Leader in ISG Private Cloud Report 2026.

UnitedLayerBlue Version
Your AI factory
In 3U.
UnitedEdge® AI puts frontier models, virtual machines, containers and
serverless in one appliance at your site. Your data stays put. Your cost
stays fixed.
Download Brochure
Contact Us
The UnitedEdge AI 3U rack appliance
1 TB of AI memory
Two of everything
Zero prompt data off-site
One fixed monthly price
1 TBAI memory
64Compute cores
3U19-inch rack
1 visitFrom crate to serving
Why the edge
AI is moving to where your
data lives.
01
Data can’t leave
Patient records, plant telemetry, trading data and IP are bound by regulation and contract to stay on site. Public AI APIs are off the table.
02
API bills never stop
Per-token pricing grows with every new use case. The more value AI creates, the more it costs, and nothing is ever owned.
03
No one on site to run it
Branches, plants and clinics have no AI or infrastructure team. Whatever goes there must run unattended and be fixed remotely.
The answer is a private AI factory small enough for the
edge, and autonomous enough to run itself.
The appliance
Two of everything. So
nothing has to stop.
Two AI Engines and two Compute Engines, side by side in one 19-inch chassis, running
active-active.
UnitedEdge® AI · 3U chassis

Side A active

AI Engine A512 GB AI memory
Compute Engine AVMs, pods, functions
Power feed AIndependent circuit

Side B active

AI Engine B512 GB AI memory
Compute Engine BVMs, pods, functions
Power feed BIndependent circuit
Clustered NVMe · software-defined network · 4×25 GbE LAN and WAN
1 TBAI memory across two AI Engines
64Cores in the Compute Engines
1 TBECC RAM for VMs, pods and functions
2×3.84 TBClustered NVMe
4×25 GbELAN and WAN
~1.4 kWPeak, on two independent feeds
3U19-inch rack · 23 in deep
Zero-touchEnrolls itself on first boot
Included in every appliance
Three clouds in one box.
Two AI Engines and two Compute Engines, side by side in one 19-inch chassis. Everything is
doubled, so nothing has to stop.
High-performance inference
Frontier and open models on 1 TB of AI memory, served as private, OpenAI-compatible endpoints.
Frontier LLMsVisionSmall modelsPrivate API
Cloud native at the edge
Run your apps where the data is: VMs, Kubernetes and serverless, highly available across both Compute Engines.
VMsKubernetesServerlessHA
Software-defined networking
Virtual firewalls, load balancers and routers, configurable per tenant and highly
available.
FirewallLoad balancerRouterPer tenant
Use cases
One box. Five front lines.
Wherever your data lives, AI can now live there too. Here is what UnitedEdge® AI does
on site, live
Use cases

🔊 Voiceover
UPC Vektor console
Run one site or hundreds
from one screen.
Follow a day in the life of an edge AI fleet. Each scene is the real console. Click any
screen to see it full size.

    vektor.unitedlayer.com/fleet

    Resilience

    Built to run with no one on
    site.
    Go ahead, try to break it. Cut a power feed, fail an engine or pull the WAN cable, and
    watch what the appliance does.
    UnitedEdge AI – Resilience simulator
    All services healthy · endpoints answering

    Side A active

    servingAI Engine A512 GB AI memory
    runningCompute Engine AVMs, pods, functions
    onPower feed AIndependent circuit

    Side B active

    servingAI Engine B512 GB AI memory
    runningCompute Engine BVMs, pods, functions
    onPower feed BIndependent circuit
    WAN uplink · LTE backup standing byprimary
    100%AI capacity on site
    0Failed requests
    00:00Session uptime
    Value
    One fixed monthly price.
    Unmetered tokens.
    As AI use grows across your sites, your cost stays the same. Model your own numbers.

    Sample figures to show the shape of the economics, not a quote. Your UnitedLayer team prices the subscription from a sizing workshop.

    –Public API cost, 12 months
    –Fixed subscription, 12 months
    ––
    Public API equivalentUnitedEdge® AI subscription
    Autonomous operations
    The edge cloud
    that runs itself.
    The SRE Orchestrator finds and fixes issues at every site, with people approving by policy. AI
    Coworker answers in plain language, acts and updates tickets.
    0 %

    L1/L2 incidents closed by AI

    0 %
    Faster to resolve
    0 X
    Engineer productivity
    0 %
    Staff hours saved
    SRE Orchestrator
    Turns millions of raw events into a handful of investigations, then plans, checks blast radius and remediates within guardrails. Anything outside policy waits for a human.
    AI Coworker
    Ask “Why did DEN-Plant-02 page at 14:07, and is it fixed?” and get the answer, the evidence and the follow-up ticket, in seconds.
    Outcomes reported in customer deployments of UnitedLayer’s autonomous operations platform.
    Security
    Sovereign by design, not by
    policy.

    Data stays on site

    Models, prompts, outputs and embeddings never leave the appliance.

    Encrypted at rest

    Hardware-sealed keys, secure boot and signed updates.

    Zero-trust access

    Your identity provider, role-based access and a full audit log.

    Isolated tenants

    Compute, storage, network and AI separated per business unit.

    Tamper-aware

    Lid, bezel and shock sensors raise alerts in the console.

    Recoverable

    Backups to an immutable cyber vault, with DR in UnitedLayer facilities.

    Deployment
    Live at your first site in a
    single visit.
    1

    Size and order

    Use cases, models and workloads agreed in a workshop.

    2

    Ships ready

    Pre-imaged, burned in and tested, in a shock-rated crate.

    3

    Plug in

    Two power cords and two network handoffs per side.

    4

    Enrolls itself

    Joins the console, pulls your policies and starts serving.

    In the crate

    Everything it needs to go live.

    One shock-rated crate, ruggedized to ship: every connector locked, every heavy component shock-mounted.

    • The appliance3U, pre-imaged and burned in
    • Rack rails19-inch, 23 in deep
    • Dual powerTwo cords per side
    • Lights-out kitRemote console, power control, LTE backup
    Stories from the edge
    Where it pays off first.
    UnitedEdge AI – Stories from the edge

    Stories are illustrative scenarios, not specific customers.

    Questions
    Good to know.
    UPC Edge AI – FAQ
    Which models can it run?

    Open-weight frontier, vision, coding, speech, industry and small models from US, EU and China publishers, curated, security-scanned and license-reviewed. With 1 TB of AI memory you can run two frontier models plus many small ones. Licenses vary by model.

    Do my apps need to change?

    No. Endpoints are OpenAI-compatible. Change one URL and your code keeps working, with failover to the peer AI Engine and then to your core AI Factory.

    What happens if a part fails?

    Both sides run active-active. Lose a feed, an engine or a cable and service continues; firewall, load balancer and AI endpoints fail over in seconds.

    Who operates it?

    Your choice: self-run, partner-managed or UnitedLayer-managed. Autonomous operations closes most routine incidents without a person on site.

    How is it priced?

    Appliance and platform in one fixed monthly subscription, with managed operations optional. Tokens are unmetered: no per-token or per-seat charges.

    How do we start?

    A workshop to pick two or three AI workloads and the VMs to consolidate, a pilot with one appliance at one site, then scale across sites under one subscription.

    ↑