durability guarantee may prove important given Core's price point—the initial batch of devices, which sold out quickly, retailed for $3,499 each.

Ghost is now accepting registrations from users interested in purchasing the second batch of devices. The company has not announced when it will begin accepting orders for the next production run, but given that the first batch sold out within hours and the startup now has several million dollars in funding to manufacture additional units, scaling production appears to be the immediate focus.