When AI capital expenditure peaks, will it be the day gold makes a comeback?

This round of gold's adjustment cannot be explained solely by "whether real interest rates are high or not." More critical are two lines: one is U.S. dollar credit, corresponding to U.S. fiscal pressure and central bank allocation demand; the other is financial attributes, corresponding to global liquidity, the dollar exchange rate, and risk-off trades. The problem now is that neither line is on gold's side.

In a study released on July 1, Zheshang Securities financial engineering analyst Chen Aolin and others gave a core judgment: "Monetary attributes and financial attributes are the two most important factors driving gold prices, and currently both are headwinds. The adjustment trend for gold may continue in the short term."

This framework incorporates AI capital expenditure into gold pricing. The logic is not convoluted: if tech giants' capital expenditure supports the U.S. economy, the U.S. government does not have to rely on larger fiscal deficits to underpin growth, market concerns about U.S. fiscal sustainability ease, and the medium-term allocation logic for gold weakens; at the same time, a strong economy and sticky inflation also compress the Fed's rate cut space, or even strengthen rate hike expectations, putting pressure on gold's financial attributes.

Therefore, for gold to regain stage-by-stage opportunities, it may not necessarily depend on geopolitical risks first, nor solely on a decline in U.S. Treasury real interest rates. A nearer observation window might be the July-August U.S. stock earnings season: storage and other segments have already seen sharp price increases. If AI capital expenditure plans are not revised significantly upward in tandem, it means the actual scale of construction is declining, which could be a signal for gold to stabilize.

Gold prices are not only priced by real interest rates; 2023 already taught a lesson.

In the past few years, the most common misjudgment about gold is that the market often compresses it into an "inverse function of real interest rates." When real interest rates rise, gold should fall; when real interest rates fall, gold should rise. This explanation works in many phases but is not complete.

2023 is a counterexample. At that time, the Fed was still in a tightening cycle, U.S. real interest rates were high, and under the traditional framework, gold should not have been strong. But gold prices clearly diverged from real interest rates, driven by U.S. fiscal policy: during the rate hike cycle, the U.S. government still adopted a significantly loose fiscal policy, debt service pressure rose, U.S. dollar credit was damaged, and gold received allocation support.

The second half of 2025 is another kind of dislocation. The U.S. fiscal deficit contracted, central bank gold purchases also declined, and the allocation logic weakened marginally. However, the Fed's rate cut expectations heated up, gold ETF inflows surged, and trading funds took over the dominant role, pushing gold prices higher. In other words, gold can be driven up by "U.S. dollar credit" or by "liquidity trading."

What do medium- to long-term capital allocations look at: how long can fiscal policy hold up, and does the central bank still trust the U.S. dollar?

Zheshang Securities' research framework clearly divides the factors affecting gold prices into two categories: one looks at medium- to long-term allocation logic, and the other at short-term trading sentiment.

The core of allocation logic is the U.S. government's ability to repay and the choices of global central banks. Specifically, two indicators:

U.S. fiscal pressure: The larger the fiscal deficit and the higher the ratio of government interest payments to revenue, the more the market worries about the U.S. being unable to repay its debts, U.S. dollar credit is damaged, and gold's allocation value becomes prominent.

Global central banks' attitude toward the U.S. dollar: If central banks around the world are reducing dollar assets and increasing gold holdings, they are voting with real money, expressing distrust in the dollar system. Once this force forms a trend, its support for gold prices is long-term.

This part explains medium- to long-term capital: it does not necessarily chase short-term trends, but once a direction is formed, it can often support gold prices in a longer trend.

What do short-term trading funds look at: dollar direction, safe-haven currencies, global liquidity tightness

Gold's financial attributes are closer to the language of short-term trading funds. The core of trading logic is whether market money is loose and whether everyone is afraid. Specifically, three signals:

First, safe-haven currencies (yen, Swiss franc). They have the same safe-haven attributes as gold, and their trends reflect market risk sentiment better than the dollar index.

Second, global liquidity. Gold does not generate cash flow. When liquidity is loose, the willingness to buy gold increases. Here, liquidity factors are synthesized from indicators such as global central bank policy expectations and global leverage levels.

Finally, financial stress. The OFR Financial Stress Index is used as a proxy for the pressure on the global financial system. When the financial stress index is above 0, gold prices tend to rise more easily. This is not the whole picture of long-term gold pricing, but under risk event shocks, it brings impulsive effects.

As long as one line is ignited, gold is not bad; when both lines are headwinds, that's the trouble.

The backtest results give a clear conclusion: gold does not need both the allocation factor and the trading factor to turn positive simultaneously; as long as one is activated, performance is usually not bad.

In the sample from January 2007 to June 2026, when both the allocation factor and trading factor were greater than 0, gold had the highest cumulative return. But when the allocation factor was positive and the trading factor negative, the monthly win rate was still 60.3%; when the allocation factor was negative and the trading factor positive, the monthly win rate was even higher, at 76.2%.

The truly unfavorable state is when both factors are below 0. This interval appeared for a total of 62 months, with an average annualized return of -8.6% and a monthly win rate of only 41.9%, lower than the full-sample win rate of 57.1%.

Based on this result, a comprehensive timing strategy is very simple: hold gold if either the allocation factor or the trading factor is greater than 0; go to cash if both are below 0. From January 2007 to June 2026, this strategy achieved an annualized return of 13.1%, while the London gold annualized return was 9.9%. It effectively avoided the sustained drawdowns of 2013-2015 and 2022, and also participated in several gold bull markets.

But the boundaries should also be clearly stated: looking at the allocation factor or trading factor alone can capture part of the upward moves, but neither can outperform the benchmark. The difficulty with gold is precisely that it is often priced by different funds in different phases in a relay manner.

The current problem for gold is that AI allows the U.S. economy to "go without fiscal stimulus."

The model has shown since March 2026 that both the gold allocation factor and the trading factor have turned negative. In other words, both monetary attributes and financial attributes are headwinds.

The key variable here is AI capital expenditure.

From 2020 to 2025, the correlation between U.S. economic performance and fiscal deficit trends was significant; loose fiscal policy was an important driver supporting economic strength. But since 2026, the two have diverged: the U.S. economic sentiment has improved, while the fiscal deficit is in a contraction trend. This means U.S. growth is increasingly driven by the private sector, not the government sector.

Breaking it down, domestic private investment has become the most core factor driving U.S. GDP growth. After tech giants' capital expenditure supports economic growth, the U.S. government does not need to continue relying on strong fiscal stimulus to maintain growth resilience. For gold, this weakens the logic of "U.S. fiscal unsustainability—U.S. dollar credit damage—rising gold allocation value."

On the other hand, AI capital expenditure supporting the economy also narrows the Fed's rate cut space. A strong economy combined with stubborn inflation raises rate hike expectations or at least suppresses rate cut expectations, and gold, as a non-yielding asset, naturally comes under pressure.

Gold's stage-by-stage opportunities may have to wait until AI capital expenditure expectations peak.

If AI capital expenditure is suppressing gold, then gold's stabilization signal may also come from changes in AI capital expenditure expectations.

The nearest observation window is the July-August U.S. stock earnings season. Storage and other segments have recently seen sharp price increases. If tech giants do not significantly revise their AI capital expenditure plans upward in tandem, it means that under rising costs, the actual scale of construction may decline.

Once the market starts trading on "AI capital expenditure peaking for now," the logic of the U.S. economy being supported by private investment will be re-examined, and the Fed's rate cut space may reopen. For gold, this could bring stage-by-stage allocation opportunities.

The risk is that this conclusion comes from historical data and model processing. Historical backtesting does not represent the future, and parameter settings can also affect results. Gold prices themselves are volatile; both factors turning negative does not mean gold prices can only fall unilaterally, but it at least indicates that a short-term reversal requires stronger catalysts.

Risk Warning and Disclaimer

        Market risks exist, and investment requires caution. This article does not constitute personal investment advice and does not consider individual users' specific investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment based on this is at your own risk.