OmnyaAI

An emerging neocloud

A new foundation
for agentic AI.

AI factories.
Agentic software.

We’re building an AI neocloud designed around the NVIDIA Cloud Partner reference architecture, and working toward NVIDIA Cloud Partner status. Project Apollyon, our planned AI factory in Albania, is designed for up to 1 GW.

Talk to our team Our NCP path
Abstract architectural layers connected by orange pathways, representing the relationship between infrastructure and software. Three pathways are labelled agentic software, neocloud and AI infrastructure. Agentic software→Software layer Neocloud→Compute layer AI infrastructure→Physical foundation
Select a layer to explore itConceptual illustration
Our path NVIDIA Cloud Partner program
In progress NCP status
Inference & training Intended workloads
Up to1GW Project Apollyon, planned

Our approach

From physical capacity
to software potential.

Our direction connects the infrastructure AI needs with the software that puts it to work.

Design basis

Purpose starts at the base.

We’re designing our infrastructure around NVIDIA’s Cloud Partner reference architecture: accelerated computing, a high-speed network fabric and closed-loop liquid cooling.

See the platform

The platform

Designed to the NVIDIA
reference architecture.

NVIDIA’s Cloud Partner reference architecture is a full-stack blueprint for AI clouds. We’re designing each layer of OmnyaAI around it, from accelerated infrastructure to the services customers use.

  1. Agentic software

    OmnyaAI software that puts models to work on real tasks.

    OmnyaAI

  2. AI platform services

    Tenant-facing services for training, fine-tuning and inference.

    Platform · PaaS

  3. Managed Kubernetes

    Containers and orchestration for multi-tenant AI workloads.

    Containers · CaaS

  4. Bare metal and virtual machines

    Dedicated GPU servers and on-demand provisioning.

    Infrastructure · IaaS

  5. Accelerated infrastructure

    NVIDIA GPU systems, a high-speed Ethernet and InfiniBand fabric, and closed-loop liquid cooling.

    Hardware design

Layers follow the NCP reference architecture. OmnyaAI services are in design; availability to be announced.

Our NCP path

On the path to NVIDIA
Cloud Partner.

NVIDIA Cloud Partners deliver AI services built on NVIDIA’s reference architecture. This is where OmnyaAI stands on the way there.

  1. Design to the reference architecture

    Platform layers, networking and cooling designed against NVIDIA’s Cloud Partner blueprint.

    In progress

  2. Pursue NCP status

    Work through the requirements of NVIDIA’s Cloud Partner program.

    In progress

  3. Build and validate

    Deploy the first clusters and validate them against the reference architecture.

    Planned

  4. Open to customers

    GPU compute for training and inference, with agentic software on top.

    To be announced

OmnyaAI is not currently an NVIDIA Cloud Partner. Status and timing depend on the requirements of NVIDIA’s program.

Project Apollyon

An AI factory for
inference and training.

Project Apollyon is our planned AI datacenter in Albania: up to 1 GW of capacity for running and training AI at scale.

Project Apollyon

Up to

1GW

Planned AI factory capacity

AlbaniaPlanned development

Inference at scale

Capacity to serve models in production, from real-time applications to agentic workloads.

Training at scale

Large GPU clusters on a high-speed network fabric for training and fine-tuning models.

Built as an AI factory

Power, accelerated computing, networking and liquid cooling designed together, around the NVIDIA Cloud Partner reference architecture.

Up to 1 GW is planned capacity for Project Apollyon. Sites, phasing, power supply and availability are subject to development and have not been confirmed.

Agentic software

Software, beyond
the infrastructure.

Agentic software is part of our direction, alongside the neocloud.

We intend to provide software that puts AI to work. The offering is being defined around the needs of future users.

Talk about your software needs

Product scope and availability to be announced.

An early look

From financial questions to working models.

Your agent builds the analysis. You explore the possibilities.

Powered by Drax Agent (Harness)

Drax Agent (Harness) running in the desktop app. Actual footage, edited for pacing. Illustrative data.

Questions

Clear from
the beginning.

Is OmnyaAI an NVIDIA Cloud Partner?

Not yet. We’re working toward NVIDIA Cloud Partner status and designing our platform around NVIDIA’s reference architecture. We’ll describe OmnyaAI as an NVIDIA Cloud Partner only once that status is granted.

What is the NCP reference architecture?

NVIDIA’s full-stack blueprint for AI cloud providers. It covers the hardware and networking, and the software layers for bare metal, managed Kubernetes and AI platform services, for large-scale training and inference.

Is OmnyaAI compute available today?

No. The OmnyaAI neocloud is in development. Capacity, pricing and availability will be announced when they’re ready.

What is Project Apollyon?

Our planned AI factory in Albania, with up to 1 GW of capacity for inference and training. It is in development: sites, phasing, power supply and availability have not been confirmed.

Is the 1 GW capacity available today?

No. Up to 1 GW is the planned capacity of Project Apollyon. It is not live capacity or an offer of immediately available compute.

What can we discuss with OmnyaAI?

Future compute requirements, the training or inference workloads you plan to run, and needs that could inform the agentic software offering. Specifications and commercial terms require direct discussion.

Is agentic software available now?

Not yet. Agentic software is part of OmnyaAI’s intended direction, and the finance demo above is an early look. Product scope, release dates and availability have not been announced.

Get in touch

Let’s start with
what you need.

Talk to us about future compute requirements
or the software you want to put to work.

Email OmnyaAI