Barclays_Global_Portfolio_Manager_s_Digest_Who_Has_the_Power

This document is intended for institutional investors and is not subject to all of theindependence and disclosure standards applicable to debt research reports prepared for retailinvestors under U.S. FINRA Rule 2242. Barclays trades the securities covered in this report for itsown account and on a discretionary basis on behalf of certain clients. Such trading interestsmay be contrary to the recommendations offered in this report.Barclays Capital Inc. and/or one of its affiliates does and seeks to do business with companiescovered in its research reports. As a result, investors should be aware that the firm may have aconflict of interest that could affect the objectivity of this report. Investors should consider thisreport as only a single factor in making their investment decision.* This individual is a member of the Product Management Group and is not a Research AnalystAll research referenced herein has been previously published. You can view the full reports,including analyst certifications and other required disclosures, by clicking the hyperlinks in thispublication or by going to our Research portal on Barclays Live.FOR ANALYST CERTIFICATION(S) PLEASE SEE PAGE 32.FOR IMPORTANT EQUITY RESEARCH DISCLOSURES, PLEASE SEE PAGE 32.FOR IMPORTANT FIXED INCOME RESEARCH DISCLOSURES, PLEASE SEE PAGE 33.Global Portfolio Manager's DigestWho Has the Power?We provide context and perspective on research across regionsand asset classes, this week highlighting: our extensive analysisof the imminent AI energy needs as adoption increases; our diveinto China's cost advantage for producing EVs; and our mid-yearthoughts on European interest rates.• Powering AI Demand: We attempted to sanity check potential U.S. data center power needsbased on 1) what is implied by NVIDIA consensus revenue, and 2) disclosure on utilitycontractual visibility for Northern Virginia. Taking a step back, following decades of no-to-lowU.S. electricity demand growth, AI adoption is leading to increasing U.S. electricity demandforecasts. While a number of top-down forecasts suggest that data centers could reach up to9% of U.S. electricity demand by 2030, these forecasts have raised more questions thananswers due to a lack of underlying assumptions. Specifically, while many investors cite theIEA's estimate that DCs already account for ~4% of U.S. electricity demand, we believe the IEAoverstates current U.S. demand at ~200 TWh vs. our ~150 TWh estimate. Further, based onimplied GPU chip sales from NVIDIA consensus revenue and power required per GPU derivedfrom NVIDIA disclosure, we calculate ~12 GW / 90 TWh from AI alone in the U.S. by the end of2027. NVIDIA's market cap suggests this is just the start. And lastly, we note that NorthernVirginia is by far the largest U.S. data center market and has disclosed "high confidence"contractual visibility for >15 GW of regional data capacity by 2030. Assuming it holds its20-30% share, this implies total U.S. data center capacity of ~50-75 GW / 380-560