Ghost Secured $11 Million for Personal AI Computer

The San Francisco-based company announced a local-first AI device called Core launching for $3,499.

Updated on Oct. 5, 2026 in Artificial Intelligence

Bold flat-color editorial illustration showing a minimalist, matte-finish aluminum computing unit, representing local artificial intelligence hardware. — San Francisco startup Ghost secured $11 million in a seed funding round led by Andreessen Horowitz to develop its Core local AI hardware device. AI Illustration. Upload story photo >

Live Poll

Would you pay for personal AI hardware to keep your data off the cloud?

Ghost emerged from stealth this week after raising $11 million in a seed funding round led by Andreessen Horowitz. The startup also unveiled Core, a personal computer designed to run AI models locally without the need for cloud infrastructure.

Why it matters

The company aims to provide users with complete ownership over their AI models and personal context. By running inference on local hardware, Ghost seeks to offer a privacy-focused alternative to cloud-dependent artificial intelligence.

The Core device features an NVIDIA RTX PRO 4000 Blackwell GPU, 24 GB of GDDR7 memory, 64 GB of DDR5 RAM, and a 6-core AMD Ryzen 5 7600 processor. The machine is rated at 770 AI TOPS and includes a 1 TB NVMe SSD.

The players

Ghost

A San Francisco-based technology startup focused on private, local AI computing.

Andreessen Horowitz

A prominent Silicon Valley venture capital firm that led the seed funding round for Ghost.

Zain Javaid

The CEO of Ghost who is overseeing the launch of the new AI hardware.

The details

Core connects directly to apps and devices to perform tasks locally, supporting models like Qwen 3.8-27B and Gemma 4-31B. The system is designed to keep data and processing within the device rather than relying on external cloud servers.

October 5, 2026: Ghost emerged from stealth.
October 31, 2026: Shipments of the Core computer are expected to begin.

The Tech Race

Ghost is entering a crowded landscape where companies are increasingly betting on local hardware to challenge the dominance of cloud-based AI providers. By utilizing NVIDIA's Blackwell GPU architecture, the firm is attempting to bridge the gap between enterprise-grade processing and home computing.

Users will be able to perform AI tasks locally, which may improve data privacy and reduce latency compared to cloud-based alternatives. However, the $3,499 price point and the complexity of local model management present significant barriers for the average consumer.

The takeaway

This development highlights a growing trend toward privacy-centric computing where users retain control over their digital environment. Prospective buyers should consider whether their specific AI workloads justify the high cost of dedicated local hardware compared to cheaper cloud subscriptions.