• Global equities have seen strong performance, and despite elevated valuations, supportive economic growth—particularly in the US—could allow robust returns to continue into 2026.

• While AI‑driven momentum remains a key theme, value‑oriented US equities and developed ex‑US markets increasingly offer relatively attractive risk‑reward profiles.

• Large‑scale AI investment carries uncertainty, and firms without strong competitive moats or pricing power may struggle to convert heavy capex into durable earnings growth.

Strong equity returns can continue – despite stretched valuations

Global equities ended 2025 with double-digit returns¹. That marked the sixth year since 2019 that global stock market performance exceeded 10% (only interrupted by the pullback in 2022, when high inflation prompted central banks around the world to sharply increase interest rates).

Underneath last year’s strong returns, US equities quickly recovered from the losses around the announced trade tariffs in April. Indeed, US stocks went on to reach all-time highs in the second half of the year – and there are good reasons why we may see a continuation of solid returns in 2026. We expect US economic growth to accelerate to 2.25%², as investment into AI continues and fiscal policy supports consumers and businesses.

This expectation is in spite of relatively high US equity valuations – but even at current stretched valuations, such momentum would not be unprecedented. In the summer of 1997, for example, the CAPE³ ratio exceeded its 100-year peak, only to rise by a further 36% over the next 30 months before the dot-com bubble eventually started to burst.

Opportunities exist regardless of AI’s impact

Despite the possibility of further rises from growth equities, our conviction is strengthening that the value-oriented parts of the US equity market and non-US developed markets provide attractive prospects. And this view stands whether we look at a scenario in which the productivity benefits of AI lead to an economic boom or a scenario in which current high expectations are not met.

We attach the highest probability to the scenario in which AI changes the way we work and live, and productivity increases significantly. But for this to happen, it would require a broadening of AI benefits to “AI adopters” and companies across all sectors of the economy, leading to higher productivity and profitability.

Our analysis also suggests that the net present value (NPV) of current and projected AI investment is far from certain. In the US alone, investment commitments in AI-related infrastructure going out to 2027 well exceed $2 trillion; however, we think it’s likely that a significant portion of this amount have a negative NPV. High initial and ongoing maintenance expenditures could erode the profit margins of companies with more limited pricing power, making it difficult to deliver the earnings growth that markets currently expect.

Positive NPV of AI spending only for firms with strong competitive moats and cheap capital

Notes: This chart aims to estimate the net NPV of AI-related investment. It assumes investment, including R&D spending and capital expenditure, by AI and AI-related companies of $3.1 trillion from 2025–2027. AI and AI-related companies include Amazon, Meta, Alphabet (Google), Tesla and other companies involved with semiconductors and semiconductor equipment, software, tech hardware and electrical utilities. Earnings before interest and taxes (EBIT) assumes revenue equal to the incremental real GDP in the megatrends “AI wins” relative to “deficits dominate” scenarios (Davis, 2025); a seven-year straight-line depreciation schedule for current and expected capital expenditure; and an earnings before interest, taxes, depreciation and amortisation margin of 37% (75th percentile of the S&P 500). We also assume that 40% of value creation is captured by shareholders. Expected EBIT over the next 25 years is discounted back at a baseline rate of 15%. “Strong moats” refer to business value capture of 70%, and “Weak moats” refer to business value capture of 20%. “Low risk” refers to a discount rate of 10%, and “High risk” refers to a discount rate of 25%.

Sources: Vanguard calculations, based on data from Bloomberg, as at 25 October 2025.

Higher long-term expected returns outside US growth equities

The above assessments inform our constructive outlook for non-AI scalers, and specifically for value-oriented US equities and developed markets outside the US. For both of these sub-asset classes, we project 10-year annualised returns of roughly 7%. These segments offer more reasonable valuations, healthy dividend yields and the prospect of realising efficiency gains.

At the same time, US value and developed ex-US equity markets could be better positioned if AI underdelivers on productivity gains, a scenario which would expose today’s high valuations in US growth stocks and parts of emerging markets.

10-year forecasts: Returns and volatility

Any projections should be regarded as hypothetical in nature and do not reflect or guarantee future results.

Notes: The forecast corresponds to the distribution of 10,000 VCMM simulations for 10-year annualised nominal returns in EUR for assets highlighted here. Asset-class returns do not take into account management fees and expenses, nor do they reflect the effect of taxes. Returns do reflect the reinvestment of income and capital gains. Indices are unmanaged; therefore, direct investment is not possible. Benchmarks used for asset classes: US equities – MSCI USA Total Return Index; US value-oriented equities – market-capitalisation weighted portfolio of stocks with a price/book ratio in the lowest one-third of the Russell 1000 Index; Developed markets ex-US equities – MSCI World ex USA Total Return Index; Euro area equities – MSCI European Economic and Monetary Union (EMU) Total Return Index; Emerging market equities – MSCI Emerging Markets Total Return Index.

Sources: Vanguard calculations, as at 31 October 2025.

IMPORTANT: The projections and other information generated by the VCMM regarding the likelihood of various investment outcomes are hypothetical in nature, do not reflect actual investment results and are not guarantees of future results. Distribution of return outcomes from the VCMM are derived from 10,000 simulations for each modelled asset class. Simulations are as at 31 October 2025. Results from the model may vary with each use and over time.

¹ Source: Refinitiv, as at 31 December 2025. The MSCI All Country World Index returned 22.9% in 2025, in total return USD terms. In GBP terms, the MSCI All Country World Index returned 14.4%.

² Source: Vanguard economic and market outlook, as at 10 December 2025.

³ The cyclically adjusted price-to-earnings (CAPE) ratio is a valuation metric that compares a market’s current price to its average inflation‑adjusted earnings over the past 10 years, helping investors assess whether equities are historically over‑ or undervalued.