HPC Data Centers
Run Cooler,
Cost Less.

A decision support tool that helps HPC operators choose and justify risky operational changes by making risk and trade-offs visible before acting — using the data they already have.

Case Study — Northern Europe HPC Centre
Air Cooling → Direct Liquid Cooling Upgrade
1,200 racks · 4.8 MW IT capacity · ASHRAE TC9.9
1.87→1.56PUE Reduction (−16.6%)
$1.25MAnnual Cost Saving
−2,425tCO₂ / Year
2.2 yrsPayback Period
Tool outcome: all 6 risk domains (thermal, energy, financial, water, stability, compliance) moved HIGH / MED → LOW. Full audit trail generated.
The Problem

AI Workloads Are Overheating Data Centers

GPU clusters run hot, cooling overhead explodes, and operators fly blind — managing multi-million dollar infrastructure with spreadsheets and intuition alone.

🌡️

Thermal Blind Spots

No unified view of rack-level thermal headroom. Risk is discovered only after something fails, throttles, or breaches ASHRAE A2 compliance limits at peak load.

Cooling Costs Spiralling

HVAC overhead averages 45–50% of total facility energy. Overprovisioning to stay safe burns budget that could fund infrastructure upgrades or hardware modernisation.

🔄

No Change Simulation

Engineers can't model the impact of a setpoint or cooling technology change before applying it. Every operational decision is a live experiment on production hardware.

🗂️

Compliance Without Evidence

Audit trails, operator sign-offs, and decision rationale exist nowhere. EU EED and ASHRAE regulatory exposure grows with every undocumented infrastructure change.

The Tool

Thermal Intelligence,
Operator-Grade

The HPC Decision Support Tool turns raw telemetry into actionable decisions. Operators analyse baseline performance, simulate changes, assess full system impact across six domains, and generate a compliant PDF decision brief — all in one structured workflow.

Tool Capabilities

Everything an Operator Needs to Act With Confidence

01

Baseline Performance Analysis

12-month KPI trends — PUE, COP, inlet temperatures, WUE, rack utilisation — benchmarked against ASHRAE, Green Grid, and LBNL targets. Know exactly where you stand before any change.

02

What-If Scenario Matrix

Model any cooling technology — air, DLC, hybrid, immersion — across setpoint and utilisation parameters. Side-by-side scenario comparison before a single rack is touched.

03

Six-Domain Impact Assessment

Simultaneous evaluation across thermal management, energy efficiency, financial impact, water sustainability, operational stability, and regulatory compliance — all scored before committing.

04

Financial Business Case

CapEx, OpEx, annual savings, payback period, and 10-year net saving tied directly to the simulated change. Present a CFO-ready business case in minutes, not weeks.

05

Operator Decision Brief

Auto-generated PDF with executive summary, change rationale, KPI before/after, compliance status, and operator sign-off — an auditable decision trail from day one.

Strategy

A Two-Phase Vision

A disciplined two-phase strategy that builds customer relationships now and converts them into hardware deployments later — de-risking adoption at every stage.

◉ Phase 1 — In Progress

HPC Decision Support Tool

MVP in active testing with HPC operators. Building relationships, validating workflow adoption, and collecting the telemetry baseline required for hardware deployment.

◯ Phase 2 — 2027+

Hardware Pilot Programme

Proprietary waste-heat harvesting modules deployed at validated customer sites. Tool relationships become the distribution channel — no cold sales required.

◯ Phase 3 — 2027+

IP Portfolio & Licensing

Patent portfolio around novel thermoelectric energy recovery opens licensing revenue alongside direct hardware sales at scale.

Market Validation Checks

Problem DiscoveryHPC operator interviews confirmed thermal management as urgent, budgeted pain with clear willingness-to-pay signals.
Tool Wedge ActiveMVP workflow testing underway. Case study generated from benchmark dataset; operator feedback loop live.
First Paid ContractTarget: signed tool subscription with HPC facility before end of 2025.
Hardware PilotDeploy first waste-heat harvesting module at validated customer site in 2027+.
Deep Tech

Proprietary Energy Recovery

Data centers dissipate an estimated 40–50% of consumed electricity as waste heat. OrionLinks is developing a novel approach to capture and convert microchip-level waste heat into recoverable electrical energy — a problem that has resisted conventional thermoelectric solutions at this scale.

Our research focuses on advanced materials science and precision thermal management at the chip interface, building toward a patentable hardware solution that can be retrofitted into existing HPC infrastructure without redesigning the facility.

  • Chip-level thermal interface energy capture
  • Advanced thermoelectric conversion materials
  • Retrofit-compatible module design for existing racks
Energy Flow Concept
💻Chip~85°C TDP
🔥Waste HeatCaptured
ElectricityRecovered
Proprietary conversion method — patent strategy in development. Technical details not yet disclosed.
Data Center Waste Heat40–50% of consumed energy
Global DC Energy Use>400 TWh / year
Addressable Recovery MarketMulti-billion €
Founders

Technical Leadership

Sebastian Irinciuc, CEO

Sebastian Irinciuc

CEO & Co-founder

Enjoys dissecting complex technical challenges and transforming ambitious ideas into working systems. Leads product strategy, customer discovery, and investor relations.

Jorge Iglesias, CTO

Jorge Iglesias

CTO & Co-founder

Has the technical depth to build from nothing and the engineering judgement to know what matters. Leads platform architecture, thermal modelling, and hardware R&D.

Institutional Support

Backed by Europe's Best

OrionLinks is built within the ecosystem of Europe's leading innovation and academic institutions — accelerating our path from prototype to market.

Get In Touch

Let's Build the Future
of Efficient AI Together.

We're actively speaking with HPC operators, data center managers, and early-stage investors who want to be part of solving AI's energy problem.