Ohio calls time on data-center tax break after cost balloons to $1.5 billion, 11x the initial estimate

Ohio is pausing a sales tax exemption for new data center applicants after the break surged far above prior projections, rising to $554 million in 2024 and nearly $1.6 billion in 2025 versus earlier estimates of $136 million and $142 million. The move increases regulatory and fiscal uncertainty for hyperscale AI infrastructure investment in the state, which has already seen about $37 billion of related investment in 2024-2025. Industry groups warned the pause could divert projects to other states, while a citizen-led ballot effort seeks to permanently ban new hyperscale data centers in Ohio.

Analysis

This is less about Ohio specifically and more about the first visible crack in the state-level subsidy regime that has underwritten hyperscale AI buildout. Once one large hub pauses incentives, neighboring states face a harder choice: either match the implicit subsidy race or accept losing capex, but in either case the margin capture shifts away from landlords and utilities toward governments and local communities. The second-order effect is that projects with marginal economics — especially those dependent on dense power interconnects, expensive cooling retrofits, or speculative pre-leased capacity — become more sensitive to timing, permitting, and financing costs.

The near-term market impact is likely not on the headline AI leaders, but on the industrials and power-adjacent supply chain that have been pricing in a straight-line pipeline of data center demand. Electrical gear, switchgear, cooling, and construction names tied to hyperscale spend can see order slippage if local opposition broadens or if developers slow final investment decisions while policy risk clears. At the same time, utilities and grid-builders with already-committed load growth should outperform peers exposed to greenfield expansion, because existing projects become relatively more valuable when new entrants face higher friction.

The biggest catalyst path is political, not economic: a ballot initiative or a change in state leadership could harden the pause into a multi-year structural headwind, while a quick administrative reversal would compress the risk window back to months. The contrarian read is that the market may be overestimating the fragility of AI capex — large platforms can redirect workloads and negotiate around state-level friction, so the lost activity may be delayed rather than destroyed. But even if aggregate demand persists, the cost of capacity is likely rising, which is bearish for local subsidy-dependent developers and bullish for incumbents with power, land, and permits already secured.

For portfolios, the cleaner expression is to fade the second-order beneficiaries of the data-center boom rather than the AI platforms themselves. The setup argues for a relative-value short in the most consensus-exposed electrical and cooling suppliers versus long utilities with locked-in load, and for buying policy volatility where optionality is mispriced. If the referendum gains traction, the overhang should persist into the election window; if it fails, the rebound could be sharp, so position sizing should assume binary headline risk rather than gradual drift.

Key Decisions for Investors

• Short SMR / PWR-style hyperscale infrastructure proxies into the next 2-6 weeks; use a tight stop if Ohio policy reverses, because the trade is driven by FID delays rather than canceled demand.
• Go long utility names with secured load and regulated returns, and short data-center-heavy power supply chains for a 3-6 month pair trade; best expression is long NEE or DUK vs short EMR/ETN on a relative basis if the market starts discounting capex slippage.
• Buy medium-dated call spreads on the main local-policy volatility proxy in industrial energy infrastructure names; the thesis is that headline risk around the ballot initiative will keep implied vol elevated through the July signature deadline and into the fall.
• Avoid chasing pure-play data-center REITs on pullbacks until permitting uncertainty clears; if you want exposure, wait for a confirmation catalyst rather than buying the first dip because occupancy is not the issue, incremental expansion is.
• If Ohio-style pauses spread to other states, rotate into national AI platforms and away from regional beneficiaries; the platforms can internalize higher capex, while subsidy-sensitive developers face a narrowing IRR band.