A procurement reference for hyperscale data center operators, EPC contractors, and ESG diligence teams building in Michigan. Prepared by MI Farm Waste Exchange, Greenville, MI.
Michigan's lower peninsula produces roughly 2.06 million acres of corn and 2.02 million acres of soybeans annually. The majority of harvest residue — corn stover, bean trash, wheat straw — is currently disked back into the field. A sustainable portion of this material (typically 30–50% removable depending on soil type) represents an annual, locally-sourced biomass feedstock pool measured in the high six figures of tons within a 100-mile radius of every announced Michigan data center site.
MI Farm Waste Exchange aggregates this supply at the farm level and matches it to operators, EPCs, and biomass processors who need verifiable, low-carbon process inputs for cooling-system support, on-site combined heat and power, or scope-3 ESG reporting. We do not bale, haul, or process. We hold the relationships, the data, and the contracts.
This document is a procurement reference, not marketing material. Sections 3 through 6 describe what we can supply today, what we are still building, and the contracting model under which we engage. Section 7 discloses what we will not claim until we have audit-grade data behind it.
1. Locality. Every announced Michigan hyperscale data center site is within 100 miles of significant corn-stover-producing acreage. Greenville (our base) sits roughly 60 miles north-northwest of Lyon Township and 90 miles northwest of Saline Township. Lake Odessa, our second pilot, is 60 miles from Saline directly. Average feedstock haul distances for participating buyers should range from 30–80 miles.
2. Sustainability documentation. Sourcing agricultural residue at sustainable removal rates avoids the soil-depletion arguments that complicate purpose-grown energy crops. It also avoids the deforestation arguments that complicate wood pellet imports from the southeastern US [2]. The carbon accounting is favorable: corn stover absorbed CO₂ during the growing season; residue that decays anaerobically in-field releases CH₄ (≈80× more potent than CO₂ over 20 years) [3]. Captured residue is a net win on both gases.
3. Political license to operate. The Michigan data center debate is publicly contested in 2026. The Michigan Attorney General has intervened in the DTE/Stargate power contract case [4]. Residents in Saline and Lyon Townships have organized. Procurement narratives that include verified local-farmer participation directly address the loudest objection — that data centers extract from rural Michigan without giving back. A farmer-paid feedstock chain is the cleanest answer available to that objection.
| Feedstock | Typical removable yield | Harvest window | Notes |
|---|---|---|---|
| Corn stover | 1.5–2.5 tn/ac | Oct–Dec | Primary feedstock. Best baling window depends on weather; storage required for off-season delivery. |
| Soybean residue | 0.5–1.0 tn/ac | Sept–Nov | Lower yield, faster degradation. Often blended with stover for pellet-grade applications. |
| Wheat straw | 1.0–1.8 tn/ac | Jul–Aug | Smaller pool in Michigan but valuable for off-season delivery scheduling. |
| Cover crop biomass | 0.3–1.5 tn/ac | Spring | Variable. Increasingly available as Michigan cover-crop adoption grows. |
MFWE's enrollment is starting in two pilot zones in West Michigan, both selected for high corn density and reasonable haul distance to announced data center sites:
| Pilot zone | County | Distance to Stargate Michigan (Saline Twp) | Distance to Verrus (Lyon Twp) |
|---|---|---|---|
| Greenville | Montcalm | ~90 mi | ~65 mi |
| Lake Odessa | Ionia | ~60 mi | ~55 mi |
Both zones sit inside the realistic biomass haul radius for combined heat and power applications (generally 50–100 miles, beyond which pelletizing and densification become economically necessary). Expansion priority counties — Kent, Ottawa, Allegan, Eaton, Calhoun — extend the network deeper into the announced data center corridor running from Ann Arbor west toward Lansing and Grand Rapids.
Hyperscale operators face growing pressure on Water Usage Effectiveness (WUE), Carbon Usage Effectiveness (CUE), and increasingly granular scope-3 supply chain disclosure. We're building the documentation chain to feed those reporting frameworks:
Closed-loop cooling has shifted from "alternative" to "expected baseline" in Michigan during 2025–2026. Stargate Michigan at Saline Township publicly committed to closed-loop cooling with daily water use comparable to a standard office building. That sets the bar for what other proposed Michigan sites should be expected to match. This section addresses what "closed-loop" actually means at engineering level, what's required to do it well, and how MI Farm Waste Exchange's residue-fueled CHP model extends the closed-loop story across additional dimensions: nutrients, methane avoidance, and farmer economics.
"Closed-loop cooling" is not a single technology. It spans at least three engineering approaches with very different implications:
Operators publicly committing to "closed-loop" should be required to specify which engineering approach, what the peak summer water consumption is, and what monitoring exists. Without that specificity, "closed-loop" is a marketing term, not an engineering commitment.
A 100 MW data center using conventional evaporative cooling consumes roughly 500,000 gallons of water per day on average, with peak summer demand often exceeding 1 million gallons per day. Some hyperscale facilities have permitted for peak demand up to 8 million gallons per day. These figures represent the cooling loop itself; total facility water consumption can be higher when humidification, sanitary, and other uses are included.
Closed-loop liquid cooling with dry coolers approaches zero ongoing water consumption. Closed-loop with adiabatic assist typically reduces water consumption by 70–90% versus full evaporative on peak days. True ground-source geothermal cooling consumes essentially no water after the initial loop fill.
The relevant procurement metric is not the cooling-system-type label but the actual gallons-per-day numbers at average and peak summer conditions, with measurement and reporting requirements written into the operating agreement.
Lower Peninsula glacial drift geology — unconsolidated sand, gravel, and clay 100–300+ feet thick over bedrock — is well-suited to vertical closed-loop ground-source heat exchange. Borehole drilling is regulated under Michigan EGLE Well Code Part 127, the same license that governs water wells. The contractor base exists. The U.S. DOE Geothermal Technologies Office is actively piloting Cold Underground Thermal Energy Storage (Cold UTES) configurations for data center cooling.
Borehole pricing in Lower Peninsula glacial drift, from Michigan contractor sources:
These figures cover the ground-side heat exchanger only. They do not include interior HVAC mechanicals, heat pump units, controls, or the connection between the borefield and the data hall. Mobilization economics favor large projects: a thousand-borehole job will typically quote per-foot rates at the lower end of these ranges due to crew efficiency.
Sizing context. General rule of thumb is approximately one borehole per ton of cooling capacity, drilled 150–250 feet deep, with 15–25 feet spacing between boreholes. A 100 MW data center cooling load translates to roughly 28,000 tons of cooling, implying a borefield in the tens of thousands of boreholes occupying on the order of hundreds of acres. This is meaningful land footprint and meaningful upfront capex. Specific sizing for any given site requires a thermal conductivity test and a 30-year thermal model. MFWE can introduce qualified EGLE-licensed Michigan drillers for site-specific quotes; MFWE takes no cut of contractor work.
Ground-source closed-loop cooling has a failure mode that responsible operators must address: a borefield absorbing heat continuously without recovery periods will warm up over time. This is called thermal depletion. The ground temperature in and around the field rises year over year, system efficiency degrades, and in undersized installations the surrounding soil and shallow groundwater can experience measurable warming that affects adjacent agriculture, septic systems, and shallow domestic wells.
This is a real engineering concern, not a hypothetical one. Three mitigations should be present in any responsible installation:
Operators evaluating Michigan sites for ground-source cooling should expect to provide: borefield sizing calculations based on site-specific thermal conductivity testing, 30-year thermal modeling with annual ground temperature projections, an off-take plan for any CHP waste heat that doesn't rely on the geothermal field as default sink, and a thermal monitoring plan that includes shallow groundwater temperature measurement at the property boundary. These are not unusual asks. They distinguish a responsibly-designed installation from a marketing claim.
CO₂ released when crop residues are combusted in a CHP plant is the same CO₂ the crop pulled from the air during the prior growing season. The carbon cycles within a single agricultural year. This is meaningfully different from fossil fuel combustion, which releases carbon sequestered for millions of years. Major carbon accounting frameworks (GHG Protocol, IPCC, EU ETS) treat properly-sourced biomass combustion as net-zero biogenic carbon at the stack, subject to specific sourcing and accounting requirements.
Crop residues left in waterlogged or anaerobic conditions decompose and release methane (CH₄). Methane has an atmospheric lifetime of approximately 9–12 years and a global warming potential approximately 81× that of CO₂ over a 20-year horizon (IPCC AR6). Combusting the same residue in a controlled CHP plant converts the carbon directly to CO₂ via the biogenic pathway above, bypassing methane formation. The avoided methane is a real, additional climate benefit beyond fossil-fuel displacement, particularly relevant for operators with 20-year emissions accounting horizons.
Combustion concentrates the mineral content of the residue. Potassium and phosphorus are theoretically retained in the ash; in practice, loss through ash dispersal during burning and handling can be significant unless deliberately captured. Nitrogen is essentially 100% lost in combustion. Sulfur is approximately 80% lost. Modern CHP plants with proper ash collection (electrostatic precipitators, bag filters) recover substantially more P and K than open-field burning, but ash return to originating fields still requires deliberate logistics: testing for trace metals, transportation routing, contractual obligations on application volumes.
Northern European wood-CHP operations have made ash return to source forests a contractual standard. The MFWE position is that any responsible residue-CHP arrangement should treat ash return logistics as a first-class part of the offtake agreement, not an afterthought. Farmers continue to need nitrogen inputs separately; the K and P recovery via ash is a partial nutrient cycle closure, not a complete one.
The model only works long-term if farmers participate, and farmers only participate if the per-acre economics work for them. Residue offtake provides a new revenue line during a sustained commodity-price compression cycle. Ash return reduces farmer fertilizer spend on K and P. Combined, residue-based revenue and avoided fertilizer cost can change the per-acre economics of corn and soybean operations meaningfully, without changing what or how farmers plant. This is the foundation of supply durability for any offtake counterparty.
Closed-loop systems are increasingly the expected baseline in Michigan. The MFWE position is that operators should be expected to (a) specify the engineering behind closed-loop claims, (b) commit to peak summer water consumption caps with monitoring, (c) document thermal management plans where ground-source cooling is involved, and (d) treat residue feedstock and ash return as contractual obligations where biomass CHP is part of the energy stack. We can facilitate introductions to EGLE-licensed Michigan drillers, residue-supplying farmers, and CHP equipment vendors. Operators contract directly. MFWE takes no commission on third-party contracts.
MFWE is positioned as an introduction and aggregation layer, not a logistics or processing company. Our role today is to connect buyers to verified farmers and to provide the documentation chain that supports ESG and procurement reporting. We do not bale, haul, store, or process feedstock — those services are arranged between the buyer and existing regional contractors, on terms the buyer controls.
Engagement options:
Standard engagement under any option includes a mutual NDA and a non-circumvention clause covering 18 months from introduction date. This protects both sides: farmers don't get cherry-picked out of the network, and operators don't get poached by competing buyers we've introduced them to.
If you're a procurement or diligence professional, you've read documents like this before. Most of them oversell. Here's what's actually true about MI Farm Waste Exchange in May 2026:
The investment thesis we're working from is that the residue supply chain in Michigan needs an aggregator, and that the political and ESG environment in 2026 specifically rewards local, transparent, farmer-paid sourcing. We think we can be that aggregator. We don't claim to be it yet.
If you're sourcing biomass into a Michigan project — or weighing one — we'd rather meet you at the diligence stage than at the contract stage. Quieter, more useful, and there's still time for the relationships to actually mean something at your permit hearing.