Morgan Stanley Research Analysis: Changing Conditions in Texas Stall 474GW of Grid Interconnections, Making Asset Revaluation from Electrification Imminent

Article by: Chaoxiang Research

The Democratic Party's unexpected victory in the Texas gubernatorial election has put the brakes on the nation's largest AI power supply tap.

Morgan Stanley analyst Stephen Byrd noted that the authority to limit data center expansion rests almost entirely with individual states. With the newly elected governor pledging to halt new approvals, 474 GW of interconnection applications within the Texas grid face sweeping delays or potential cancellations.

This local election has shattered the illusion of unlimited AI power supply. Compute projects lacking established power access face standstills, while existing power plants with secured transmission capacity are commanding unprecedented premiums.

Local Executive Orders Directly Block Interconnection Channels

The powers of the White House and Congress do not extend to local transformers.

The report indicates that the true approval leverage governing data center placement lies not in Washington, but in state public utility commissions, legislatures, and county-level administrative bodies. Incoming Texas Governor Gina Hinojosa explicitly campaigned on implementing a Moratorium for new data center approvals.

Upon taking office, the governor can act unilaterally via executive order without Republican-controlled legislative approval, directing state agencies to halt the issuance of pending permits.

This directly freezes the large-scale interconnection process at the Electric Reliability Council of Texas (ERCOT). Texas had previously paused certain interconnection approvals to clear backlogs in the grid queue, which then ballooned to 474 GW. This figure exceeds five times the state's historical peak electricity demand, with data centers accounting for 90% of the new power requirements.

With the executive order in place, the ongoing review of the zero-priority Batch Zero transmission plan could grind to a complete halt. Timeline extensions for unstarted projects to secure interconnection permits will stretch from previously expected months into indefinite limbo.

The Compute Landscape is Fragmenting into Two Opposing Camps

Policy risks are fracturing the United States into two distinct categories of regions: states eager to welcome compute capacity, and those actively driving it away.

In Ohio, Democrat Amy Acton’s gubernatorial win introduces similar headwinds to project deployment. The state’s most high-profile stalled project is NVIDIA’s planned 10 GW mega compute cluster developed alongside a leading U.S. AI laboratory. Meanwhile, Amazon’s 1.5 GW construction project in Fayette County is also feeling the pinch as tax incentive policies tighten.

However, in Pennsylvania and Louisiana, policy headwinds remain relatively manageable. Pennsylvania Governor Shapiro advocates a standardized certification framework focused on grid affordability and environmental compliance, rather than imposing direct bans on project approvals.

This rapid policy divergence across states is fundamentally rewriting the site-selection logic for AI compute infrastructure. While developers previously prioritized Texas’s low electricity rates and lax regulatory environment, uncertainty regarding approval timelines now overshadows pure cost calculations. Although the aggregate capital expenditure budget for AI across the U.S. may not contract, its geographic distribution is being forcibly reshaped.

Existing Assets Carry a ~30% Premium Over Current Stock Prices

As new project approvals stall, the intrinsic value of Independent Power Producers (IPPs) holding reliable, grid-connected assets surges abruptly.

Morgan Stanley re-evaluated the existing assets of key power sector equities. These estimates factor solely in cash flows from current assets, executed contracts, and forward power price curves, entirely excluding any potential premiums from future data center agreements.

The calculations indicate that the intrinsic value of these existing assets significantly surpasses their November 3 closing prices:

Vistra (VST): Intrinsic value stands at approximately $180 per share, a 28.57% premium over the $140 closing price.

NRG Energy (NRG): Intrinsic value stands at $129 per share, representing a 34.38% premium over the $96 closing price.

Talen Energy (TLN): Intrinsic value reaches $430 per share, a 20.45% uplift from the $357 closing price.

Constellation Energy (CEG): The calculated asset value is $249. While slightly below the $263 closing price, this figure entirely excludes potential nuclear direct-to-data-center contracts in the Pennsylvania PJM market. Factoring in anticipated nuclear contracting activity, the firm maintains its $378 price target.

Current market pricing for IPPs reflects extreme pessimism, assigning zero premium for future compute power supply contracts and even applying discounts to the baseline cash flows of existing power plants. Once the pricing power of these legacy generators is realized, they will inherently possess robust fundamental value support.

Divergence in Compute Infrastructure Chain Accelerates

Despite sharing the same AI power sector aura, underlying equities are exposed to policy shocks to vastly differing degrees.

Data Center Service Providers (PSPs) that rely exclusively on expanding through the Texas grid interconnection queue bear the initial impact. Galaxy Digital’s (GLXY) planned Helios II project (830 MW), initially assigned a 95% probability of timely commissioning, now faces rigorous reassessment. Similarly, lease probability assumptions for Cipher Mining’s (CIFR) Colchis and other new sites must be sharply lowered, with its valuation models reflecting a bear case of zero new projects at just $5 per share.

Conversely, the Behind-The-Meter (BTM) power supply model, which operates independently of Texas grid constraints, demonstrates clear advantages.

Williams Companies (WMB) leverages nationwide natural gas pipeline infrastructure to offer custom gas turbine cogeneration solutions to hyperscale cloud providers. By pre-ordering gas turbines, WMB can deliver power within 18 to 24 months of contract execution, entirely circumventing the multi-year waitlist associated with public grid interconnections.

Bloom Energy’s (BE) solid oxide fuel cells similarly boast near-zero emissions, negligible water consumption, and low acoustic footprints. In heavily regulated environmental states governed by Democrats, fuel cells are rapidly emerging as a direct replacement for conventional gas turbines.

TeraWulf’s (WULF) primary development projects are situated in Kentucky. Having previously navigated construction moratoria in New York, its management team possesses cross-jurisdictional regulatory navigation experience that insulates it from setbacks driven by Texas-specific policies.

Institutional Ledger

Chaoxiang Analysis

The weakest premise of this report is conflating the incoming governor’s anti-development rhetoric with permanent regulatory paralysis. Texas’s Republican-dominated state legislature remains deeply aligned with powerful oil, gas, and commercial lobbies; a blanket blockade on compute investment would inflict severe harm on state revenues and job markets.

Should the new governor fail to codify these restrictions into legislation within 12 to 24 months, the executive freeze will likely expire, allowing a massive backlog of projects to resume operations under bundled minimum-tariff contracts. Investors should stop fixating on nationwide AI compute sentiment and instead monitor the 2027 appointment cycle for the Public Utility Commission of Texas (PUCT), alongside the post-audit final authorization list for ERCOT’s inaugural Batch Zero transmission plan.

Disclaimer

This article compiles and interprets third-party broker research reports on behalf of Chaoxiang Research. All cited ratings, price targets, earnings forecasts, and analytical conclusions reflect solely the perspectives of the originating broker’s analysts, represent only their institutional viewpoints, do not constitute the views of Chaoxiang Research, and shall not be construed as investment advice.

Markets carry risks; decisions must be independent. This article should not serve as a basis for buying or selling any securities.