Olares, an open-source personal cloud server and AI workstation manufacturer focused on data privacy, has unveiled its flagship device: the Olares One.
Millions of users interact daily with AI services such as OpenAI Group PBC’s ChatGPT, AI-powered search engines like Perplexity AI Inc., and image/video generators including Sora, Vidu, and Higgsfield Inc. While these third-party platforms deliver impressive performance, they require users to transmit personal or sensitive information over the internet.
“We believe Olares is entering a niche but high-potential market,” founder Peng Peng told SiliconANGLE. “Consumers want change but struggle to identify a clear solution—so they stick with the status quo.”
Today, Olares introduces an alternative: a privacy-first, open-source personal server that enables users to instantly deploy their own local AI assistant capable of accessing and utilizing personal files—keeping full control of their data in their hands.
The Olares One is engineered for everyday users to build and run AI models on powerful hardware without requiring deep technical expertise. As large language models become smaller and optimized for local deployment, intelligent and accurate on-device AI is becoming a practical reality. For instance, Qwen3-30B-A3B offers performance comparable to Deepseek-R1—a 671-billion-parameter model typically requiring high-end AI workstations or cloud infrastructure—yet it’s nearly 20 times more compact.
Peng emphasized that the Olares One is fundamentally designed as a “personal cloud solution, not just a personal computer,” enabling seamless access from smartphones, laptops, or tablets anytime, anywhere.
The AI assistant ingests files from the user’s PC and other sources, organizes them into a personalized knowledge base, and trains locally to understand daily routines, preferences, writing style, and workflows. Equipped with enterprise-grade hardware and production-ready open-source software, Olares allows users to extend their AI assistant’s capabilities through a marketplace of plug-and-play “skills” and services—no AI programming required.
Launched on Kickstarter, the device starts at $2,999, positioning it competitively among recently released AI workstations, including the NVIDIA DGX Spark personal AI supercomputer.
The Olares One features a high-performance NVIDIA GeForce GTX 5090 Mobile GPU with 24GB of GDDR7 memory and an additional 96GB of system RAM. According to the company, this robust configuration enables the device to run open-source large language models at speeds up to 1,824 teraFLOPS.
Built atop this hardware is an advanced GPU management system featuring a time-sharing mode that dynamically allocates compute resources among applications and AI tasks—maximizing efficiency while maintaining cool, quiet operation.
The open-source Olares OS adds another layer of flexibility: it can be installed on any AI-capable hardware with full functionality and zero vendor lock-in. This empowers advanced users to build custom setups and run the OS wherever they choose, ensuring continuity of access for those who prefer a DIY approach to hosting their AI assistant.
Why would consumers or developers consider a device like the Olares One? Beyond the obvious privacy advantages, current AI services operate on subscription-based models.
“In today’s AI era, a single user’s computational demand can be 100 times greater, forcing them not only to trade privacy for service but also to pay recurring fees,” Peng explained. “In this context, while the upfront hardware investment is higher, the total cost of ownership drops significantly over the device’s lifespan.”
Peng stated that the Olares One is built to disrupt the status quo—providing early adopters and current AI users with their own dedicated personal workstation and democratizing access to AI technology, models, and software ecosystems.
“We anticipate widespread adoption of the Olares One as model sizes continue to shrink through 2026,” Peng added.
