The Second Half of AI Investment: Global Market Rotation and the Monetary Policy Dilemma

In the first half of 2026, global capital markets underwent a profound gear shift beneath a seemingly calm surface. The S&P 500 recorded a 9% gain, but the former "Magnificent Seven" fell 1% overall, with June alone dropping nearly 9%—the worst monthly performance in over a year. At the same time, a broader AI investment cluster—from memory chips to electrical equipment to server makers—surged at an astonishing pace, delivering a 100% return in the first half. This stark contrast not only depicts a dramatic shift in capital flows, but also signals that the AI narrative is moving from a "dream premium" to "delivery pressure."

From "Storytelling" to "Counting Inputs and Outputs"

The market's attitude toward AI is undergoing a fundamental transformation. Over the past two years, investors were willing to pay hefty premiums for large tech companies because the room for growth imagination was large enough. However, when "hyperscalers" such as Alphabet, Amazon, Meta, Microsoft, and Oracle committed to unprecedented levels of capital expenditure, the market began to calmly calculate: how much sustainable return can these investments actually generate? The answer is not clear. Combined with rising costs for memory chips and electrical equipment, Apple and Microsoft have announced price increases on some products, and profit-margin compression signals have further reinforced cautious sentiment.

Meanwhile, the Fed's hawkish pivot in mid-2026 sharply increased the discounting pressure on high-valuation growth stocks. Investor positions were highly concentrated in a few high-market-cap tech names, and this crowded trade itself amplified the scale of the correction. As a result, capital began to exit the "Magnificent Seven" and instead flowed into global companies that are actually building AI's physical foundation—high-bandwidth memory suppliers SK Hynix and Micron, lithography giant ASML, electrical infrastructure provider Schneider Electric, and server makers Cisco and Dell. This industrial cluster, called the "AI complex" by Vanguard, returned 100% in the first half excluding hyperscalers, far outperforming the traditional semiconductor index. It represents a more practical path to profits: regardless of whether AI applications ultimately deliver, the hardware and energy networks required for training and inference must be built first.

This rotation is not a simple style switch. It reflects the market's repricing of AI capital expenditure efficiency. As the market shifts from "imagination" to "discounted cash flow," companies providing physical infrastructure earn a premium because of earnings certainty. Whether high-bandwidth memory or electrical equipment, they directly benefit from the wave of data center construction, and their earnings visibility is far higher than that of large platforms dependent on advertising and cloud subscriptions. This is also why, even during the June tech stock correction, capital did not leave the AI track—it reallocated along the industry chain instead.

The Fed's Wait and Global DivergenceThe AI investment cycle does not operate in isolation. Behind it lies the complex intertwining of the global macroeconomic and policy environment. The U.S. economy has shown resilience, but inflation remains above target, and the labor market has seen intermittent cooling. June employment data was soft, consistent with seasonal expectations of slower summer hiring; the unemployment rate is expected to edge higher over the next few months, but the Vanguard team judges this to be only a "healthy adjustment," projecting that the unemployment rate will stabilize in the mid-4% range and maintain full employment through 2027. Faced with this combination, the Fed's choice is not hard to understand: since price pressures have not yet been reliably eliminated, excessive easing could reignite inflation, while premature tightening would hurt employment—holding steady until the end of 2026 is the most realistic policy path.

From a regional forecast perspective, the backdrop for this policy choice becomes even clearer. U.S. real GDP growth is projected at 2.3% in 2026 and 3.0% in 2027; China is at 4.7% and 4.8%, still an important engine of global growth; the euro area is only 0.8% and 1.3%, Japan 0.8% and 1.2%, and the UK 1.1% and 1.2%. Differences among advanced economies, compounded by divergent inflation paths, have led to a pronounced mismatch in interest rate cycles—the European Central Bank's deposit rate is expected to be cut from 2.5% to 2.0%, the Bank of England from 4.25% to 3.75%, while the Bank of Japan is expected to raise from 1.25% to 1.75%. Capital will accordingly flow from low-return regions to markets with greater return certainty, and the strength divergence between the U.S. dollar and regional currencies will persist. Mexico, meanwhile, keeps its interest rate at 6.5% due to elevated core inflation, highlighting localized pressures in emerging markets.

Short-Term Optimism, Medium-Term Caution

For today's market, short- and medium-term judgments need to be separated. In the short term, the AI investment cycle is still deepening, and oil prices have retreated from the highs reached during the Middle East conflict, which also supports risk sentiment. Therefore, maintaining a constructive view on equities in the short term is reasonable. But the medium-term picture is far more complex. The next phase of the AI story is a test of whether massive investment can translate into productivity growth for the global economy. Historical experience shows that the dividends of general-purpose technologies are not distributed evenly or immediately—they tend to spill over slowly from the sectors that initially innovate to the broader economy. If technology diffusion proves weak, the "optimistic assumptions" embedded in current valuations will face a reassessment.

Accordingly, Vanguard prefers U.S. value stocks and developed market equities outside the U.S. over the medium to long term. Against the backdrop of growth stock valuations approaching historical highs, this tilt carries a distinctly defensive flavor. Notably, in mid-July, the Philadelphia Semiconductor Index and Asian tech stocks experienced a notable pullback, while a Chinese AI model named Moonshot sparked new discussion about the global competitive landscape. The volatility of AI and its infrastructure will not disappear simply because the theme is hot.

ConclusionMidway through 2026, AI investment logic has reached a fork in the road: on one side is physical-layer infrastructure construction with continuously rising capital expenditures; on the other is application-layer promises with increasingly uncertain returns. Capital has made its choice between the two worlds, and this choice will profoundly shape future global capital allocation. Meanwhile, the Federal Reserve's patient wait between inflation and employment continues to shroud the market in monetary policy uncertainty. For investors, understanding the macro logic behind rotation matters more than predicting the next market move. What truly warrants attention is not whether AI is great, but when and how it will translate into sustainable economic returns.