2026全球展望

BlackRockInvestmentInstitute Introduction Twice a year, BlackRock’s senior portfolio managers and investment executives gather for two days to debate the outlook for theglobaleconomy and markets–and its implications for portfolios. The Global Outlook is the culmination ofthatdebateand discussion at ourlatest Forum and informs our macroframing. What hasbecomeclearthis year: ourmegaforcesframework, launched a few years ago, is centered in reconciling the clashing orders of magnitude on investment and potential revenues in the AI buildout–and how it isintersecting with other mega forces, such as the energy transition and future of finance. Authors Investment Institute Investment Strategist—BlackRock InvestmentInstitute Quantitative Analysis–BlackRock Glenn PurvesGlobal Head of Macro—BlackRock Investment Vivek PaulGlobal Head ofPortfolio Research–BlackRock InvestmentInstitute Rick RiederHead of BlackRock Raffaele SaviGlobal Head ofSystematic– Rich KushelHead–BlackRockPortfolioManagement Group Scot FrenchFoundingPartner and Co-President, HPS, apartofBlackRock Raj RaoFounding Partner,PresidentandChiefOperating Officer,Global InfrastructurePartners, a part of BlackRock The global economy and financial markets are being transformed by megaforces, especially AI. Technology is becoming capital-intensive, and the AIbuildout could be unprecedented in both speed and scale. With a few megaforces driving markets, it is hard to avoid making a big call on their direction–and as such, there is no neutral stance, not even exposure to broad indexes. We Contents ThemesMicro is macroLeveraging up Mega forcesAI buildout faces constraintsFragmentation: AI and defenseFuture of finance evolvingquicklyPrivate credit enters new phaseInfrastructure: a timely momentRoom to run for EM9-1491011121314 AI is the dominant mega force right now, helpingpropel U.S. stocks to all-time highs this year. Inrecent months, investors have started to fretabout equity valuations and whether an AI bubblefront-loaded–and unavoidable to realize thebenefits–while revenues are back-loaded. Alongwith highly indebted governments, this creates amore levered financial system vulnerable toshocks–including bond yield spikestied topolicy is forming. TheShillerprice-to-earnings ratioshows U.S. stock valuations are the mostexpensive since the dot-com and 1929 bubbles.Market bubbles have arisen in all major historicaltransformations–and that could happen again.tensions between inflation and debtsustainability. We see private credit andinfrastructure supporting this financing. We gounderweight long-term U.S. Treasuries tactically. A rapid transformation Clashing ordersof magnitude Length and capital deepening of notable innovations, 1760-2040 The AI buildout could be faster andgreater than all past technologicalThe gap in time between capexneeds and eventual revenues means U.S.$5-8 trillion globally through2030, most of thatin the U.S. Thechallengefor investors: reconcilingthe huge capital spending planswith the potential AI revenues.Willtheir orders of magnitude match? necessary to realize eventual gains.Private sector leverage will add to aheavily indebted public sector,creating financial system vulnerabilities. Bond yield spikescould pose a risk to this financing. Chart takeaway:Getting inflation all the way back down totarget–the dotted green line–would require the Fed to deala significant blow to the economy.Chart takeaway:The AI transformation is unfolding fast and isdriving record investment—we could reach a level close to thelargest buildout in half the time. makes it conceivable for the firsttime. See next page. Even ifspending and revenues reconcile atthe macro level, there is still atiming mismatch. The AI buildoutSecond, a broadly higher cost ofcapital as this big borrowing putsupward pressure on interest rates.Third, before new pools of AIrevenues spread across the Forward-looking estimates may not come to pass.Source: BlackRock Investment Institute with data fromCrafts (2021),December 2025. Notes: The chart shows the average annual contribution of capital spendingto GDP growth for previous U.S. technologies (except “steam,” for the UK) against the length of time thecapital was spent. Estimates for steam, electricity and ICT are taken from Crafts (2021). The spend needed forartificial intelligence (AI) is calculated using realized capital spending between 2022-2024, the upper end ofthe U.S.$5-8 trillion range of total capital spending intentions spanning the period 2025-2030, and anassumption that capex intensity continues at a similar rate through 2030-2040. This investment environment has some requiring the need to make big calls, andmore opportunities for active views. Never broken out Growth breakout:now conceivable U.S. GDP per capita and long-term trend, 1870-2024 We see AI capital spending stillsupporting growth in 2026, withBut the big question is whether it’spossible for U.S. growth to eventually break out of