Value for business
A private AI assistant, local data processing, and predictable costs without expensive external APIs.
Backed by EuroHPC · LUMI AI FactoryKickstarter — November 2026
Your AI. Your data.
Your hardware.
The DeAI Node is not just a device for local AI — it is an entry point into a new infrastructure where users simultaneously gain privacy, autonomy, access to more powerful models, and the ability to monetize idle computing resources.
Two modes. One node.
DAI is a network of autonomous AI devices where users can choose to run tasks locally and privately, or connect to the DAI Network for heavy models, deep research, content generation, or model training.
No cloud required
Run AI locally
Private, powerful, and always available on your own node.
Run powerful AI models directly on your DAI Node. Your AI stays local and private - with no API limits and no data leaving your device.
Deploy autonomous assistants that can read your files and interact with your tools while keeping every byte on your device. Your documents, source code, and business data are never uploaded to the cloud.
The compute is already yours. Run the node all day and generate as many tokens as you like — nothing per token, nothing per request, no surprise invoice at the end of the month.
Chat with your documents without uploading them to the cloud. Search, summarize and analyze your private files locally.
Install DAI-optimized models or any other supported AI models with a single click - no command line, complex setup, or manual configuration required.
Around 80 W power draw makes local AI efficient and renewable-energy friendly.
Up to ~1 PFLOP
FP4, theoretical
128 GB LPDDR5x
Unified memory
CPU + GPU
Coherent unified memory
~273 GB/s
Memory bandwidth
Distributed access
Access AI through the network
Tap decentralised AI capacity, advanced learning modes, and lower-cost large-scale workloads.
Large AI models are split across multiple DAI Nodes. Each node processes its part of the model and passes compact intermediate results to the next node. Together, the network can run models that are too large for any single machine.
DAI Network gives you access to our own 2.4-trillion-parameter model - built to match the capabilities of the world’s leading open AI models.
Train and improve AI models across the DAI Network without sending your private data anywhere. Nodes share only what the model has learned, while sensitive data and processing can stay on your own device.
Green Mode can reduce the cost of scheduled AI workloads by up to 125× compared with traditional API-based inference.
Your DAI Node doesn’t have to sit idle. Let it contribute spare computing power to the DAI Network and earn rewards by processing AI tasks, hosting model layers, and supporting other users.
Distributed LLM
Access without a data centre
Federated + split
Training without raw data
Up to 125x
Green Mode savings
Idle → income
Monetise spare capacity
Switching modes is a physical switch on the node, fully under your control.
Value for cities & infrastructure
Local AI services that keep running through poor connectivity or outages.
Value for energy
Monetise surplus green energy through digital services and EV charging.
Intelligent P2P routing, no single point of failure
Requests automatically reroute through available nodes, keeping the network fast, resilient, and always connected.
We are building AI infrastructure that does not need a data centre
DAI is a decentralised AI infrastructure where computation is not performed in one central data centre, but distributed across autonomous AI devices, local nodes and energy-independent clusters.
A single node gives you private, local-first AI. Many nodes together create something a single device cannot: infrastructure for large models, served by splitting the work across the network. That is the part most local-AI products stop short of, and it is the reason DAI exists.
DAI offers an alternative to full dependence on cloud AI providers: its modularity and P2P architecture enable scaling AI capacity without building traditional data centers, while energy independence reduces bureaucratic barriers and simplifies deployment.
DAI d.o.o. is registered in Celje, Slovenia, and develops its intelligent layer on European public supercomputing infrastructure.
AI lives closer to the user, and the network's power emerges through the cooperation of many Nodes.
The node always maintains local control: what to run locally, what to allow the network to process, which models to install, which adapters to keep private.
Local infrastructure ensures full control over data, resilience against external interference, and independence from global providers.
With the support of the European Union and EuroHPC.
DAI received a EuroHPC allocation on LUMI, one of Europe’s fastest supercomputers.
This grant turned an architecture that previously existed only on paper into real MVP nodes and enabled experiments that can help bring the DAI Network to scale.
Thanks to access to supercomputing resources, the team is working on:
- Building the Distributed Model Layer: splitting large AI models across multiple nodes and enabling recovery when one of those nodes goes offline.
- A network simulator with thousands of virtual peer-to-peer devices: testing unstable connections, node churn, and partial network outages.
- Federated and Split Learning experiments across heterogeneous nodes: including devices running on battery power.
This non-dilutive R&D support validates the project as an innovative AI infrastructure initiative with significant potential for further development — not simply an early-stage commercial idea.
We built our first working off-grid DAI Node — powered by its own battery and rechargeable from solar panels.
With Starlink connectivity, it can operate privately and locally or connect to the DAI Network, keeping AI available even during power outages.
The same low-power, renewable-energy setup can also help make Green Mode compute cheaper across the network.
Where we are
-
March 2026DAI d.o.o. founded
DAI d.o.o. is established in Slovenia to continue the development and commercialization of decentralized AI infrastructure.
-
Summer 2026EuroHPC allocation on LUMI
DAI receives access to LUMI through the LUMI AI Factory / EuroHPC program. Work begins on the decentralized DAI Network, distributed AI architecture, and DAI's own AI models.
-
Summer 2026First autonomous DAI Node MVP
The first working autonomous DAI Node prototype is built — operating on battery power, rechargeable from solar panels, and capable of running AI locally or connecting to the network.
-
Summer 2026Distributed Model Layer on LUMI
Large-scale R&D begins on splitting large AI models across multiple nodes, automatic failure recovery, network simulation, and Federated & Split Learning across heterogeneous devices.
-
November 2026Kickstarter campaign launch
DAI Node becomes available for crowdfunding. Early supporters and waitlist members receive access to special launch pricing and early hardware reservations.
-
January 2027First commercial DAI Nodes
Production of the first commercial DAI Nodes begins, moving the project from MVP hardware toward real-world deployment.
-
Q1 2027Early adopter deployment
The first DAI Nodes are deployed with early users, developers, and pilot partners, providing real-world feedback on local AI and network workloads.
-
Q1 2027DAI Network Beta
Early DAI Nodes begin connecting into a shared peer-to-peer compute network, enabling distributed inference, resource sharing, and network-based AI workloads.
-
2027Green Mode & Compute Marketplace
Node owners can contribute spare compute capacity, while users can schedule non-urgent workloads when renewable or lower-cost compute is available.
From one autonomous AI node to a global decentralized AI network.
Kickstarter — November 2026
Join the waitlist, get the launch discount
DAI Node goes live on Kickstarter this November. Join the waitlist to unlock your exclusive launch discount and be the first to know the moment the campaign opens.