Converting AI data centre waste heat into electricity.

Data centers are the energy infrastructure challenge of our time. Currently, their excess heat is vented as waste. But we see it as a fuel source.Our technology will harvest data centre heat as an asset and convert it back into electricity to power each AI data centre more efficiently and sustainably.As we move toward our MVP and initial pilots, we are building the new standard for energy recovery. Join us.

Explore the story

MotionLab Berlin hardtech accelerator graduates. Based at Future Energy Labs (dena, Berlin).

The Problem.

Data centres are the engine rooms of our time, yet they are running into a hard wall: power. Jensen Huang identifies it as the single biggest bottleneck for scaling AI. Data centres consumed 415 TWh of electricity in 2024, projected to reach 945 TWh by 2030 as AI workloads explode. That is nearly double Germany’s total annual electricity consumption.The constraint is not just total energy: it is the grid’s ability to deliver it. New data centres face multi-year waits for grid connections. Operators are hitting peak demand ceilings that limit how many racks they can fill. Every kilowatt-hour generated on-site is a kilowatt-hour not drawn from the grid, reducing dependency, lowering peak demand charges, and unlocking rack density that grid constraints currently block.Meanwhile, 10-30% of that consumed power goes to cooling, which vents heat as waste. All the while, the EU’s Energy Efficiency Directive mandates energy performance reporting for data centres, and Germany’s Energy Efficiency Act requires new data centres to reuse waste heat from July 2026. Regulation is compounding economic pressure that already exists.The heat is there. The need for power is desperate. No viable technology bridges the two.

Data Centre Electricity Consumption, 2020–2030 (TWh), IEA — Data Centre Electricity Consumption (2020-2030) (TWh)
_{IEA (2025), Global data centre electricity consumption, by equipment, Base Case, 2020-2030, IEA, Paris. Licence: CC BY 4.0}

Physics created limitations.So why aren't people converting waste heat into electricity already? The barrier to recovering waste heat as an energy source has always been the "Thermodynamic Dead Zone." At 40–65°C rack output, data center heat is too cool for traditional turbines (ORC) while standard thermoelectric devices often operate at less than 1% efficiency. This physics constraint makes the conversion to electricity economically unviable for the legacy market.FluxTech is solving this. By utilizing supercritical CO₂ (sCO₂) as a working fluid, we gain the thermodynamic leverage required to operate with high efficiency in this temperature range. Combined with our proprietary AI control systems, we can close the loop by capturing the surplus heat and converting it into electricity.

The Solution.

FluxTech bridges the "Thermodynamic Dead Zone" by converting low-grade server heat into high-value electricity that can help power the data centre in a closed, sustainable loop. While legacy systems require temperatures above 80°C to function, our technology is intended to operate with high efficiency in the 40–65°C range common to modern data centres.The sCO₂ Advantage.We utilize supercritical CO₂ (sCO₂) as a working fluid. Near its critical point, sCO₂ gains immense thermodynamic leverage: small temperature changes trigger massive state changes. This sensitivity allows us to extract energy where traditional turbines find only "waste".Engineered for Reliability.Our thermoacoustic engine contains practically no moving parts. No turbines, no pistons, and minimal mechanical friction. By converting heat into acoustic waves to drive a linear alternator, we ensure operational longevity sustainably.AI-Driven Precision.Operating near the CO₂ critical point is inherently nonlinear. FluxTech’s core proprietary technology utilizes real-time AI control systems to stabilize these dynamics, optimizing efficiency every second based on live operational data. This creates a technical moat: the hardware cannot function without the control system, and the system is trained on data unique to our architecture.Seamless Integration.

Operational Cooling System. FluxTech is the site's cooling pathway in normal operation, sized to site load.Backup-Redundant. The existing cooling stack stays in place as backup. On fault, the engine decouples automatically and uptime is unaffected.Grid Ready. We turn a thermal liability into a measurable reduction in grid dependency.

The Business Model.

Energy as a Service.We install, own, and operate the hardware. Customers pay for every kWh generated from their waste heat. All performance risk sits with FluxTech. If the engine produces nothing, the customer pays nothing.The engine modulates between electricity recovery and heat delivery, capturing whichever revenue stream is paying more.This model aligns incentives across all three stakeholders: every efficiency improvement flows directly into revenue, with no intermediary step.

For the Customer.No capital outlay. No operational burden. Revenue from infrastructure they already own. FluxTech operates as the cooling pathway. If the engine stops, uptime is unaffected and the customer owes nothing.

For FluxTech.Hardware ownership captures the full value of every efficiency gain. Revenue scales directly with technology performance. Fleet ownership enables modular upgrades in the field and circular lifecycle management: replaced units are refurbished and redeployed.

For the Investor.Recurring revenue tied to a cost the customer already pays. Performance compounds into revenue without new customer acquisition. Fleet growth enables project finance (borrowing against contracted revenue to fund expansion), reducing equity dilution in later phases.

Pricing Structure.Two components: a fixed base rate covering cooling provision, agreed at signing, and a floating component tied to the market value of recovered electricity and heat passed through for reuse. The fixed component provides the revenue predictability that project finance requires. The floating component aligns incentives on the variable streams: transparent benchmarks against local market prices mean neither party can game the rate.

Three-year initial agreements with renegotiation at renewal. Early-phase contracts prioritise adoption speed; as the technology matures, terms shift toward longer-duration agreements suited to project finance.Zero downtime risk to the customer. If a unit is not generating electricity, the customer pays nothing. The engine decouples from the cooling loop; data centre operations are unaffected.Fleet scaling via project finance. Once units generate revenue under contract, the revenue stream becomes an asset for debt financing. Each performing cohort strengthens borrowing terms for the next. Equity funds the company through technical de-risking; growth capital increasingly comes from debt markets.

The Physics & Engineering.

FluxTech's system is a closed-loop thermoacoustic engine. It connects to the data centre's cooling fluid on one side and produces electricity on the other. The sections below trace the conversion end to end.

Thermal Capture.
Low-grade waste heat (40–65°C) is drawn from the server cooling circuit through high-efficiency printed circuit heat exchangers. This thermal energy is transferred to our working fluid: supercritical CO₂ (sCO₂).
Acoustic Amplification.
Near its critical point, sCO₂ becomes hyper-sensitive to temperature changes. As the fluid absorbs heat, it expands and contracts, creating self-sustaining acoustic pressure waves within a resonant cavity. By using sCO₂, we gain the thermodynamic leverage required to operate in the "Dead Zone" where traditional turbines fail.
Energy conversion.
These high-energy acoustic waves drive a linear alternator to generate electricity. Because the conversion is driven by sound rather than rotation, the engine contains no moving parts: no turbines, no pistons, and no bearings. Reliability is structural.
AI Optimization.
Operating near the critical point is inherently nonlinear and unstable. Our proprietary AI control system monitors operational data in real-time, adjusting valve timing and flow rates every second to keep the system in its highest efficiency band.
Seamless Integration.
FluxTech is the cooling pathway in normal operation. The existing cooling stack stays in place as backup; on fault, the engine decouples automatically and uptime is unaffected. Residual heat after electricity extraction is passed through at the appropriate temperature for secondary reuse.
Heat Modulation.
Our core differentiating mechanism. Controlled, phase-locked thermal exchange synchronises heat input with the engine’s acoustic cycle, maximising energy transfer at each oscillation. This architecture has no published precedent and is the subject of our first ZIM feasibility study.

The Market.

The market for FluxTech’s technology is not just large, it is structurally inevitable due to the convergence of two things: surging demand for AI and the avoided load advantage.The scale of the opportunity is defined by the explosive growth of data centres, which are projected to consume 945 TWh of electricity by 2030, driven largely by energy-intensive AI workloads.The avoided load advantage is the ability to generate power on-site to bypass grid capacity constraints. By reducing the total draw from the grid, operators can lower peak demand charges that currently limit rack density.We are seeing immediate market pull for three specific reasons.

Economic Compulsion.
High energy prices, particularly in Europe, make efficiency economically valuable independent of regulation. In our initial addressable segment alone, every percentage point of engine efficiency adds approximately €4M in annual revenue at scale. In grid-constrained markets like Frankfurt and London, every recovered MW unlocks compute capacity worth far more than retail electricity.

Commercial Validation.
We have already secured three signed LOIs with deployment partners Deep Green, Leafcloud, and Heata, whose deployment pipelines represent up to 2.6 TWh of recoverable low-temperature waste heat, anchored by Deep Green's publicly stated 300 MW buildout across the UK and North America.

Binding Sustainability Mandates.
The EU’s Energy Efficiency Directive mandates energy performance reporting for data centres, with minimum standards planned for 2026. Germany’s Energy Efficiency Act requires new data centres to reuse waste heat, creating direct regulatory demand for energy recovery infrastructure.

The Team.

John Therrien | Co-Founder & Engine Architect.Where thermodynamics meets control theory.John holds a double major in Physics and Mechanical Engineering from UC Santa Barbara, a five-year program spanning two colleges. That dual foundation converges directly in the nonlinear physics at the core of FluxTech’s engine architecture.Before FluxTech, John spent five years as a professional musician, touring and working in the studio. The experience sharpened project management, rapid iteration, and comfort with ambiguity. FluxTech’s thermoacoustic engine architecture, working fluid choice, and AI control approach emerged from independent inquiry into low-temperature heat conversion; self-starting in a technical arena the industry had written off.“This whole field of problems keeps me awake at night in the best way possible; it’s truly fascinating. The ecological impact feels tangible, the economic side gains strength every time I check the news, and I love where it sits with compelling interdisciplinary depth. It’s a perfect engine to drive the kind of obsession necessary for a successful startup.”

Florian Schmack | Co-Founder & Materials Lead.Materials define limits. Engineering overcomes them.Flo holds a master’s degree in materials science from TU Berlin, specialising in light metals and high-purity synthesis. At PSC Technologies, he pioneered a method for identifying silicon impurities below 0.1%, a breakthrough that eliminated false positives in battery testing and set a new standard for material analysis.In 2024, Florian integrated his materials expertise with advanced data science at the WBS Coding School, specialising in Machine Learning and Cloud Computing. This intersection of physical metallurgy and predictive analytics is the engine behind FluxTech’s proprietary control systems. Today, Flo leads the technical development of our supercritical CO₂ systems and the construction of the TRL 4 research platform.“We are not just managing heat, we are harvesting a previously invisible resource. Let’s turn the waste of today into the power of tomorrow.”

The Ask.

Strategic Investment Round.FluxTech is raising up to €500k to join committed capital.This is a leverage round. Every euro of equity activates a cascade of non-dilutive grants, scaling total deployment from ~€460k to ~€1.2M depending on round size.

The Funding Cascade.This capital unlocks two parallel federal workstreams that transition FluxTech from feasibility to industrial readiness:

Grant Leverage Calculator

Investment size €250k

Total Dilutive

—

Total Non-Dilutive

—

Total Deployment

—

Grant Leverage

—

Private Capital GründungsBONUS ZIM Feasibility ZIM R&D

Upcoming Milestones.The funded program runs two parallel workstreams over a 12-month cycle to retire core research risks:

Feasibility Study 1 | Heat Modulation: Validates our controlled, phase-locked heat exchange on the existing research platform.Feasibility Study 2 | Near-Critical sCO₂: Integrates supercritical sCO₂ to activate the engine’s thermodynamic leverage at target temperatures.

The Technical Gate.Successful completion of these studies positions FluxTech for the next funding layer. This includes SPRIND (up to €1M) and EIC Pathfinder (up to €3M), all without requiring additional equity.

We are looking for a partner who understands the value of entering at maximum leverage.Join Us.Email john@flux-tech.de to set up a meeting.