Higher capital costs shift equity leadership from multiple expansion toward earnings, diversification and stock selection

Overview

This global strategy report examines how AI-driven private-sector investment, expanding government borrowing and higher interest rates are raising the cost of capital. Goldman Sachs sees near-term bond-market risk but expects earnings growth, nominal GDP growth and broader market opportunities to support equities over a 12-month horizon.

Core views

Goldman Sachs identifies two linked themes dominating investor discussions: the impact of AI and the rise in interest rates. AI infrastructure investment is increasing private-sector capital expenditure and reducing free cash flow, prompting more debt and equity financing. At the same time, government borrowing needs are rising as spending priorities shift toward infrastructure, energy security and defense. Higher energy prices are also contributing to cyclical inflation pressure and higher policy rates. Together with geopolitical and AI-related uncertainty, these forces have raised the cost of capital; 30-year German and Japanese government bond yields were still close to zero as recently as 2022. The report notes that the increase in yields follows a near-record period of equity outperformance relative to bonds over 10-year holding periods. Equity risk premia have fallen back to levels last seen in the late 1990s, making equities more exposed to further yield increases, although the equity risk premium in the US and Japan has rebounded from recent lows. The effect of rising yields depends on their level, cause and, especially, the speed of adjustment. Historically, equities have generally delivered positive returns alongside rising rates unless the pace of the increase exceeded two standard deviations. A two-standard-deviation move in 10-year Treasury yields currently equates to roughly 50 basis points over a month or 30 basis points over two weeks; adjustments of that scale in recent days help explain the equity-market pullback. So far, strong nominal GDP and corporate-profit growth have offset much of the pressure from higher yields. Earnings growth has been the main contributor to equity returns in every region over the past 18 months. As a result, P/E multiples have been flat in Japan and Europe and have fallen in the US, Asia and emerging markets. In the US, the S&P 500 forward P/E declined from 22x at the start of the year to 19x, in line with its long-run average, even as the market remained near its all-time high. Better earnings across regions and a sharper US de-rating have also broadened geographic performance: since 2025, US equities have been the weakest among the major regions, reversing the post-financial-crisis pattern and increasing the contribution available from geographic diversification. Positive earnings revisions for 2026 and 2027 extend across major regions. Goldman Sachs attributes much of the profit growth to four areas: resilient technology earnings; higher commodity-sector profits driven by energy prices; strong bank earnings supported by nominal GDP growth, steep yield curves and healthy private-sector balance sheets; and industrial revenue gains from AI infrastructure spending by the sector's "pick and shovels" suppliers. This breadth increases the opportunity to diversify across both sectors and countries. AI investment is also altering financing conditions. Hyperscaler capex is increasingly consuming free cash flow, pushing companies toward credit and equity markets. Aggregate capex growth among AA-rated issuers was 65% year on year in the second quarter, the 10th consecutive quarter in which aggregate AA capex growth exceeded 35%. US convertible-bond issuance had reached $135 billion year to date, with AI-related borrowers accounting for an estimated 44% of issuance. Goldman Sachs' Credit team raised its full-year US investment-grade gross issuance forecast by $200 billion to $2.3 trillion; AI-related issuers account for one-quarter of USD investment-grade gross supply this year. Higher capital intensity and a higher cost of capital have reduced the value placed on distant cash flows and led to a de-rating of the largest technology companies, whose forward P/E is now close to the average for the rest of the US market. The report distinguishes this from a classic technology valuation bubble. Software has de-rated amid AI-disruption concerns, while memory and chip stocks have seen lower multiples despite strong earnings. Global technology's P/E has fallen below its 20-year median. In contrast, Healthcare, Consumer Discretionary and Industrials are more expensive than IT in absolute terms, and Industrials have re-rated to a P/E near the 90th percentile of their 20-year range. Goldman Sachs argues that the larger risk is an unsustainable earnings bubble rather than excessive technology valuations. It contrasts today's technology sector with banks before the 2008-09 financial crisis, when earnings were boosted by leverage and US real-estate financing and later collapsed with housing. Goldman Sachs sees important differences today: technology profits remain robust, balance sheets are broadly strong, and interest-coverage ratios for the aggregate S&P 500 and its median stock rank in the 99th and 68th percentiles, respectively, relative to the past 20 years. Compute demand is accelerating faster than supply, with Microsoft intending to triple data-centre capacity over six years and Nvidia reiterating a 2030 AI total-addressable-market outlook of $3 trillion-$4 trillion. Constraints increasingly relate to power, land and data-centre capacity rather than lack of compute demand. Nonetheless, a slowdown in profit growth amid a much higher cost of capital could pressure equity prices and weaken confidence in future cash flows across both hyperscalers and AI-infrastructure suppliers. Against this backdrop, Goldman Sachs remains neutral on equities and other asset classes over a three-month horizon because of near-term bond-market risks, but overweight equities over 12 months. It expects profit growth to slow from its recent pace but believes nominal GDP growth, a low perceived recession risk, earnings growth and return on equity can continue to support equities without valuation expansion. Low stock correlations across major equity markets, and sharply lower correlations even within technology, reinforce the report's preference for diversified exposure across geographies, sectors and factors. With higher bond yields limiting the scope for multiple expansion, differentiation, diversification and alpha generation are expected to be the main drivers of equity returns.

Analysis framework

The report links macro funding demand and interest rates to equity valuation through the cost of capital, then assesses whether earnings growth can offset that valuation pressure. It compares equity and bond performance, equity risk premia, the pace of yield changes, regional and sector valuation multiples, earnings revisions, financing issuance and historical banking-sector experience. It then uses current profit, balance-sheet, demand and correlation evidence to frame its short- and medium-term asset-allocation stance.

Methodology notes

• Valuation methods

  Forward P/E and equity risk-premium comparison

  The report compares forward P/E multiples with long-run and 20-year historical ranges, and uses equity risk premia to assess how much compensation equities offer relative to bonds as yields rise.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI-capex transmission across the technology ecosystem

  Goldman Sachs traces how hyperscaler spending reduces their free cash flow and increases financing needs while supporting revenues and cash-flow visibility for AI infrastructure suppliers and industrial beneficiaries.

• Event-Driven and Behavioral Finance

  Historical comparison of earnings bubbles

  The report contrasts current technology conditions with bank-sector earnings before the financial crisis to distinguish valuation risk from the risk that elevated earnings prove unsustainable.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Global equities

  Core asset class assessed relative to bonds and other asset classes.

  Strengths

  Earnings growth, positive 2026-27 estimate revisions, continued nominal-GDP growth and lower correlations across markets.

  Weaknesses

  Lower equity risk premia and limited scope for valuation expansion at higher bond yields.

  Comparison

  US equities have been the weakest major region since 2025 after years of relative dominance.

  Risks

  A sharp rise in yields or a slowdown in earnings growth could pressure prices.

• Technology equities

  Central beneficiary and funder of AI infrastructure investment.

  Strengths

  Robust profits, strong balance sheets, accelerating compute demand and supply constraints in power, land and data-centre capacity.

  Weaknesses

  Rising capex is reducing free cash flow and increasing reliance on debt and equity funding.

  Comparison

  Global technology P/E is below its 20-year median, while Industrials trade near the 90th percentile of their 20-year P/E range.

  Risks

  Profit growth could slow, reducing confidence in future cash flows throughout the AI ecosystem.

• Industrials

  Beneficiary of AI infrastructure investment and related demand for equipment and services.

  Strengths

  Improved revenues from AI-related capital spending.

  Comparison

  Industrials have re-rated and trade at a P/E close to the 90th percentile of their 20-year range, above Technology.

• Banks

  Sector whose earnings are supported by nominal GDP growth, steep yield curves and strong private-sector balance sheets.

  Strengths

  Strong current earnings backdrop.

  Weaknesses

  The report uses the pre-financial-crisis banking experience to illustrate the risk of earnings becoming unsustainable.

  Comparison

  Unlike today's technology sector, pre-crisis bank earnings were supported by rising leverage and financing of a US real-estate valuation bubble.

  Risks

  An earnings reversal can be damaging even without a major valuation bubble.

Key data

• S&P 500 forward P/E19xDown from 22x at the start of the year and in line with its long-run average.
• AA-rated issuer capex growth65% year on year in Q2The 10th consecutive quarter in which aggregate AA capex growth exceeded 35%.
• US convertible-bond issuance$135 billion year to dateAI-related borrowers accounted for an estimated 44% of issuance.
• US investment-grade gross issuance forecast$2.3 trillionGoldman Sachs raised the full-year forecast by $200 billion; AI-related issuers represent one-quarter of USD investment-grade gross supply.
• Two-standard-deviation 10-year Treasury yield moveAbout 50bps over one month or 30bps over two weeksThe report uses this pace as the threshold associated with recent equity pressure.
• Nvidia 2030 AI total addressable market outlook$3 trillion-$4 trillionReiterated at Goldman Sachs' annual Communacopia Technology Conference.

Impact & implications

The report argues that higher yields make valuation expansion less likely, placing greater weight on earnings delivery, return on equity and the sustainability of AI-related profit growth. It sees broadening earnings and lower correlations as supportive of diversification across regions, sectors and factors, while identifying a profit-growth slowdown and further rapid yield increases as potential sources of pressure.

Risks

• A continued sharp increase in bond yields could put pressure on equities, particularly if the speed of the move is unusually large.
• A slowdown in technology and AI-related profit growth could reduce confidence in future cash flows across hyperscalers and AI-infrastructure suppliers.
• Higher capital intensity and financing needs may further weigh on free cash flow and valuations for major technology companies.
• The report warns that apparently strong sector performance can become vulnerable if earnings prove unsustainable.

What to watch

• The pace and drivers of further increases in government-bond yields.
• Whether 2026 and 2027 earnings estimates continue to rise across major regions.
• Hyperscaler capital expenditure, free-cash-flow pressure and resulting debt, equity and convertible-bond issuance.
• Technology profit growth, compute demand and constraints in power, land and data-centre capacity.
• The persistence of low stock correlations and return dispersion across regions, sectors and technology companies.