The dominance of AI themes over stock market performance, corporate profitability, economic growth, and investor attention has become so extreme that some very big structural questions appear urgent and imperative. Goldman Sachs entered this week with a comprehensive analysis of this AI advantage across several sectors. Commenting on the S&P 500’s narrow gains so far this year, the firm found that “TMT names within the S&P (which includes Amazon and Tesla, in addition to both the tech sector and communications services) have accounted for 87% of the year-to-date rally.” The same market segment that generated 87% of the S&P’s gains now accounts for about 54% of the S&P 500’s weight. Framed this way, the market is not “ignoring” or “ignoring” potential dangers such as conflict in the Middle East, rising oil prices, or rising bond yields. The part of the market where this matters — the median consumer cyclical stock, which is down 14% from its all-time high — is noticing such pressures. Wall Street bulls argue that this divergence is logical and healthy because it reflects an evolving earnings outlook. That’s true, but so far most of the expected earnings upside has come from companies whose stock prices are on the rise. Here, Goldman plots 2027 earnings revision trends for AI infrastructure businesses, energy companies, the entire S&P 500, and the rest of the S&P outside of AI and energy. In the last category, next year’s profit forecasts have not been raised at all since the start of 2026. Even if fundamentals point to a bumper crop of profits, stock prices can overshoot. Hedge funds collectively manage 20% of their net market exposure in semiconductor stocks, TD Cowen noted on Monday. The strong long-short basket (a strategy that holds the most persistent outperformers and shorts the deepest laggards) has only rallied more than 20% 11 times in the three months to date. The current move is well above the average for these spikes, suggesting there is plenty of room for a rebound before even the underlying trend is threatened. This Goldman chart ranks today’s episode compared to the average of previous extreme gains. Besides whether it is right for stocks to take advantage of such large expected profits, there is also the question of whether this entails dangerous imbalances in the real economy. AI capital spending next year is expected to reach $1 trillion, or about 3% of U.S. GDP. While the impact of trade accounting on reported GDP may not apply exactly to that number (much of the AI hardware is imported and is counted in domestic production), it is a useful scale. During the 19th century, investment in railroads peaked at 5% to 6% by most estimates, creating a bubble. Nvidia Warning The magnitude of this commitment is forcing even those betting big on a perpetual AI capex cycle to examine some of their underlying assumptions. John Belton, a portfolio manager at Gabelli Funds, detailed a potential red flag for NVIDIA, the largest holding in the growth fund he oversees, and said in a report on Wednesday after the bell: “About half of NVIDIA’s business comes from about five companies. Therefore, the negative case for the stock is essentially that for half of NVIDIA’s business to continue to grow, those customers would have to become negative free cash flow or become free cash flow. You need to make sure that your cash flow generation accelerates. I think this is a more complicated situation for NVIDIA at this point, and one of the reasons why the stock has lagged other semiconductor stocks recently. ” Is the market’s recent preference for lower value-added groups in the tech industry (memory manufacturers, Intel, CPU providers) a sustainable change, or does it simply reflect the temporary ability of these companies to take advantage of the shortfall and tax infrastructure builders? The voracious demand for capital among builders of AI computing power also threatens to strain investors’ ability to make it affordable. While the big tech platform companies aren’t having trouble finding buyers for their debt, issuers are increasingly spreading out around the world to issue debt. Interest rates around the world are rising towards the upper end of multi-year ranges as inflationary pressures mount due to global wars, large budget deficits and shortages of several manufactured goods. I’m not a big believer in “bond vigilantes,” or investors who collectively punish government profligacy by raising interest rates. But the bond market can act as the neighbor who calls the police if the party gets too rowdy. Police may tell revelers to turn down the music a little, or they may break up the party. Shadow Supply Next, there is the shadow supply of stocks from large IPOs that are expected to come to market. If SpaceX’s market value really reaches the rumored $1.75 trillion, and the companies behind it, OpenAI and Anthropic, each have at or near $1 trillion, that would represent about 7% of the S&P 500’s total value. It’s not like they’re going to be in the index right away or at that level. The index weights companies only by the value of their publicly traded stocks, so it may start out small. But if such a deal were to materialize, it would certainly make the mega-cap segment of the market even more volatile and fundamentally more expensive. The market is happy to ride the trend of valuing capital investment over consumers, tacitly supporting the idea that companies can continue to build new capabilities and earn higher profits while making profits. This is despite corporate profits as a share of GDP already reaching record levels. These big-picture questions are hovering over the market, which is also grappling with more mundane short-term issues. Will the reopening of the Strait of Hormuz prompt the pendulum to swing away from technology and back to the rest of the tape, and what does that mean for large-cap benchmarks? Will rising yields expose deeper vulnerabilities or will rising inflation expectations cause a ripple in equity valuations? And will the extreme technical divergence and multiple overbought conditions improve without much pain, as suggested by Monday’s mild rotational action in the face of significant declines in semis and other momentum stocks?
