US data center capacity could triple in five years! After the "blue wave," who can fully capture the power dividend? Goldman Sachs is bullish on these seven companies.

Title context: US data center capacity could triple in five years! After the "blue wave," who can fully capture the power dividend? Goldman Sachs is bullish on these seven companies. Text: As Democrats won control of both the US Senate and House of Representatives and secured state-level election victories in data center hubs such as Texas, Georgia, and Ohio, AI power investment now faces two forces simultaneously: long-term AI data center capacity demand continues to be significantly revised upward, while project interconnection, cost sharing, and regulated returns face stricter scrutiny. A Goldman Sachs research report dated November 2 sharply raised its 2030 global data center capacity forecast from 217GW to 273GW, an increase of about 26%; among this, Goldman Sachs expects US data center capacity to surge from 49GW in 2025 to 152GW, reaching about 3.1 times the original level, with the massive expansion space continuing to support long-term demand across the power chain. Two days later, the Goldman Sachs research team led by Carly Davenport noted that state-level political changes could delay the inflection point for load growth, affecting Texas wholesale electricity prices and the pace of utility investment, but still viewed Texas as an important long-term growth market and maintained "Buy" ratings on core data center power chain companies such as Vistra and NRG Energy. What these two Goldman Sachs reports jointly point to is that the global acceleration of AI computing infrastructure expansion continues to provide enormous demand for power resources, while the value of power assets depends on whether these demands can be quickly converted into electricity sales cash flow on schedule and into officially approved data center investment returns. After Democrats won congressional control and key state elections, the market is reassessing the preconditions for converting these demands into profitspotential development restrictions in Texas affect new load interconnection and forward electricity prices, changes in Georgia's regulatory landscape bear on cost recovery and reasonable returns for power investment, policy proposals in Ohio increase uncertainty for project advancement, and California wildfire liability reform may ease valuation discounts for relevant companies. In this latest post-midterm-election power stock research report, Goldman Sachs maintained "Buy" ratings on NRG Energy, Vistra, PG&E, Sempra, Xcel Energy, FirstEnergy, and Duke Energy. The investment thesis behind this list is to find, within continuously growing AI electricity demand, companies capable of converting power supply resources, customer load, and approved investment into sustainable cash flow. In Goldman Sachs' view, future differentiation among power stocks will depend more on when projects connect, how electricity prices form, who bears investment costs, and ultimately how much growth can be realized as shareholder returns. The market has already begun distinguishing between two logics: incremental and existing data center power supply. In October, the Nasdaq 100 and the Philadelphia Semiconductor Index fell 6.6% and 15.4%, respectively, while Sempra, Xcel Energy, and American Electric Power rose about 10.4%, 13.3%, and 9.8%, respectively. On November 4, after election results were announced, Vistra, NRG Energy, and Constellation Energy plunged 13.83%, 13.52%, and 6.71%, respectively, but Xcel Energy, Duke Energy, and Edison International still rose 2.21%, 2.43%, and 3.12%, respectively. This shows that the shock was concentrated in market expectations for new load, electricity prices, and project delivery, but not all utilities should be treated as the same AI trade; power chain leaders that have already secured massive data center orders remain favored by the market. This severe divergence in data center power chain company stock prices echoes the differences in policy impact emphasized by Goldman Sachs: slower new load interconnection may depress forward electricity prices and earnings expectations for some generators; the value of regulated utilities depends more on whether investment can be approved into rate base, whether costs can be recovered, and the regulatory environment in their regions. Therefore, AI electricity demand remains a long-term growth source, but the market is raising its requirements for growth realizationcompanies with reliable power supply resources, clear interconnection arrangements, and sustainable return mechanisms are better positioned to sustain valuation support. Texas reassesses electricity price outlook, Georgia reassesses investment returns: ballots reshape the path to power growth realization The key variable in Texas is when data center demand actually connects to the grid. Governor-elect Gina Hinojosa had advocated pausing development until more protections are established by legislation in 2027. Goldman Sachs believes the main risk is delayed construction and interconnection timelines rather than the disappearance of long-term demand: delayed load growth would weaken expectations of tightening supply and demand in the Texas electricity market (ERCOT), affecting forward electricity prices and contract values for companies such as Vistra and NRG Energy; for Sempra, Xcel Energy, and American Electric Power, it could delay transmission and distribution investment. The Goldman Sachs research report lists the data center audit study results expected before December 10 as a subsequent observation point. Meanwhile, KeyBanc analyst Sophie Karp downgraded Vistra from "Overweight" to "Sector Weight" due to policy uncertainty, indicating that Wall Street investment institutions already have significant differences in their tolerance for short-term risk. Georgia's impact goes deeper into the regulated return mechanism. In addition to the governorship, Democrats also won a majority of seats on the Public Service Commission, meaning large load interconnection, cost allocation, and rate approvals may face stricter scrutiny. About 56% of Southern Company's rate base is located in the state, and its potential large-load project pipeline of more than 75GW has therefore become a key focus for Goldman Sachs. The 75GW here is an opportunity pool, not all of which is signed or energized. For regulated utilities, increased demand can only profit after investment is approved, assets are placed in service, and costs are recovered; if regulators require developers to bear more infrastructure costs, project economics and commissioning pace will change, while the controversy over ordinary customers subsidizing large customers may also ease. Ohio data center expansion faces policy test, California seeks wildfire liability reform: power stock re-ratings each have key variables Ohio's core issue is coordination between large load growth and residential electricity bill affordability. Goldman Sachs expects about 8GW of data center capacity to come online locally by 2030, but after Amy Acton's victory, stricter environmental, disclosure, community benefit, and cost-sharing requirements may increase development uncertainty, relatively unfavorable for FirstEnergy and American Electric Power, with Duke Energy less affected. The Goldman Sachs research report shows that American Electric Power's 84GW of signed load goes further than its potential project pipeline, but still requires completed interconnection before contributing revenue; in addition, Goldman Sachs emphasizes that faster interconnection and additional investment from projects such as Bloom/Piketon are upside sources, while financing needs, the gap between actual and authorized return on equity, and data center rate policy determine whether growth can fully flow into shareholder profit. California's valuation theme is wildfire liability system reform. Governor-elect Xavier Becerra said relevant policies still need improvement, so Goldman Sachs continues to watch the special legislative session and the scope of reform. If institutional adjustments reduce uncertainty over major payouts and improve investment cost recovery, PG&E, Sempra, and Edison International may receive support from a decline in risk premium. In Goldman Sachs' view, especially if wildfire liability reform reduces the risk of massive compensation and increases certainty of reasonable cost recovery, PG&E, Sempra, and Edison International are expected to ease valuation discounts and create conditions for share price recovery. Investors worry that power companies may bear massive wildfire compensation or that invested costs cannot be recovered through electricity rates, so they demand higher expected returns as risk compensation, reflected in a lower share price they are willing to pay. This additional required return is the "risk premium" here. If institutional reform makes liability clearer and reasonable cost recovery more assured, the uncertainty borne by investors declines, and the required risk compensation also decreases. With future earnings expectations unchanged, a lower discount rate can support a higher valuation. Therefore, a "decline in risk premium" does not mean company profits are falling, but that the market may reduce its valuation discount on the stock. Sempra especially reflects regional divergence: its Texas business faces delayed load realization, while its California business may benefit from liability reform. Factors determining the company's valuation therefore also include the combined result of regional earnings contributions, regulatory arrangements, and cost of capital. From "capacity planning" to "chargeable assets": which data center-related power companies does Goldman Sachs favor GPU training clusters, continuously running inference services, and agent workflows convert computing demand into high-density, continuous power loads, while also driving construction of substations, transmission lines, distribution equipment, cooling, and backup power. But the profit mechanisms differ across segments: market-based generators mainly profit through generation volume, electricity prices, capacity revenue, and long-term contracts; regulated utilities mainly build rate base through approved investment, then recover costs and earn returns under authorized mechanisms. Therefore, project delays first affect electricity sales and pricing expectations for the former, and first affect the timing of capital deployment and earnings contribution for the latter; the extent also depends on hedging, contractual protections, and regulatory approvals. The Goldman Sachs capacity update reveals the scale of this investment cycle: global data center capacity is expected to rise from 103GW to 273GW, a net increase of 170GW over five years, with the US contributing 103GW, about 61% of global additions; by 2028, large projects above 75MW are expected to account for 75% of new US capacity. These figures point to larger-scale power source and grid supporting demand, but GW measures capacity and does not directly equal actual full-year electricity consumption. Assuming a full-stack cost of US$50 billion per GW, 170GW corresponds to about US$8.5 trillion in cumulative investment, though this also includes IT hardware and other expenditures and cannot all be regarded as utility electricity revenue. New cloud vendors have expanded potential demand sources, Asia-Pacific construction expectations have been revised upward, and European additions are more concentrated in the longer term. At the same time, the report cites average annual capital expenditure of about US$1.51 trillion from 2026 to 2030 for eight US AI and cloud companies, covering these companies' global investment and not all directed to the US domestic market. In Goldman Sachs' view, the most valuable power growth is growth that has locked in high-quality customers, clarified interconnection timing, implemented cost sharing, and can earn reasonable capital returns. Long-term expansion forecasts and post-election policy pressure converge precisely at this realization stage. Goldman Sachs' latest research report lists ratings on seven "Buy" and four "Neutral" data center power chain companies, and did not directly downgrade these ratings or target prices because of this midterm election. The following upside is uniformly calculated based on the November 4 closing price; target market value is approximately estimated assuming disclosed share counts remain unchanged, excludes dividends, and is not a target price reissued by Goldman Sachs after the closing plunge. The valuation methods reflect business differences. Vistra uses a blended valuation of 10x EV/EBITDA and a 7% free cash flow yield, while NRG uses 8.5x and 9%, so changes in electricity prices, cash flow, and the PJM capacity market are especially important; Constellation uses 21x price-to-earnings and a 5% free cash flow yield. Regulated companies are valued more on forecast earnings for the 5th to 8th quarters: Xcel, FirstEnergy, Duke, and PG&E are assigned 19.5x, 17.5x, 19x, and 10.5x price-to-earnings, respectively, while Southern and Edison are assigned 19.5x and 10x, respectively; Sempra and American Electric Power use sum-of-the-parts valuation. The premiums Goldman Sachs assigns to Xcel and Duke reflect transmission and renewable growth, generation investment, and regulatory improvement opportunities, respectively; PG&E's lower multiple highlights the constraint of wildfire liability on valuation. The expanded target price upside after the election means the market has raised its discount for realization risk, and subsequent recovery still requires evidence from interconnection, contracts, and regulated returns.