Could a stock market correction dilute the impact of AI capex on the US economy?

A single fact dominates US economic prospects for 2026: private investment continues its upward trajectory with exceptional vigour. Major technology companies have announced capital expenditure (capex) plans of a magnitude rarely seen outside recessionary periods. At the national level, investment spending serves as a direct driver of growth. Over an economic cycle, it also functions as an indirect driver through employment and income, boosting consumption dynamics. This provides the US economy with a form of short-term cyclical armour. 

This protection, however, is not unconditional. It depends on a variable that the economy cannot directly control: the reaction of financial markets. The reason is simple: for several years, US consumption has been fuelled by wealth effects. A severe – and especially persistent – market correction could therefore weaken consumption and reduce some of the gains associated with the investment stimulus. 

The relevant question is not whether markets can decline; they always can. The real question is quantitative: what level of sustained market decline would offset AI-related capex? Simply put analyses this scenario.

Read more: Multi asset: responding to extreme investment scenarios

An investment shock of macroeconomic magnitude

Figure 1 presents the most striking recent US macro news, which comes from the micro side of the economy. Aggregate announcements from major technology players point to a 50% increase in investment between 2025 and 2026, reaching approximately USD 700 bn. This represents a 2.3% contribution to US GDP and is therefore comparable to an economic policy shock. The surprise element of this capex is somewhat smaller, amounting to roughly USD 300 bn or slightly more than 1% of US GDP. 

These significant figures could grant the US economy a certain level of immunity to potential cyclical shocks in 2026. In a nutshell, US tech is not just a stock market story; it's the story of an entire economy. So, will the US economy be exceptional once again?

FIG 1. Announced investment plans and capex, and operating cash flow ratio¹

Figure 1 also depicts the relationship between the magnitude of company expenditure and its ability to finance the capex through internal cash flows. The capex-to-operating cash flow ratio should increase significantly in 2026, approaching 80%, before gradually declining in longer-term projections. Two key interpretations emerge, which are a potential source of market tension.

1.	In the short term, the spending is exceptional and mobilises a very high proportion of operating cash flow, making financial discipline a cause for concern
2.	In the medium term, an eventual rise in the ratio implicitly assumes an improvement in cash flow, which should provide sufficient future returns on these investments.

Can these plans be justified in terms of future profitability? The issue is not merely microeconomic. It could become macroeconomic if a market correction impacts consumption amid investor doubts about the profitability of the whole AI project. 

The wealth effect: a systemic transmission channel

Underestimating this wealth effect would be an analytical error. US consumption – particularly discretionary consumption – has been driven in recent years by the valuation of financial assets. Many recent macroeconomic developments can be explained by a simple mechanism: rising markets strengthen the sense of wealth, and a fraction of this wealth translates into spending, with a certain delay. 

Without delving into the methodological debates, the following pattern emerges: an increase in financial wealth translates into increased consumption, measurable over several quarters, with a stronger sensitivity to durable goods. This pattern aligns with classic research on the wealth effect and its transmission to consumption, particularly in syntheses and empirical analyses that measure the marginal propensity to consume (MPC) from variations in financial wealth.

Figure 2 illustrates this relationship: the correlation between quarterly S&P 500 Index returns and durable goods consumption (with a one-quarter lag) is positive across different sub-samples. Conversely, the relationship is nil or even negative when changing the lead or lag of the relationship. Thus, if a market correction were to occur as investment accelerates, the macroeconomic outcome would depend on the balance between the ‘real’ support from capex and the ‘financial’ brake of the negative wealth effect.      

FIG 2. A positive correlation between quarterly S&P 500 Index returns and subsequent durable goods consumption²

Read more: Higher real rates globally are boosting the outlook for large cap equities

Measuring the wealth effect

A pure wealth effect factor is difficult to construct. The financial wealth of American households is not limited to a single asset class or a single form of holding. Exposure is distributed across direct holdings, investment funds, pension funds and insurance, and varies considerably according to wealth levels. The complete mapping of this exposure is complex, incomplete and evolving.

Our simple approach to approximate the wealth effect begins with a target variable, durable goods consumption – which tends to increase when the sense of wealth improves – and then estimate its sliding sensitivity to several representative asset classes. Our framework uses four markets: bonds, equities, gold and bitcoin. Figure 3 presents the composition of the reconstructed factor for a US investor.

FIG 3. Composition of the wealth effect factor: bonds, equities, gold and bitcoin³

Such historical analysis is instructive. In the aftermath of the technology crisis, the implicit exposure of investors shifted away from equities (where it was previously significant) and towards bonds and gold. From 2010, exposure to equities was rebuilt. More recently, gold has regained a more visible place, and a bitcoin component has emerged (while remaining marginal). 

The key point is that US consumption currently appears sensitive to all four segments. In other words, the wealth effect channel has become multi-asset. Some would argue that its risk remains dominated by higher volatility assets: equities, gold and bitcoin – indicating multiple sources of vulnerability. It's worth noting simply that bitcoin currently seems to be a relatively minor player on this chessboard, and its recent 40% drop has had little impact on consumption.

Read more: Investing in a K-shaped world of fragmented returns

What level of correction might offset capex support?

We now consider our initial question: what level of sustained market decline would offset AI-related capex? Based on an SVAR model with contemporary restriction between investment and consumption (they tend to grow in tandem), our analysis in Figure 4 compares:

• The magnitude of the investment shock (1.1% of GDP)
• The estimated spillover effect on consumption (approximately 2% over a year, for a multiplier effect of 2, which remains reasonable)
• The market decline necessary to generate, via the wealth effect, an equivalent contraction in consumption.
   

FIG 4. Expected effects of a 10% investment shock on consumption and simulated equivalent decline in equities⁴

The result is notable – a sustained decline of approximately 18% (over 6 months) and 28% (over 12 months) in equity markets would be needed to neutralise the positive effect of the investment cycle on consumption. This does not describe an intraday correction or a temporary volatility episode – it represents a persistent decline that is long enough to transmit to consumption decisions. 
 
The current composition of the wealth effect factor further reinforces this diagnosis. More than a simple market accident is required. A prolonged sequence of declines would be needed, without bonds tempering the effect. Therefore, caution is warranted: declining markets can moderate the positive effects of the current investment cycle (not to mention that these investments could be cancelled or lowered), but a correction of around 20% would be necessary and, importantly, be sufficiently broad to affect the large portfolios held by US investors.

The AI-related investment programme constitutes exceptional US macroeconomic support in 2026. It directly stimulates activity and, through spillover effects, should support consumption in the medium term. The main limitation is unlikely to come from the real economy, but from a possible financial correction sufficiently durable to trigger a negative wealth effect of great magnitude. The quantitative benchmark is clear: only a sustained decline of about 20% in global markets would cancel the effect of the investment cycle on consumption.

Simply put, planned capex from large tech names makes the US economy highly resistant to a short-term slowdown, except if there is a severe and persistent correction in financial markets.

Macro/nowcasting corner

The most recent evolution of our proprietary nowcasting indicators for global growth, global inflation surprises and global monetary policy surprises is designed to track the recent progression of macroeconomic factors driving the markets.

Our nowcasting indicators currently show:

• Our growth nowcaster was flat this week, with a small increase observed in China. The signal continues to point to a low but declining regime
• Our inflation indicator increased slightly, approaching the high and rising regime. Only China showed a small decline
• Our monetary policy nowcaster declined slightly in China, while remaining flat in the US and the eurozone. Overall, it signals a low but rising regime.
   

World growth nowcaster: long-term (left) and recent evolution (right)

World inflation nowcaster: long-term (left) and recent evolution (right)
 
World monetary policy nowcaster: long-term (left) and recent evolution (right)

Reading note: LOIM’s nowcasting indicators gather economic indicators in a point-in-time manner in order to measure the likelihood of a given macro risk – growth, inflation surprises and monetary policy surprises. The nowcaster varies between 0% (low growth, low inflation surprises and dovish monetary policy) and 100% (high growth, high inflation surprises and hawkish monetary policy).