Distributed AI Inference · Grid Edge · Calgary + Tulsa

Let's build the
inference edge.

RyuGen turns commercial real estate into high-yield digital infrastructure — deploying modular, high-density AI compute where demand already is, and bypassing the grid's multi-year queues with proprietary thermal and power orchestration.

Partner with us → Explore the technology

Built on NVIDIA-Certified systems · RTX PRO 6000 Blackwell

49 GW

Projected U.S. capacity shortfall

7+ yr

Hyperscale interconnect queue

2–4 mo

RyuGen time-to-revenue

90%+

GPU waste heat recovered

01 — The Challenge

The bottleneck isn't GPUs. It's power.

AI inference demand is outrunning the grid. New capacity stalls behind multi-year interconnection queues and heavy capital — while commercial real estate sits on rising vacancy and underused power. RyuGen connects the two, putting idle power to work for the workload that needs it most.

74GW

U.S. data-center demand projected by 2028 — ~49 GW of it unmet

20k MW

Data-center load requests in Alberta against ~12k MW peak capacity (AESO)

$2.25/hr

GPU inference floor, +39% YoY — demand outrunning supply

02 — The Solution

Two pillars, one foundation — and no shortcut around it.

The Helix Node makes compute distributable. HeatSync makes it deployable inside real estate it would otherwise cook. Both stand on RyuGen Command — the control plane that turns scattered nodes into one network we can dispatch and monetize. Each is necessary; none is sufficient alone.

The Network Pillar

Helix Node

Modular sub-MW compute that drops into a building as-is — 16× NVIDIA RTX PRO 6000 Blackwell per node, NVIDIA-Certified, with direct liquid-to-chip cooling and cold-climate operation to −30 °C. It deploys at the metro edge and skips the interconnection queue entirely.

The Deployment Pillar

HeatSync™

Proprietary thermal energy storage that lets dense compute live inside real estate it would otherwise overwhelm. It captures 90%+ of GPU waste heat at PUE 1.10–1.15, banks it in a proprietary storage medium, and dispatches a stable 60–70 °C loop into building systems — turning a cooling cost into landlord value. Provisional patent; 5–7 year moat.

The Foundation

RyuGen Command

The control plane the pillars stand on: multi-objective workload scheduling, predictive thermal control, and behind-the-meter power balancing across every node we operate. It turns distributed hardware into one dispatchable network we can monetize across channels. Without it, the pillars are just hardware.

Pull any one and the structure falls. The hardware, the thermal IP, and the control plane only create an advantage together — and no competitor stands on all three.

03 — The HardwareHX-ES-001

The Helix Node.

A distributed AI-inference appliance engineered for warm-water cooling and grid-edge deployment — the constant unit that scales from a 2-node pilot to 40-node commercial sites without redesign.

GPUs / node16 × RTX PRO 6000 Blackwell

Form factor7U · 19" commercial-aisle

CoolingDirect liquid-to-chip · ASHRAE W5+

Coolant loop60 °C supply / 70 °C return

Heat captured95%+ to liquid loop

Ambient envelope−30 °C to +35 °C

Acoustic< 65 dBA @ 1 m

Power input480 V 3-phase · behind-the-meter

CertificationNVIDIA-Certified server platforms

04 — Thermal AdvantageHeatSync™

Every kilowatt of compute heat does two jobs.

Conventional liquid cooling rejects GPU heat to the environment. Helix runs hot on purpose — a 70 °C return loop that's warm enough to drive real building thermal systems. HeatSync's proprietary thermal energy storage banks that heat and dispatches it on demand — the core moat that turns intermittent waste heat into a stable, year-round source of value.

40 °CCOOL — REJECTED HEAT→USEFUL THERMAL80 °C

~55 °C

Conventional reject

58 °C

Adsorption min

70 °C

Helix return

The 70 °C return clears the adsorption-cooling threshold by 12 °C — opening direct hydronic heating, domestic hot-water pre-heat, and refrigerant-free cooling, all from the same loop. That's a thermal revenue stack and carbon-credit attribution no heat-rejecting competitor can claim.

05 — The MoatDefensibility

Why hasn't anyone else done this?

Vacant, grid-connected buildings have existed for years. Heat is the reason they're still empty — and the reason we can fill them.

The hidden wall

Buildings can't shed compute heat

A single Helix Node pushes ~10 kW+ into a space designed for people and plug loads. Commercial HVAC was never built for continuous rack-density heat. Drop GPUs in as-is and you cook the room.

Why the obvious fix fails

Bolt-on cooling breaks the math

Add chillers and you add 20–40% parasitic load — PUE collapses. You pay to make the heat, then pay again to dump it. The landlord inherits the noise, permitting, and roof work — and says no.

The RyuGen unlock

We turn the wall into the wedge

Direct liquid-to-chip handles the density that kills air-cooled retrofits. HeatSync captures 90%+ and reuses it — winter heat, summer cooling, hot water. Heat becomes landlord value, not an objection. The lease gets signed.

Heat is the gate. We hold the key. Replicating the integrated stack — the thermal-storage IP, the retrofit playbook, and a landlord with no reason to say no — costs a fast follower 5–7 years and real capital.

06 — Why Now

Three inflection points are converging.

/ 01

The Inference Shift

+39% YoY

Inference now exceeds training. Distributed metro-edge compute meets latency requirements that centralized facilities structurally can't.

/ 02

The Power Bottleneck

2–4 months

Sub-MW behind-the-meter deployments bypass 5–10 year interconnection queues. RyuGen reaches revenue in months, not years.

/ 03

Sovereign Compute

$2B CAD

Canada's AI Compute Access Fund is actively distributing grants. Calgary positions RyuGen for sovereign infrastructure demand on home ground.

07 — Execution

From a 25 kW proof-of-concept to a continental fleet.

The pilot puts 2 Helix Nodes — 32 GPUs at 25 kW — into a single commercial building in Tulsa or Calgary, behind the meter. It validates the unit economics, the thermal recovery, and the deployment playbook before commercial scale.

GPUs across 2 Helix Nodes · 25 kW IT load

Target Helix sites across North America by 2032

Q3 2026

Pre-seed close + site selection

Lock the pilot site and regulatory posture across the Calgary and Tulsa markets.

Q1 2027

Hardware procurement & build-out

OEM co-development of the 7U Helix Node enclosure; site retrofit and liquid-cooling loop.

Q3 2027

25 kW pilot operational

First reference deployment online. Third-party PUE and heat-recovery metering.

Q3 2028

First commercial site

Seed round funds the first sub-MW commercial Helix deployment.

2032

Ten+ Helix sites

A distributed grid-edge inference fleet across North America.

08 — Ecosystem

Backed by the right ecosystem.

RyuGen is built on a validated hardware path and a network of energy, compute, and thermal partners.

Hardware Platform

NVIDIA RTX PRO 6000 Blackwell

The Helix Node is built on NVIDIA-Certified server platforms — the same Blackwell inference GPU now being deployed across major global inference grids.

Validated hardware path

Venture Studio

Avatar Innovations

RyuGen is a resident company at Avatar Innovations — Canada's energy-transition venture studio, in Calgary's Energy Transition Centre.

Lead investor target

Thermal Integration

To Be Disclosed

Discussions underway on adsorption-cooling integration — turning the 70 °C HeatSync loop into refrigerant-free chilled water for a closed-loop, ultra-low-PUE site.

Partnership discussion

EntityRyuGen Energy Solutions Canada Inc.

HeadquartersCalgary, Alberta · Canada

StagePre-seed · raising

StudioAvatar Innovations portfolio

Core IPHeatSync™ · provisional patent

09 — Company

An energy company
for the compute era.

RyuGen Energy Solutions is a Calgary-based deep-tech company commercializing the convergence of distributed AI compute and waste-heat recovery. We don't compete for grid capacity — we build the infrastructure that operates around its limits, and we turn the heat that everyone else throws away into a second revenue stream.

10 — Team

Why this team, for this problem.

RyuGen sits exactly where energy infrastructure meets AI compute — and so does the team. Between us we've run midstream gas operations, built energy-trading and risk systems, engineered pipelines and thermal processes, and shipped data-science work. The rare combination this thesis demands is already in the room.

CEO & Co-Founder · Inventor, HeatSync™

Sean Brown, MBA

The throughline of the company. Eighteen-plus years across engineering and energy — seven of them in midstream natural gas, alongside a Senior Data Scientist role at Williams Companies. That pairing of energy-infrastructure fluency and applied data science is exactly what RyuGen runs on. Dual B.S. in Aerospace & Mechanical Engineering and an MBA; sole inventor on the HeatSync™ provisional patent and resident at Avatar Innovations, Calgary.

Midstream gas · WilliamsApplied data scienceMechanical & aerospace eng.HeatSync™ inventor

Tristan Walker

Operations

M.Sc. Sustainable Energy. Eight years in renewable energy leading large-scale energy development. Two-time startup founder.

Samuel Duperré

Compute

Energy-trading analytics and risk management experience across natural gas, power, crude, and FX, bringing a commodity-markets discipline to inference capacity. Startup founder.

Hadi Bagherzadeh

Technology

PhD researcher at the University of Calgary and reservoir engineer, with hydrocarbon-upgrading and thermal-process expertise.

Josh Harris

Engineering

Mechanical engineer and pipeline project engineer at Venture Global LNG — field ops, hydraulics, and project development.

Backed & advised by

Avatar InnovationsExergy Solutions

Let's build theinference edge.