TonoForge™ – Decentralized AI Factory | Tonomia Product Catalogue 2025–2026
TonoForge AI Factory
Product Catalogue 2025–2026
Green.
Sovereign.
Resilient AI.

The world’s first containerized, off-grid, distributed AI factory — deployable anywhere in 12 weeks. From a single TonoForge™ unit to a 180 MW TonoFactory™ campus.

🌿 80% Heat Recovery ⚡ LFP / Na⁺ Battery Backup 🔒 Sovereign AI Infrastructure 🌍 Europe · USA · MEA · Asia 🚀 12-Week Deployment 🤝 NVIDIA Inception Member
0.3MW
per TonoForge™ 20
180MW
TonoFactory™ campus
1.1EFLOPS
peak AI compute / rack
15yr
system lifetime
About Tonomia

The AI Infrastructure Layer
for a Sovereign Digital Future

Tonomia SRL is a Belgian sovereign AI infrastructure company headquartered in Battice, Liège. We build containerized, renewable-energy-coupled, distributed AI factories that bring exascale compute directly to where it’s needed — with zero dependency on centralized hyperscale clouds.

Every TonoForge™ unit is a fully self-contained AI data center — shipped as an ISO container, connected to renewable energy, and orchestrated by TonoFabric™, our proprietary GPU orchestration software. Deployable in 12 weeks. Operable for 15 years.

🏭

Containerized AI Factory

ISO 20ft HC container with integrated cooling, power, networking and AI compute. Ships anywhere in the world via standard freight logistics.

🌿

80% Heat Recovery

Industry-leading 3-way valve heat recovery system. Winter heating. Summer absorption cooling. Net-zero ambition by design, not by offset.

Permanent Battery Power

LFP or Na⁺ battery arrays with up to 4h backup — no traditional UPS required. N+N power module topology. 10-year battery warranty.

🔒

Sovereign by Design

Your data never leaves your jurisdiction. TonoFabric™ provides GPU orchestration and monitoring with full sovereignty over compute and data.

📡

Scalable to Gigawatt

From a single 0.3 MW node to a 180 MW TonoFactory™ campus. Modular architecture scales linearly — add units as demand grows.

🤝

Best-in-Class Compute

NVIDIA Inception member. AMD partner. MiTAC partner. Certified to host GB300 NVL72 and MI355X — the world’s most powerful AI accelerators.

Product

TonoForge™ 20
0.3 MW · ISO 20 ft HC

A fully autonomous AI data center in a 20-foot container. Two OCP 48U racks at 150 kW each, liquid-cooled, battery-backed, heat-recovering — ready to connect to solar, wind, or grid power anywhere in the world. Deployable in 12 weeks, operable for 15 years.

TonoForge 20 3D Render
TonoForge™ 20 — Dual AC Chiller Configuration
TonoForge Top View
Top view — dual cooling units with heat exchanger pipework

Technical Specifications

0.3 MW
General
Container TypeISO 20 ft HC
RegionEurope, USA, MEA & Asia
Voltage / Frequency400V/50Hz – 480V/60Hz
Ambient Range-40°C – +60°C
Physical
Height6.06 m
Length2.44 m
Width2.90 m
IT Racks
No. of Racksmax. 2 (OCP, 48U)
Rack Density150 kW per rack
PDU4× per rack, Switched
Cooling & Power
Max IT Load0.3 MW
CoolingLiquid 70–80% + Air 20–30%
Cooling TypeAC Chiller / Absorption
A/C RedundancyN + N
Heat Recovery80%
Battery BackupUp to 4h (default 2h)
Battery TypeLFP or Na⁺ Ion
Power TopologyN + N
0.3MW
Max IT Load
80%
Heat Recovery
4h
Battery Backup
12wk
Deployment Time
15yr
System Lifetime
N+N
Power & Cooling Redundancy
GPU Configurations

Choose Your
AI Powerhouse

TonoForge™ is hardware-agnostic. We qualify and integrate the world’s leading AI accelerators — ensuring you get maximum performance within the container’s power and cooling envelope.

Infrastructure

TonoForge™ 20
0.3 MW · ISO 20 ft HC · 2 × OCP 48U racks

AI Compute

MiTAC AMD · 64 × MI355X
G4826Z5 · 8 nodes · AMD CDNA™ 4 · ~1.3 EFLOPS FP4
GPU System Specs
Total GPUs64 × AMD Instinct™ MI355X
GPU ArchitectureAMD CDNA™ 4
Total GPU Memory18.4 TB HBM3e
Peak AI (FP8)~644 PFLOPS
Peak AI (FP4)~1.3 EFLOPS
InterconnectInfinity Fabric 1,075 GB/s
Total CPUs16 × AMD EPYC™ 9005
Total System RAM~48 TB DDR5-6400
Network400 / 800 Gb/s
SoftwareAMD ROCm™ 7+
Cooling (rack)Full cold-plate DLC
Rack Power~120–150 kW
MiTAC AMD MI355X Rack

Infrastructure

TonoForge™ 20
0.3 MW · ISO 20 ft HC · 2 × OCP 48U racks

AI Compute

SMC NVIDIA · GB300 NVL72
SRS-GB300-NVL72 · 72 B300 GPUs · Blackwell Ultra · 1.1 ExaFLOPS
GPU System Specs
Total GPUs72 × NVIDIA B300 Blackwell Ultra
Total CPUs36 × NVIDIA Grace™ (Arm)
GPU Memory / GPU288 GB HBM3e
Total GPU Memory~20.7 TB HBM3e
Combined Fast MemoryUp to 40 TB
NVLink / GPU1.8 TB/s (5th Gen)
NVLink Aggregate~130 TB/s
Peak AI (FP4)1.1 ExaFLOPS
Network / GPU800 Gb/s ConnectX-8
vs. H100 output50× AI factory output
Cooling (rack)DLC 250 kW CDU
Rack Power132–140 kW
Supermicro NVIDIA GB300 NVL72 Rack
TonoFactory 180MW campus
Scale

From One Unit
to a 180 MW
AI Campus

TonoForge™ units cluster into TonoFactory™ — a full-scale AI park with integrated solar, BESS, substation and operations centre. Modular expansion from day one, with no stranded capacity.

180
MW campus capacity
600+
TonoForge units
3
continents covered
TonoFactory aerial view
TonoFactory™ aerial view — solar + AI containers
TonoForge golden hour
TonoForge™ 20 — deployable in any environment
Service & Warranty

Built to Last 15 Years

Market-standard warranties aligned with IEC 62619, IEC 62040, EN 50600, ASHRAE TC 9.9, and Uptime Institute Tier III/IV. 24/7 Tonomia NOC included.

10yr
Battery System
LFP / Na⁺ · 80% retained capacity · ≥4,000 cycles · IEC 62619 / UL 1973
3yr
Power Conversion
Rectifiers, inverters, control boards · IEC 62040-1 / EN 50160
5yr
Structural Container
ISO 1496-1 · CSC recertification every 5 years · 25+ yr structural life
3yr
Cooling Systems
AC Chiller + CDU + Absorption · EN 378 / ASHRAE 15 · quarterly service
3yr
OEM GPU Servers
MiTAC / SMC pass-through · 4h NBD on-site · extendable via TSC
24/7
Tonomia NOC
P1: 15min remote · 4h on-site · Monthly SLA report · ISO 20000-1