全球投资组合经理摘要：处处有转折点

Global Portfolio Manager's Digest Inflections Everywhere SIGNATURE We provide context and perspective on research acrossregions and asset classes. This week, we examine theaccelerated ramp in AI demand; discuss future AI winners infinancial services and introduce the AI harness; and analyse Equity Product Management Group Terence Malone*+ 1 212 526 7578terence.malone@barclays.com Rob Bate*+44 (0)20 7773 3576rob.bate@barclays.comBarclays, UK •AI Acceleration:We believe that AI adoption is accelerating faster than expected, withdemand materially exceeding prior assumptions and steepening the industry’s S-curve,supported by native AI applications such as Claude Code and Codex, alongside underlyingmodel improvements. We highlight that gains in orchestration, tool calling, and modelaccuracy are driving enterprise adoption much faster than previously forecast, pullingforward revenue expectations across the broader AI stack. As a result, we expect AI revenue FICC Product Management GroupJennifer Cardilli*+1 212 526 8351jennifer.cardilli@barclays.com Jill Nentwig*+ 1 212 526 5129jillian.nentwig@barclays.comBCI, US •The AI Harness:We think that winners in financial services AI will be those that can mosteffectivelyconvert expert workflows into governed AI-native systems. We believe the AIHarness is a secure, observable, model-agnostic control layer that connects AI systems tofinancial workflows in a governed manner. We think the concept is not simply another Thisdocument 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 its Barclays Capital Inc. and/or one of itsaffiliatesdoes 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 couldaffectthe objectivity of this report. Investors should consider this * This individual is a member of the Product Management Group and is not a Research Analyst All research referenced herein has been previously published. You can view the full reports,including analyst certifications and other important disclosures, by clicking the hyperlinks in FOR ANALYST CERTIFICATION(S) PLEASE SEE PAGE 34.FOR IMPORTANT EQUITY RESEARCH DISCLOSURES, PLEASE SEE PAGE 34.FOR IMPORTANT FIXED INCOME RESEARCH DISCLOSURES, PLEASE SEE PAGE 35.Completed: 06-Jun-26, 00:55 GMTReleased: 07-Jun-26, 13:00 GMT governance and control plane will increasingly determine which financial institutions lead or Recalibrating Personal Auto:We conducted a deep dive of the personal auto landscape toreassess loss cost trends, pricing dynamics, and reserve adequacy to inform our updatedmargin outlook. Recent company commentary and data points suggest a more challengingbackdrop; as a result, we recalibrate our attritional loss ratio assumptions, driving up our lossratio outlook for 2027 and reducing our EPS estimates. That said, carriers retain meaningfulcapacity for favorable prior-year development throughout 2026, although we expect a much Events Best of Barclays U.S. Internet & Semiconductors AI's S-Curve Is Steepening EQUITY RESEARCHU.S. InternetPositive Ross Sandler BCI, US | Alex Hughes BCI, US | Michael DiSanto BCI, US | Owen Clendenin BCI, US |Alexander Kessinger BCI, US | Tom O'Malley BCI, US | Kyle Bleustein BCI, USExcerpted from U.S. Internet & Semiconductors: AI's S-Curve Is Steepening, published on June 1, The ramp in AI demand thus far in '26 has been astonishing. We estimate that the threeleading labs could generate over $130B in revenue this year, with exit ARR near $200B, The Key Takeaway:The AI adoption "S-curve" has bent up quite a bit lately on the back ofnative AI applications like Claude Code and Codex, coupled with underlying modelimprovements. Advancements in orchestration, tool calling, and the underlying model accuracyare driving up enterprise adoption far faster than previous forecasts. This dynamic has pulledrevenue estimates forward across the entire AI ecosystem, from all the major labs to many Key Points From This Report Include: 1.We have revised up our framework for modeling AI industry growthand, as mentionedabove, now estimate over $130B in revenue in 2026 from the largest labs, a milestone that 2.Our overall AI capex forecast nudges a bit higher, but is starting to reflect this bull case.We now think the capex peak gets pushed farther out beyond 2028 with ~36 GW of next-gencompute infrastructure, and levels outthereafter.The revenue ramp is driving up inferencedemand near-term, and putting stress on the overall AI capacity picture (i.e., demand 3.The unit economics are improvingat the AI labs and at the hyperscalers. Recent modelimprovements point to better tokenefficiencyper task combined with higher revenue pertoken, which is driving inference gross m