Zain Javaid Launches Core: A Local AI Computer

Overview

Zain Javaid has unveiled Core, a new class of personal AI computer designed to run large language models entirely on-device. The launch marks a significant pivot in consumer technology, emphasizing local processing over cloud-dependent services. By keeping data within the home, Core aims to address growing privacy concerns while delivering high-performance AI capabilities directly to individual users.

The device is positioned as more than just a hardware accessory; it is intended to be the central node for a user’s digital life. According to Javaid, the most powerful technology humanity has built should belong to the individual, not the corporation hosting the servers. With orders now open through ghost.ai, Core enters a market increasingly defined by the tension between convenience and data sovereignty.

Technical Specifications and Performance

Core is built around a robust internal architecture that balances computational power with energy efficiency. At its heart lies an AMD Ryzen 5 7600 CPU, paired with 64GB of DDR5 RAM to handle multitasking and complex model inference simultaneously. Storage is provided by a 1TB NVMe M.2 SSD, ensuring fast read/write speeds for immediate access to local files and model weights.

The graphical processing unit is the centerpiece of the system: an NVIDIA RTX PRO 4000 Blackwell SFF GPU. This professional-grade, small-form-factor card is capable of running substantial open-weight models locally. During initial benchmarks, Core achieved inference speeds of over 87 tokens per second on Qwen-3.8-27B and 32 tokens per second on the larger Qwen-3.8-Next model.

Core runs local models 24/7 entirely on-device. No data ever leaves your home.

These performance metrics place Core in a competitive tier for desktop AI workstations. However, its value proposition extends beyond raw speed. The device is designed to operate continuously as a always-on assistant, leveraging the Blackwell architecture’s efficiency to maintain low power consumption while delivering sustained inference rates.

Ecosystem and Connectivity

One of Core’s defining features is its connectivity layer. Users interact with the device through their existing digital habits, connecting via laptop, a dedicated mobile application, or even iMessage. This seamless integration ensures that the AI assistant feels like a natural extension of the user’s current workflow rather than a separate, cumbersome tool.

The software ecosystem supports popular custom harnesses such as OpenCode, OpenClaw, and Hermes. This flexibility allows developers and advanced users to build custom AI agents and workflows on top of the Core infrastructure. The device acts as a local inference engine, powering these applications without requiring an external internet connection for every query.

Javaid emphasized that Core is not merely a passive storage device. It functions as a full computer with its own file system, browser, and action-taking tools. The AI can browse the web, retrieve information, and execute tasks autonomously. In a recent demonstration, Core was shown researching moving companies in San Francisco, aggregating their contact information, and making outbound calls to negotiate availability and pricing on behalf of the user.

Privacy and Personal Context

The primary selling point of Core is its commitment to local-first privacy. Unlike cloud-based assistants that transmit personal data to remote servers for processing, Core keeps all information local. This architecture ensures that sensitive conversations, personal files, and behavioral data never leave the user’s home network.

Beyond privacy, the device excels at contextual awareness. Core aggregates data from various apps and devices to build a comprehensive understanding of the user’s life. For instance, it can analyze sleep patterns from a Whoop device and cross-reference them with calendar entries to identify trends. If the AI detects that late meetings consistently correlate with poor sleep quality, it can proactively suggest schedule adjustments to improve well-being.

This level of personalization is achieved without third-party data brokers. The AI learns from the user’s own digital footprint, creating a tailored intelligence that becomes more useful over time. By maintaining complete ownership of this context, users avoid the surveillance capitalism model that characterizes most current AI products.

Market Position and Availability

Core is available for order at a price point of $3,499. This places it in the premium segment of the personal computing market, targeting professionals, developers, and privacy-conscious individuals who require high-performance AI capabilities. The initial batch of orders represents the first phase of production, signaling a direct-to-consumer launch strategy similar to other recent hardware ventures in the AI space.

The announcement has generated significant interest, with Javaid’s initial posts garnering hundreds of thousands of views and extensive engagement. Supporters have highlighted the potential for Core to democratize access to advanced AI tools, allowing individuals to run models that previously required enterprise-grade infrastructure.

We believe that AI will do incredible things for people. Our goal is to create an entire ecosystem around personal intelligence you own.

Zain Javaid and the Ghost team view this launch as the beginning of a broader movement. They anticipate a new generation of consumer software and hardware built specifically around personal, local AI. As cloud computing faces increasing scrutiny over data privacy and latency, Core represents a tangible alternative for those seeking to reclaim control over their digital interactions.

The response from early observers has been largely positive. Critics of the current cloud-dominated AI landscape have welcomed the shift toward on-device processing. However, the premium price tag may limit adoption to early adopters and enterprise users initially. As production scales and the ecosystem matures, the hope is that such devices become more accessible to a broader audience.

In conclusion, Zain Javaid’s Core represents a bold statement in the evolving AI hardware sector. By combining powerful local inference with a strict privacy-first philosophy, it offers a compelling vision for the future of personal computing. Whether it can sustain momentum beyond the initial release remains to be seen, but it has undeniably shifted the conversation toward user ownership and data autonomy.