In recent years, gold and AI technology have almost simultaneously become primary investment themes for global capital. Superficially, one represents a safe haven, while the other signifies growth. On a deeper level, they correspond to two entirely different worldviews: buying gold is a bet on a continued loosening of global order and ongoing questioning of US dollar credibility, whereas investing in AI is a wager on technology's ability to once again elevate the efficiency and return rate of the US economy.
In a strategy report dated July 13, a strategist from SDIC Securities wrote: "As we enter 2026, investors are approaching the ultimate showdown between gold and AI technology. In the current context of AI's comprehensive acceleration and deepening penetration, the transformation from micro-level evidence to macro-level efficiency is evident. Our position clearly leans towards the side of AI technology, and the scales are already tilting in its favor."
This is not to say gold will crash immediately. A more precise description is that gold may have reached a position analogous to the 2021 'Mao Index': the peak has likely already occurred, but the conclusive evidence that makes the market completely abandon it still requires time. The 'Mao Index' peaked after the 2021 Lunar New Year, but the falsification of the consumption narrative only became gradually clear in the second half of 2022 after pandemic restrictions were lifted. The situation with gold is similar now, where initial 'loosening evidence' appears first, not 'crash evidence'.
The most critical loosening stems from the US dollar. A significant portion of gold's past rally was built on narratives of a weak dollar, central bank gold purchases, and de-dollarization. Now, the 'weak dollar' is transitioning into a 'non-weak dollar'. If this is coupled with a potential restoration of US dollar credibility driven by AI capital expenditures and productivity improvements, the monetary premium gold previously enjoyed would be systematically compressed. The essence of a so-called 'M-top' pattern lies not in the chart shape, but in the sequence of trading logic reversals: the policy peak arrives first, followed by the data peak.
Gold's Current Resemblance to the 2021 'Mao Index': Belief Loosens First, Evidence Follows
The peak of the 2021 'Mao Index' did not occur only after everyone recognized the deterioration in fundamentals. On the contrary, at the time of the peak, only signs of loosening were evident: some baijiu (Chinese liquor) wholesale prices failed to rise, duty-free sales data declined, and expectations for overseas interest rate hikes intensified. What truly triggered large-scale institutional selling was the subsequent real-world verification that came later.
The current parallel with gold lies in the fact that the short-to-medium-term contradictions that previously supported it still exist: central bank gold buying, US dollar credibility debates, geopolitical conflicts, and fiscal deficits. However, these factors are already fully priced in, even amplified. The new variable is that the US dollar is not continuing its trajectory of weakness.
The upward shift in the oil price center is a key reason for the dollar's non-weakness. High oil prices elevate inflation expectations, reduce the scope for interest rate cuts, and even strengthen expectations for rate hikes. Following the shale revolution, as a major energy producer, high oil prices also support the US current account. For gold, this means its financial attributes are beginning to be constrained, while its commodity attributes are reasserting themselves.
This is the meaning of 'one peak is largely confirmed': it does not mean gold cannot rally, but rather that the previously unilaterally reinforcing narrative is beginning to loosen. The definitive evidence determining whether gold enters a major bear market may need to wait for the verification of AI productivity and US dollar credibility over the next 1 to 2 years.
The M-Top: Not a Simple Pattern, But a Timing Gap Between 'Policy Peak' and 'Data Peak'
Historically, gold bear markets have mostly originated from two forces: Federal Reserve rate hikes, or the rise of technology. The former is gold's most direct adversary; the latter's capital siphoning effect often causes larger drawdowns.
Since the collapse of the Bretton Woods system, gold has experienced multiple bull markets. Apart from a few special periods, most bull-to-bear transitions were sharp peaks. Only the periods 1975-1981 and 2010-2013 more closely resembled M-tops. The commonality of M-tops is: policy turns first, but the market does not fully believe it; subsequently, gold makes a secondary surge until economic and inflation data confirm the policy's effectiveness, at which point the right-side top is established.
In 1979-1980, after Paul Volcker's aggressive rate hikes, gold first formed its initial peak, but the market remained concerned about inflation resurgence. It was only after real interest rates turned positive, inflation was confirmed to be declining, and rising unemployment signaled economic cooling that gold's right-side peak was confirmed.
The period 2009-2012 was similar. Concerns over monetary oversupply from QE drove gold higher. In 2011, with the disappointment over QE3 and the introduction of Operation Twist, gold formed its left-side peak first. When QE3 was actually implemented in 2012, gold surged to form its right-side peak. Subsequently, as the US unemployment rate fell below 8% and consumer confidence recovered, the market confirmed that US dollar credibility was not collapsing, and the gold bear market began.
Currently, gold more resembles an M-top than a sharp peak. Pressure has already emerged on the policy front: the June FOMC meeting released hawkish signals, with half the FOMC participants projecting at least one rate hike this year according to the dot plot, and the median policy rate projection was revised upward; forward guidance was weakened. The data side is still undergoing verification. If subsequent inflation stickiness and employment resilience persist, the confirmation of the right-side peak will occur faster.
The Dollar's Shift from 'Weak' to 'Non-Weak': Gold's Most Comfortable Environment is Gone
The dollar framework can be broken down into four states:
Rate Cuts + Persistent High Inflation: Weak Dollar.
No Rate Cuts + Persistent High Inflation: Non-Weak Dollar.
Rate Cuts + Declining Inflation: Not-So-Strong Dollar.
No Rate Cuts + Declining Inflation: Strong Dollar.
Gold's most comfortable state is the first: falling interest rates, persistent inflation expectations, depressed real interest rates, and a pressured dollar. For a period, gold enjoyed precisely this environment.
The current situation is closer to the second state: no rate cuts, persistent high inflation, and a non-weak dollar. Oil prices, geopolitical conflicts, and US economic resilience are all compressing the room for rate cuts. Simultaneously, Europe faces weak growth, energy dependency, inflation pressures, and limited fiscal space, while Japan contends with low growth and high debt pressure. A stronger dollar does not necessarily mean the US has no problems, but rather that other major economies have more issues.
A more distant risk lies in the fourth state: if AI drives productivity improvements, leading to a downward shift in the long-term inflation center, and US interest rates are in no hurry to decline, the dollar would enter an even stronger state. That would be a more lethal combination for gold, as it would lose both the support of a weak dollar and the narrative of US dollar credibility collapse.
Technology's Rise May Hurt Gold More Deeply Than Rate Hikes
Gold's largest drawdown may not come from interest rates themselves, but potentially from the sustained capital drain towards technology assets.
The 1990s serve as an example. In the 1980s, computers proliferated in US businesses, but macroeconomic productivity showed no significant improvement for a long time—this was the Solow Paradox: computers were everywhere except in the productivity statistics. The reason was not that the technology was useless, but that businesses had not yet completed the restructuring of their processes, job roles, and organizational structures.
Around 1993, US information technology reached a critical inflection point. Businesses stopped merely buying computers and began redesigning their business processes around them. After 1995, productivity dividends were concentratedly released, the Nasdaq and the US dollar strengthened in tandem, and gold was sold off continuously. During the internet boom from 1995 to 2000, gold fell by approximately 25%.
The lesson from this history for today is direct: if AI remains merely a chat tool or a search replacement, it can hardly shake gold. However, if AI enters workflows such as customer service, R&D, coding, sales, legal affairs, and investment research, changing how businesses are organized, it will transform from a 'tool' into a 'productivity variable'. Once capital believes that the US economy's efficiency will be re-elevated by AI, gold's monetary attributes will be weakened.
AI is Building a New Capital Cycle for the Dollar
AI's support for the dollar is not simply about 'US tech companies being stronger'. More importantly, a new cycle centered around AI capital expenditures is taking shape.
US hyperscale tech companies are investing massive capital expenditures to purchase AI computing infrastructure, servers, storage, and networking equipment from Asia, particularly South Korea and Taiwan, China. After earning export surpluses, the Asian computing supply chain sees part of these funds flow back to purchase US dollar assets, indirectly supporting the next round of capital expenditures for US tech giants.
Estimates show that in 2025, the combined capital expenditures of five major tech companies—Microsoft, Amazon, Google, Meta, and Oracle—will reach approximately $449 billion, with a growth rate of 72%. This figure may further rise to around $805 billion in 2026. This scale is no longer just an internal tech industry cycle but a macro-level capital formation.
The proportion of AI-related exports in the GDP of South Korea and Taiwan, China, is also rising rapidly. The export surpluses are not fully converted into local currency appreciation or domestic money supply expansion but are retained in offshore markets in the form of overseas equities, deposits, and US dollar assets. As long as US AI capital expenditures do not systemically collapse, this cycle will provide a buffer against dollar depreciation.
This does not mean the dollar will only rise, but it weakens the one-sidedness of the 'de-dollarization' narrative. What gold fears most is not a single rate hike, but the market regaining belief in the long-term returns of US dollar assets.
The Key for 2026: Not Whether AI is Useful, But Whether Productivity is Delivered
The real threat of AI to gold lies not in how much stock prices have already risen, but in whether productivity gains transition from micro-level experiences to macro-level data.
Some early evidence is already visible: corporate AI adoption rates are increasing, with 65% of respondents reporting a positive impact of AI on work efficiency; efficiency improvements are more pronounced for structured tasks like customer service, writing, consulting, and coding. Related research from the BEA also indicates that since 2021, total factor productivity (TFP) in AI-intensive industries has grown at an average annual rate of 2.01%, while it declined by 0.41% in non-AI-intensive industries.
However, this is still not the final outcome. AI agent deployment is still in its early stages, and corporate process restructuring has just begun. In the short term, demand for data centers, electricity, copper, servers, and chips may actually push up costs, creating inflationary pressures. AI's productivity dividends are not realized linearly; they follow a sequence of capital expenditure first, then process restructuring, and finally entry into macro statistics.
Over the next 1 to 2 years, the key metrics to watch are not major model launches, but several hard indicators: non-farm business labor productivity, unit labor costs, prices and TFP in AI-intensive industries, the degree of AI workflow integration in enterprises, and data center and cloud capital expenditures. If these indicators collectively point towards efficiency gains, the US might replicate the combination seen in the mid-to-late 1990s: growth resilience stronger than expected, inflation pressures weaker than traditional models suggest, and US dollar credibility reinforced.
A Second Peak is Still Possible, But Gold's Winning Conditions Have Narrowed
An M-top pattern does not preclude another surge in gold. The second peak could come from subsequent Fed rate cuts or from setbacks in the AI industry's evolution. If AI capital expenditures are questioned, application-side delivery falls short of expectations, or the dollar weakens again, gold still has reasons to rally.
However, a rally and a primary trend are two different matters. In the past, gold's rise relied on three lines: central bank purchases, de-dollarization, and safe-haven demand. Now, it must simultaneously face several countervailing forces: a non-weak dollar, not-so-low real interest rates, AI capital siphoning, and a re-evaluation of US productivity.
The implications for equity markets are also clear: during a phase of no rate cuts and wavering expectations for hikes, gold tends to be weaker, while high-growth assets are more favored. Historically, during policy watch periods, the Nasdaq 100 has often outperformed the S&P 500, and growth styles have been stronger relative to value styles. The structural shift in 2026 may correspond precisely to the peaking of 'old Mao Index' assets like gold, and the continued rise of the 'new Ning combination' in technology and outbound investment directions.
Boundaries must also be acknowledged: if overseas monetary policy changes exceed expectations, or if AI productivity gains fail to materialize in macro data for an extended period, the verification of gold's M-top will be prolonged. However, if AI data continues to prove that the US economy is being reshaped by technology and US dollar credibility is being reinforced, the core monetary attribute pricing that gold has enjoyed over the past few years will be sealed away once again.
Comments