The Midwest's rise to America's second-biggest data center hub is fueled by these key states

Texas and the Midwest now account for one-third of hyperscale data center capacity and are expected to capture 53% of new capacity in the next few years, as AI-driven demand shifts buildouts away from Northern Virginia. Power availability is the key constraint, with some projects in Texas bypassing the grid via on-site natural gas plants under the BYOP model. The report highlights Wisconsin, Indiana, Michigan, and Missouri as growing hubs, while Ohio continues to attract large projects through tax incentives.

Analysis

The center of gravity in AI infrastructure is moving from network-adjacent markets to power-adjacent markets, which changes the value chain more than the headline geography implies. In the near term, the scarce asset is not land or even chips; it is firm electricity with acceptable interconnect latency and permitting certainty. That favors vertically integrated cloud players and developers that can self-provision power, while compressing economics for markets whose advantage was historically fiber density and existing colocation ecosystems.

The second-order winner is likely the industrial energy stack rather than the hyperscalers themselves: gas turbines, switchgear, transformers, EPC contractors, and behind-the-meter generation suppliers should see a multi-year demand wave as more projects bypass the grid. A less obvious beneficiary is natural gas infrastructure in the relevant states, because on-site generation still needs fuel logistics, compression, and long-duration supply contracts. By contrast, regulated utilities in constrained regions face a paradoxical mix of higher capex demand and political backlash if load growth raises retail rates faster than they can recover costs.

The key risk is that the market is extrapolating a supply bottleneck into a permanent geography shift. If transmission buildout, interconnection reform, or state incentives accelerate over the next 12-24 months, the inland premium could fade and recent winners may see multiple compression. Another risk is policy: large on-site gas generation increases scrutiny around emissions, water use, and local tax concessions, which could slow permitting or force a costlier power mix.

From a positioning standpoint, this is a better relative-value than outright beta trade: the most attractive expression is long the infrastructure suppliers and power-enablers versus short the “picks and shovels” of conventional colocation. The hyperscalers remain net beneficiaries of AI capacity expansion, but the differentiator is who can secure megawatts fastest, so near-term project wins matter more than long-term addressable market narratives. The consensus is probably underestimating how durable the capex cycle becomes once firms internalize that power itself is the gating factor, not demand.

Key Decisions for Investors

• Long ETN or VRT into the next 3-6 months: both are levered to data center electrical intensity and should benefit if the power bottleneck persists; use pullbacks to build, target 15-20% upside with stop-loss on any evidence of order-book normalization.
• Pair long XLU-selected utilities with strong industrial load growth exposure against short colocation REITs over 6-12 months: the market is likely to overpay for legacy network proximity while underpricing grid-capex recovery in power-constrained states.
• Long AMZN and GOOGL on any 5-8% drawdown over 1-2 quarters: their scale and balance sheet give them the best optionality to self-fund power, but cap position size because near-term returns depend on permitting and capex discipline, not revenue reacceleration.
• Buy call spreads on NEE or other utility-adjacent power infrastructure names over 6 months: if utility interconnect and transmission spending accelerates, the rerating can be meaningful; define risk tightly because regulatory pushback can cap multiples quickly.
• Avoid chasing the obvious Texas/Midwest land play; instead look for upstream beneficiaries such as grid equipment and gas midstream names on weakness, since the more durable trade is the tools required to solve the bottleneck, not the geography itself.