← All docs

The proponent case — a neutral, precise read

Scope: the four proponent submissions to the Ohio Select / Joint Committee on Data Centers — Google (Liz Schwab, 6/4), QTS (6/3), the Data Center Coalition deck (5/27), and Ohio Dept. of Development / the DCTE (Director Lydia Mihalik, 5/27). Sources: select-committee-2026/; catalog in the hearing index.

Posture: neutral but precise. This does not argue for or against data centers or the abatement. It steelmans the proponent case, then assesses each load-bearing claim by how well it is supported — solid, contested, unverifiable-at-site-level, or outdated/mismatched-basis — and notes where the claim engages, or sidesteps, the disclosure question the committee is weighing. Figures are the witnesses’ own unless tagged otherwise.

The steelman (the proponent case at its strongest)

Stated as strongly as the record fairly allows:

1. The demand is real and structural. ~5.4B people online, 21 connected devices per US household, and cloud/AI/streaming growth drive a genuine, non-speculative need for compute (DCC). Compute capacity is genuinely constrained and leasing-based (a point the critic Parent also concedes).
2. The capital is enormous and largely privately financed. Google cites >$20B in Ohio (New Albany/Columbus/Lancaster since 2019); QTS a $10B Van Wert campus. These are decade-plus, multi-phase builds, not short-term plays.
3. Construction and induced employment are substantial, even if on-site operations jobs are few — a point all proponents concede rather than hide (Google: 4,200 construction jobs/avg US facility, 9 indirect per 1 direct; QTS: up to 4,500 construction jobs).
4. Some operators directly accept cost causation. QTS says it pays for interconnection and upgrades, agrees existing residential/small-business customers shouldn’t subsidize large loads, and cites BYONG (Bring Your Own New Generation) to “deliver more than we utilize.” Google takes service under PUCO-approved tariffs, runs a 100 MW PJM Virtual Power Plant, and invests in SMRs/geothermal. This is real engagement with the grid-cost problem, not denial of it.
5. The program has already been tightened. ODOD concedes the DeWine administration moved from 100% exemptions for up to 40 years to 50% for 10–15 years — a self-correction toward the critics’ fiscal concerns.

A fair reading credits all five. The precision below is about the gap between these claims and what the record actually establishes — and about a question none of the four answer.

Google (Liz Schwab) — the benefits-and-stewardship case

Argument: Google builds “the right way” — local investment (>$20B Ohio), 3,000+ workforce, Grow with Google (379,000 trained; $16M to nonprofits), 100% competitively procured power under PUCO tariffs, grid innovation (VPP, SMR/geothermal, HB 15), and water stewardship (replenish 120% of water consumed globally; Slim Creek wetlands). National security gets one sentence (“increasingly essential to securing American technological leadership and national security”); the workforce note includes “mission-critical fields like the military.”

Precise assessment:

• Rate relief (contested). The claim that large load “spreads fixed infrastructure costs… putting downward pressure on rates,” citing Georgia Power / Indiana-Michigan / DTE, is the strongest pro-industry economic claim — and it directly contradicts ELPC (AEP +$126M transmission, +$7.91/mo) and OCC in the same record. Both can be true conditionally: large load lowers rates only if it pays its full cost of service; otherwise it shifts costs. The claim is therefore conditional on exactly the cost- causation enforcement (HB 706) the critics want — Google’s own VPP/tariff posture implicitly accepts that.
• Water 120% (unverifiable-at-site-level). Replenishment is global and offsite (wetlands in the Buckeye Lake watershed), not a statement of Ohio-site operational consumptive use. It says nothing about how much the Lima/American Township design will evaporate locally — the precise figure the corpus’s cooling-basis work estimates at ~3–10 MGD for that site (36 cooling towers). A 120%-global-replenishment pledge and a large local consumptive draw are not mutually exclusive.
• The BOSC omission (notable). Google’s testimony names New Albany, Columbus, and Lancaster — but not the American Township / Lima project under scrutiny in the committee’s own home-county example, the one negotiated ~15 months under NDA. Not necessarily evasive (Lima is the newest site), but conspicuous given the hearing’s theme.
• Disclosure (not addressed). The testimony does not mention beneficial-ownership disclosure, per-site end use, or authorization posture at all. It argues benefits, which is orthogonal to Parent’s who-is-the-customer ask.

QTS — the most candid on the defense dimension

Argument: A diversified operator (90+ campuses) across hyperscale, colocation, and federal customers; $10B Van Wert campus; pays interconnection/upgrades; closed-loop cooling; community engagement.

Precise assessment:

• Federal/defense candor (solid, and corpus-relevant). QTS openly lists “federal customers, where we operate secure, compliant facilities that support national security and government operations” as one of three core segments. This is the clearest industry confirmation in the record that government/defense hosting is a real, named end use — corroborating the dimension Parent raised and Google only gestured at. It does not, however, disclose which facilities or what share — the per-site opacity Parent targets remains.
• Closed-loop water (solid-but-design-specific). QTS’s claim that operations use no additional water beyond office domestic use (~600,000 gal/yr) is credible for a closed-loop design — but it is design-specific, not an industry constant. The DCC’s own slide lists evaporative cooling alongside closed-loop, and the BOSC/Lima air permit shows 36 cooling towers (evaporative), which the corpus estimates consume millions of gallons per day. So “data centers barely use water” is true for some designs and false for others; the only way to know which a given Ohio site is, is facility-level reporting — Parent’s Recommendation 3. QTS’s strong number actually strengthens the case for per-site disclosure, because it is the exception that proves the variance.
• Cost causation (solid). The BYONG / “deliver more than we utilize” / “pay for interconnection” posture is genuine alignment with HB 706’s objective.
• Disclosure (not addressed). Like Google, no beneficial-ownership discussion.

Data Center Coalition deck — the macro evidence base

Argument: A sourced macro case — demand drivers (people/devices/AI), efficiency, jobs, and a rate-neutrality study.

Precise assessment:

• “Highly efficient consumers of energy” (outdated basis). The efficiency slide cites Science (2020) — data from the pre-generative-AI era. It does not capture AI-class rack densities (40–120+ kW), which is precisely the load growth now straining the grid. Using 2020 efficiency to characterize 2026 AI buildout is a basis mismatch.
• Jobs: 24,120 Ohio direct / 106,000 total / $8.2B GDP (mismatched basis, via PwC 2024). These are industry-wide installed-base figures (all facilities, direct + indirect + induced). They are not comparable to the 356 new permanent jobs from the 13 abated agreements Parent cites (Policy Matters / ODOD) — that is the marginal jobs-per-subsidy number. Both are real; they answer different questions. The policy question (what did the abatement buy) is the marginal one, where the proponents’ macro figure does not speak.
• Rate neutrality (contested). The E3 (May 2026) slide — “unclear relationship between [data-center] growth and rate hikes; state-level analysis shows no [clear correlation]” — is the industry’s rate defense, and it directly opposes ELPC’s specific AEP tariff figures. Macro non-correlation and specific tariff increases can coexist; this is a genuine empirical dispute, not a settled point in the industry’s favor.
• Water 39 BGY (US, 2025) (solid as aggregate) — a real national figure (Bluefield), but national aggregates say nothing about a specific Ohio watershed’s assimilative limit.
• Ownership models (slide 4): the deck distinguishes self-perform/enterprise vs multitenant/build-to-suit — i.e., it is fluent in facility/lease structure, which makes the absence of any beneficial-ownership discussion a choice, not an oversight.

ODOD / DCTE (Director Mihalik) — the program defense

Argument: The DCTE (HB 59, 130th G.A.) made Ohio competitive; Amazon (2014), Meta, Google followed; 18 agreements now (first 2013); the administration has shifted to a more targeted 50% / 10–15-year structure.

Precise assessment:

• The concession is the most informative part (solid). Moving from 100%/40yr → 50%/ 10–15yr is an implicit admission that the early deals were over-generous — aligning with the critics’ fiscal concern and with Parent’s “treat each abatement as a negotiation.”
• What’s absent (not addressed). The testimony gives no cost/revenue accounting (no foregone-revenue figure, no jobs-per-dollar) — the very numbers (≈$281.9M loss / 356 jobs; ~$554.9M in 2024 exemptions) that Parent and Policy Matters supply. The agency that administers the exemption did not quantify its cost to the committee. And it does not address who the beneficiaries ultimately are.

Claim scorecard

The precise neutral conclusion

Three findings a careful, side-neutral reader can state with confidence:

1. The proponent case is strongest where it is conditional. Its best claims — rate relief, “we pay our own way,” water efficiency — are true only under the safeguards the critics are asking for (enforced cost causation, facility-level reporting, a workable definition). QTS’s BYONG and Google’s VPP/tariff posture are the proponents themselves conceding the conditions. So the strongest pro-industry argument and the strongest pro-oversight argument point the same way: verify the promises.

2. The evidence is thinner than presented in three specific places — efficiency on pre-AI data, jobs on an industry-wide denominator offered against a marginal-subsidy question, and water on global/aggregate figures rather than Ohio-site consumptive use. None of these are false; each is a basis mismatch that a precise committee should not let pass as responsive.

3. None of the four engage beneficial-ownership disclosure — the actual question on the table. That is the precise reason the proponent case does not rebut Parent’s: it argues a different proposition (data centers bring benefits) than the one in dispute (Ohio should be able to name who it subsidizes and verify per-site claims). The two are orthogonal, which is exactly Parent’s thesis — disclosure doesn’t compete with the benefits case; it is the mechanism that would let Ohio confirm it. QTS’s water number and Google’s rate claim are the cleanest illustrations: both are checkable only with the per-facility transparency none of the four proposed and one (Google) had 15 months to provide locally and did not.

Bottom line for a neutral-but-precise stance: credit the demand, the capital, the cost-causation engagement (QTS especially), and the program’s self-correction as genuine. Discount the efficiency, jobs, and water claims to the extent they rest on mismatched bases. And note that the dispositive question — who is the customer, and what is the per-site end use and resource draw — is one the proponents did not answer, leaving the disclosure case substantially unrebutted on its own terms.

Cross-refs: mandamus-analysis.md (§II legislative slate; §III defense nexus), HYDROLOGY.md (the cooling-basis figures that test the water claims), and the witness catalog in the hearing index.