Project Title: Integration of Waste-Heat-Driven Atmospheric Water Harvesting (AWH) into Greenhouse-Coupled Datacenter Infrastructure
Implementation Partner: K51 AG, Überlandstrasse 129, 8600 Dübendorf Research Partner: Empa — Swiss Federal Laboratories for Materials Science and Technology, Urban Energy Systems Lab (Dr. Binod Koirala) Technology Partner: Solabs Nanotechnology (Fabrice Bagnoud, ETH Zurich)
K51 AG operates datacenter containers as fossil-free heating systems for commercial greenhouses. The company's infrastructure is deployed directly at agricultural sites, where computing hardware generates waste heat that replaces fossil fuels for greenhouse climate control.
K51's current value chain position is that of an integrated infrastructure operator, combining:
The proposed Innosuisse project adds a third value layer: waste-heat-driven water production and dehumidification through Atmospheric Water Harvesting (AWH) based on sorbent technology developed at ETH Zurich.
This transforms K51's business model from a dual-revenue system (compute + heat) to a triple-revenue system (compute + heat + water/dehumidification), fundamentally improving the economic viability and year-round operability of the infrastructure.
The research foundation is provided by Empa (Urban Energy Systems Lab), which will validate system performance, energy balances, and integration parameters under controlled and real-world conditions. The sorbent technology is developed by the Solabs team (ETH Zurich), who bring deep expertise in thermochemical absorption systems. K51 acts as the implementation partner, providing real datacenter-greenhouse infrastructure, operational know-how, and the commercial pathway to market.
A potential spin-off company ("NewCo") may emerge from the project to commercialize AWH technology independently in broader markets (e.g., standalone datacenter cooling). K51's interest, however, is focused on securing exclusive usage rights for the greenhouse-integrated application in Switzerland and selected European markets.
Swiss greenhouse operators face a convergence of economic and regulatory pressures that fundamentally threaten their business model:
Mandatory decarbonization. Major Swiss retailers (Migros, Coop) increasingly require their agricultural suppliers to eliminate fossil fuel use. Erdgas (natural gas), historically the dominant heating source for greenhouses, is becoming a non-option. Operators must find fossil-free heating alternatives or risk losing their primary sales channels.
High energy costs for dehumidification. Greenhouse crops continuously release moisture. The standard practice for humidity control is "dry heating" combined with roof venting — heating the air to absorb moisture, then venting the warm, moist air outside. This process wastes up to 30–40% of total heating energy and causes significant CO₂ losses, which must be compensated by purchasing additional CO₂. In Switzerland, CO₂ emissions from greenhouses are increasingly regulated, making CO₂-neutral climate control economically critical.
Increasing water scarcity. While Switzerland is not traditionally considered water-stressed, seasonal droughts are increasing, particularly during summer months. Greenhouse irrigation demands are substantial, and on-site water production represents both an economic advantage and a resilience factor.
No integrated solution available. Greenhouse operators today must procure heating, dehumidification, and water separately from different providers or systems, each with its own cost structure. No commercially available system delivers all three services from a single waste-heat source.
K51's current business model faces a structural constraint that limits profitability and hardware choices:
Seasonal operation only. Under Swiss zoning law (RPG), datacenter infrastructure in agricultural zones must provide a direct agricultural benefit. Since K51's primary service is heat delivery, computing operations are strictly heat-driven — the datacenter runs only when the greenhouse requires heat. This limits annual operating hours to approximately 4'500–6'000 hours (heating season only), with no operations during summer.
Low-value computing hardware. Because uptime cannot be guaranteed year-round, K51 cannot offer Service Level Agreements (SLAs) required by high-value computing customers (AI inference, rendering, scientific computing). The company is limited to low-end, interruptible workloads (e.g., Bitcoin mining, batch processing) that generate only approximately CHF 0.10 per kWh of electrical input — barely covering operating costs.
Marginal profitability. With current unit economics (see financial plan below), a 1 MW installation generates a Free Cashflow of only approximately CHF 30'000 per year — insufficient for meaningful reinvestment or risk buffering.
The core problem is not heat — it is the inability to operate year-round. K51 needs a legitimate, regulation-compliant agricultural service that requires waste heat during summer months. AWH-based water production and dehumidification is precisely this service.
K51's USP is not the AWH technology itself, but the system-level integration of computing, heating, dehumidification, and water production into a single infrastructure that serves two customers simultaneously:
1. "Three-in-one" infrastructure for the farmer. A single K51 installation replaces the greenhouse operator's heating system, dehumidification system, and part of the water supply. No other provider offers all three services from waste heat.
2. Year-round datacenter operation in agricultural zones. By adding AWH, K51 transforms seasonal heating infrastructure into permanent agricultural infrastructure (heating in winter, water production and dehumidification in summer), enabling continuous operation within the existing regulatory framework.
3. High-value computing through guaranteed uptime. Year-round operation enables K51 to deploy high-performance GPU servers with SLA-grade availability, unlocking compute revenues of CHF 0.80/kWh instead of CHF 0.10/kWh — an 8× improvement per kilowatt-hour.
4. Shared-cost economics. The AWH unit serves two beneficiaries (datacenter cooling + agricultural water/dehumidification), allowing costs to be shared. This creates economics that neither standalone datacenter cooling nor standalone agricultural dehumidification can achieve independently.
5. Proprietary integration know-how. K51 has operational experience integrating liquid-cooled computing hardware into agricultural heating systems. The AWH integration adds a new layer of system complexity that represents a significant barrier to entry for competitors.
K51's primary competitors for greenhouse heating are:
K51 is unique in providing fossil-free heat at a fixed price below natural gas (CHF 0.08/kWh including hardware allocation vs. CHF 0.08–0.10/kWh for natural gas), while simultaneously generating computing revenue that subsidizes the heat price.
Current dehumidification approaches:
The proposed AWH system is the only solution that provides heat-driven dehumidification without additional electrical input, using datacenter waste heat as the sole energy source.
Existing AWH companies (Genesis Systems, Uravu Labs) operate standalone, electrically powered units for drinking water production. None are designed for integration with datacenter thermal systems, and none offer combined dehumidification + water production from waste heat.
K51 generates revenue from three independent streams per installation:
Heat: Swiss greenhouse operators currently pay CHF 0.08–0.10/kWh for natural gas. K51's price of CHF 0.08/kWh for fossil-free heat is competitive and eliminates decarbonization risk. Willingness to pay is validated through existing contracts.
Water: Municipal water in Switzerland costs CHF 1.50–3.00/m³. K51's price of CHF 5.00/m³ is above municipal water rates but includes:
When accounting for the combined value of water + dehumidification + reduced CO₂ loss + reduced heating energy, the effective cost to the farmer is significantly below the standalone cost of each service procured separately.
Profit-sharing: In addition to the fixed-price heat and water contracts, K51 offers greenhouse operators a 30% share of the Free Cashflow generated on their site. This profit-sharing model aligns incentives and compensates the farmer for hosting the infrastructure. It is modeled on K51's proven approach in France, where a 50% FCF split is offered to early adopters.
Capital expenditure (one-time):
| Item | CHF |
|---|---|
| AWH sorption unit (200 kW, containerized) | 75'000 |
| GPU hardware upgrade (Low-End → High-End, 100 kW) | 150'000 |
| Total CAPEX per site | 225'000 |
Annual operating costs (steady state, Year 3+):
| Item | Calculation | CHF/year |
|---|---|---|
| Electricity ASIC | 4'500'000 kWh × CHF 0.13 | 585'000 |
| Electricity GPU (heating season) | 500'000 kWh × CHF 0.13 | 65'000 |
| Electricity GPU (summer, AWH-enabled) | 376'000 kWh × CHF 0.13 | 48'880 |
| Total electricity | 698'880 | |
| Infrastructure maintenance | Flat rate | 40'000 |
| IT operations (K51 remote management) | Flat rate | 80'000 |
| AWH sorption maintenance | Flat rate | 10'000 |
| AWH depreciation (CHF 75k / 10 years) | Linear | 7'500 |
| Total operating costs (excl. electricity) | 137'500 | |
| ASIC hardware reserve | 4'500'000 kWh × CHF 0.02 | 90'000 |
| GPU hardware reserve (High-End) | 876'000 kWh × CHF 0.15 | 131'400 |
| Total hardware reserves | 221'400 | |
| TOTAL ANNUAL COSTS | 1'057'780 |
The total project budget for the Innosuisse innovation project is estimated at approximately CHF 820'000 over 24 months, covering:
Budget allocation between partners is subject to finalization (see Section: Open Items).
Per 1 MW installation, all values in CHF
| Year 1 (2026) | Year 2 (2027) | Year 3 (2028) | Year 4 (2029) | Year 5 (2030) | Cumulative | |
|---|---|---|---|---|---|---|
| REVENUE | ||||||
| Compute ASIC (900kW × 5'000h × 0.10) | 450'000 | 450'000 | 450'000 | 450'000 | 450'000 | 2'250'000 |
| Compute GPU Low-End (100kW × 5'000h × 0.10) | 50'000 | 50'000 | 50'000 | 50'000 | 50'000 | 250'000 |
| Heat sales (5'000'000 kWh × 0.08) | 400'000 | 400'000 | 400'000 | 400'000 | 400'000 | 2'000'000 |
| Water sales | 0 | 0 | 0 | 0 | 0 | 0 |
| Total Revenue | 900'000 | 900'000 | 900'000 | 900'000 | 900'000 | 4'500'000 |
| COSTS | ||||||
| Electricity ASIC (4'500'000 kWh × 0.13) | -585'000 | -585'000 | -585'000 | -585'000 | -585'000 | -2'925'000 |
| Electricity GPU (500'000 kWh × 0.13) | -65'000 | -65'000 | -65'000 | -65'000 | -65'000 | -325'000 |
| Infrastructure maintenance | -40'000 | -40'000 | -40'000 | -40'000 | -40'000 | -200'000 |
| IT operations | -80'000 | -80'000 | -80'000 | -80'000 | -80'000 | -400'000 |
| ASIC reserve (4'500'000 × 0.02) | -90'000 | -90'000 | -90'000 | -90'000 | -90'000 | -450'000 |
| GPU reserve (500'000 × 0.02) | -10'000 | -10'000 | -10'000 | -10'000 | -10'000 | -50'000 |
| Total Costs | -870'000 | -870'000 | -870'000 | -870'000 | -870'000 | -4'350'000 |
| Free Cashflow | 30'000 | 30'000 | 30'000 | 30'000 | 30'000 | 150'000 |
Conclusion: Without AWH, K51's greenhouse-coupled datacenter model operates at near break-even. The business is sustainable but generates insufficient cashflow for meaningful growth, reinvestment, or risk buffering.
Per 1 MW installation, all values in CHF
Year 1: Innosuisse project start, Empa prototype — no commercial AWH operation. Year 2: Field deployment at Imhof Bio from Q2 — 75% operational year for AWH. Years 3–5: Full commercial AWH operation.
| Year 1 (2026) | Year 2 (2027) | Year 3 (2028) | Year 4 (2029) | Year 5 (2030) | Cumulative | |
|---|---|---|---|---|---|---|
| REVENUE | ||||||
| Compute Revenue | ||||||
| Compute ASIC (900kW × 5'000h × 0.10) | 450'000 | 450'000 | 450'000 | 450'000 | 450'000 | 2'250'000 |
| Compute GPU Low-End (100kW × h × 0.10) | 50'000 | 12'500 | 0 | 0 | 0 | 62'500 |
| Compute GPU High-End (100kW × h × 0.80) | 0 | 525'600 | 700'800 | 700'800 | 700'800 | 2'628'000 |
| Total Compute Revenue | 500'000 | 988'100 | 1'150'800 | 1'150'800 | 1'150'800 | 4'940'500 |
| Heat & Water Revenue | ||||||
| Heat sales (5'000'000 kWh × 0.08) | 400'000 | 400'000 | 400'000 | 400'000 | 400'000 | 2'000'000 |
| Water sales (m³ × 5.00) | 0 | 5'913 | 7'884 | 7'884 | 7'884 | 29'565 |
| Total Heat & Water Revenue | 400'000 | 405'913 | 407'884 | 407'884 | 407'884 | 2'029'565 |
| Total Revenue | 900'000 | 1'394'013 | 1'558'684 | 1'558'684 | 1'558'684 | 6'970'065 |
| COSTS | ||||||
| Electricity | ||||||
| Electricity ASIC (4'500'000 kWh × 0.13) | -585'000 | -585'000 | -585'000 | -585'000 | -585'000 | -2'925'000 |
| Electricity GPU heating season (500'000 kWh × 0.13) | -65'000 | -65'000 | -65'000 | -65'000 | -65'000 | -325'000 |
| Electricity GPU summer / AWH (kWh × 0.13) | 0 | -36'660 | -48'880 | -48'880 | -48'880 | -183'300 |
| Total Electricity | -650'000 | -686'660 | -698'880 | -698'880 | -698'880 | -3'433'300 |
| Operating Costs | ||||||
| Infrastructure maintenance | -40'000 | -40'000 | -40'000 | -40'000 | -40'000 | -200'000 |
| IT operations (K51 remote management) | -80'000 | -80'000 | -80'000 | -80'000 | -80'000 | -400'000 |
| AWH sorption maintenance | 0 | -7'500 | -10'000 | -10'000 | -10'000 | -37'500 |
| AWH depreciation (75k / 10 years) | 0 | -5'625 | -7'500 | -7'500 | -7'500 | -28'125 |
| Total Operating Costs | -120'000 | -133'125 | -137'500 | -137'500 | -137'500 | -665'625 |
| Hardware Reserves | ||||||
| ASIC reserve (4'500'000 kWh × 0.02) | -90'000 | -90'000 | -90'000 | -90'000 | -90'000 | -450'000 |
| GPU reserve Low-End (kWh × 0.02) | -10'000 | -2'500 | 0 | 0 | 0 | -12'500 |
| GPU reserve High-End (kWh × 0.15) | 0 | -98'550 | -131'400 | -131'400 | -131'400 | -492'750 |
| Total Hardware Reserves | -100'000 | -191'050 | -221'400 | -221'400 | -221'400 | -955'250 |
| Total Costs | -870'000 | -1'010'835 | -1'057'780 | -1'057'780 | -1'057'780 | -5'054'175 |
| Free Cashflow (before split) | 30'000 | 383'178 | 500'904 | 500'904 | 500'904 | 1'915'890 |
| Farmer profit share (30%) | 0 | -114'953 | -150'271 | -150'271 | -150'271 | -565'767 |
| K51 net cashflow (70%) | 30'000 | 268'225 | 350'633 | 350'633 | 350'633 | 1'350'123 |
Capital expenditure (Year 2):
| Item | CHF |
|---|---|
| AWH sorption unit (200 kW) | -75'000 |
| GPU upgrade (Low-End → High-End, 100 kW) | -150'000 |
| Total CAPEX | -225'000 |
K51 net cashflow after CAPEX (5 years cumulative): CHF 1'125'123
| Scenario | NPV K51 (5 years) |
|---|---|
| A: Without AWH, without Innosuisse | CHF ~123'000 |
| B: With AWH, with Innosuisse | CHF ~813'000 |
| Delta | CHF ~690'000 |
Without public co-funding, development would proceed at a slower pace:
Innosuisse funding is therefore critical not only for reducing financial risk, but for accelerating market entry by enabling parallel development of the laboratory prototype and commercial deployment.
K51 is an established, revenue-generating company. The AWH integration does not require K51 to find a new business model — it enhances an existing, proven model.
Self-sustainability is achieved through:
After the Innosuisse project concludes (end of Year 2), K51 operates the AWH-integrated system commercially without further public funding.
B2B. K51 sells infrastructure services (heat, water, dehumidification) directly to commercial greenhouse operators under long-term contracts.
K51's market access is relationship-driven and infrastructure-based, not mass-market:
| Phase | Timeline | Milestone |
|---|---|---|
| Innosuisse project start | Q1 2026 | Funding approved, team mobilized |
| Laboratory prototype (10 kW) | Q2–Q4 2026 | Performance validation at Empa |
| Containerized pilot (200 kW) | Q4 2026–Q1 2027 | System built, integration tested |
| Field deployment at Imhof Bio | Q1 2027 | First real-world operation |
| Performance validation & optimization | Q1–Q4 2027 | 12 months of operational data |
| Commercial rollout (2nd installation) | Q1 2028 | Proven system deployed at new site |
| Scale-up (5+ installations) | 2028–2030 | Series production of AWH units |
Proof of early market traction:
Existing installation at Imhof Bio, Schwerzenbach: 2 × 1 MW K51 containers operational, with thermal integration into the greenhouse heating system. This site is confirmed as the location for the AWH field pilot.
Active expansion in France: 4 × 1 MW installations planned in Nantes region for 2026, with farmer partnerships established through Les Maraîchers Nantais (Flavie Morin, Directrice).
Institutional partnerships: Formal collaboration with Empa (Dr. Binod Koirala, Reto Largo) and Solabs (Fabrice Bagnoud, ETH Zurich environment).
Customer demand validated: Greenhouse operators have expressed interest in combined heat + dehumidification solutions. The decarbonization pressure from retail customers (Migros, Coop) creates urgency.
MoU signed: A Memorandum of Understanding between K51 and the Solabs technology team governs the collaboration framework, IP principles, and commercialization path.
The project transforms datacenter waste heat — currently the largest source of inefficiency in digital infrastructure — into clean water and climate services for agriculture. It demonstrates that computing and food production can be symbiotic rather than competing uses of energy and land.
The project creates value at multiple levels:
The investment is justified by the combination of strong commercial viability (see financial plan), significant environmental benefit, and the creation of exportable know-how through the Empa–ETH–K51 collaboration.
| Item | Status | Timeline |
|---|---|---|
| IP ownership and licensing terms (K51 exclusive rights for greenhouse application) | To be discussed with Solabs team | Next week |
| Detailed budget allocation between K51, Empa, and Solabs | To be jointly developed and challenged | Before submission |
| Letter of Intent from Imhof Bio (pilot site confirmation) | To be obtained | Before submission |
| Exact Innosuisse co-funding ratio and K51 cash contribution | To be confirmed with Innosuisse | Before submission |
| NewCo spin-off structure (if applicable) | Preliminary, not required for submission | Post-project |