VIRTUAL CONNECTED EDGE

Meet the Cube.

Your on-site AI engine for real-world operations—anywhere, at any scale.

Bring AI to your worksite.

The Cube is a real-world AI engine for physical environments—factories, hospitals, labs, power plants, and other mission-critical facilities. It connects directly to your environment and runs on-prem, tapping into data from your machines, devices, and on-floor software. The Cube bridges the gap between software in the cloud and your real-world sites, enabling a highly secure environment for your AI to work with your data.

A solution that works for you.

Connects with your data, equipment, sensors locally. Meets your constraints. Compatible with your software tools and ecosystem.

Deploys in under an hour. It just works.

A plug-and-play complete, autonomous stack frees your team to focus on business value.

Fully operational offline.

Built for resilience and works seamlessly in intermittent connectivity or disconnected state and automatically syncs when back online.

Scales to your entire footprint.

The Cube dropships to your sites – whether one or thousands – and deploys with zero technical overhead.

Fully local + fully private = data sovereignty.

Data, logic, and models stay inside your facility and implemented with industry-leading security principles—no exceptions.

PRIVATE AI

Working for you—and you alone.

Private and dedicated inference, running leading models or your proprietary models, enabling data sovereignty and optimized performance. The more you use the Cube, the better it works.


The Cube uses a fast, efficient, vLLM runtime and OpenAI-compatible API. Models are deployed for you with and include native guardrails to keep results safe and aligned, benchmarking to measure performance, and embedded MCP servers and third-party tools to bring your data seamlessly into your workflows.

DATA MANIFOLD

All your data, how you need it.

The Cube brings all your data together. It connects, aggregates, and organizes information from any source — without complex pipelines or data engineering — and makes it readily usable by your applications.


The Data Manifold operates a local cache of data, leveraging multiple brokers and databases and a leading implementation of modern data principles for transparency, consistency and lineage. The Data Manifold connects to hundreds of common data sources autonomously without the need for endless data piplines and adopts your preferred formats and schemas.

EDGEPORT

Close the loop on every issue.

One-click deployment for apps or agents — from your ecosystem partners or developed in-house — on the Cube.


Apps, models and agents are seamlessly deployed from local or cloud-based development environments and operated on the Cube for you. The Cube enables a next-generation AI-assisted software development lifecycle with built in testing, evaluation. and monitoring.

DATA SECURITY & GOVERNANCE

Security and control by design.

All data, logic, and models remain fully on-site. The Cube uses a zero-trust architecture purpose-built for critical infrastructure, with complete auditability and operational control.


  • Immutable environment
  • Secure boot (TPM 2.0)
  • Encryption in transit (TLS 1.2/1.3) and at rest (AES-256)
  • Default-deny firewall
  • Micro-segmented networking
  • Intrusion detection & DoS protection
  • Signed firmware & software updates
  • Continuous vulnerability scanning
  • Comprehensive audit logging

your cube ai co pilot

Hello, Samantha.

Your operating system agent. Built into the Cube and always on — running the platform for you and helping you get things done effortlessly. Designed to maximize reliability and embody responsible engineering.

What Samantha does:

Interactive

Ask for system configuration, check status, or request clear, actionable performance reports.

Monitoring

Continuous oversight of system health, diagnostics, and anomaly detection — with end-to-end support.

Delivering

Maximizes uptime with system performance engineered for 99.9%–99.99% availability or higher.

Operating

Handles updates, improvements, and integrations with your approval — connecting seamlessly to third-party systems and data sources.

Product Specifications

Built for real work.

Cube appliances deploy in under one hour, require no external dependencies, and operate offline in your physical environments.


Operating System

  • RHEL base
  • Red Hat CoreOS with hardened, immutable Kubernetes
  • Apollo continuous delivery
  • Embedded operators for health & flow monitoring

Networking

  • Ethernet: 1 × 2.5 Gbps WAN port (RJ45)
  • Cellular: 5G NR (n1, n2, n3, n5, n7, n8, n12, n20, n25, n28, n38, n40, n41, n48, n66, n71, n77, n78, n79); LTE Cat 20 (B1–B71); 3G fallback B1, B2, B5, B8
  • Wi-Fi: Wi-Fi 6 (802.11ax), dual-band 2.4/5 GHz, 2×2 MU-MIMO, AP or Wi-Fi as WAN

Compute

  • CPU: Intel Xeon 6520P — 24C/48T, 2.4–4.0 GHz
  • Memory: 128GB DDR5
  • Storage: 12 TB NVMe PCIe 5.0
  • GPU: NVIDIA RTX Pro 5000 Blackwell — 48GB GDDR7 ECC

Form Factor

  • Dimensions: 3U — 19 in (48.3 cm) W × 18 in (45.7 cm) D × 5.25 in (13.3 cm) H
  • Mount: Rack, wall, rail, surface, or portable
  • Power: 100–240V AC input, avg ~500 W, peak ~850 W

Unlock AI-powered output anywhere.

Talk to us