An Unprecedented Spike in Earnings Estimates
At the start of the year, Wall Street was expecting roughly 16% earnings growth. Today that figure sits closer to 25% — an extraordinary upward revision. The question is what structurally changed during first-quarter reporting to produce such an unprecedented spike in forward estimates.
The short answer is that the story is "brought to you by the letters A and I." Artificial intelligence was the dominant driver. There was a mass underestimation of just how parabolic earnings growth would become. But it wasn't AI alone — there was also a decent amount of breadth to the earnings improvement. Most sectors, not just the AI-linked names, saw their results improve.
Part of the estimate gap also came from analyst psychology. Analysts had a heightened sense of caution heading into the reporting period, in part because the third month of the quarter coincided with the first month of the war. In that kind of uncertain environment, analysts were simply not inclined to raise the estimate bar. Then earnings season began, and the reality of the results forced revisions higher. On the chart, the first quarter produced a line that went parabolic — an "upside-down hockey stick" trajectory — and that surge dragged up estimates for the remaining three quarters of the year as well as the full-year calendar 2026 numbers. This earnings strength has been the single most important underlying fundamental support for the bull market.
Is the Pace Sustainable, or Have Analysts Overcorrected?
A natural worry is whether that rate of change is sustainable into the second half, or whether analysts overcorrected and set the market up for disappointment. There is indeed some risk of extrapolation — of assuming the current pace of growth will simply persist. That caution is a core message worth imparting.
Crucially, though, the situation differs fundamentally from the late 1990s. Back then, the bubble was in the P — valuations were a bubble precisely because there was no E in many of the leadership companies; they lacked real earnings. In the current case, there is an extraordinary amount of E in the P/E equation. As a result, multiples have actually come down even as prices rose, because earnings have grown so fast.
The real concern is different: the expectations bar has climbed so high that the margin of error has narrowed relative to the past. When expectations are this elevated, even a marginal miss — a one-off company stumble or a sector-specific problem — can trigger outsized reactions. That creates the potential for volatility beneath the surface, driven not by weak fundamentals but by the sheer height of the expectations bar.
Inflation, Rates, and the Fed
Inflation and interest rates loom as a genuine threat to these elevated estimates. The 10-year Treasury yield carries the highest correlation to equity market performance, and it has calmed down somewhat recently, which has been supportive of the market — and of the strong small-cap rally in particular. A chart of the relationship shows the inverse correlation between the 10-year yield and the S&P 500 sitting in steeply negative territory. That relationship has bounced around, but it may represent a secular change: a regime in which inflation volatility is a higher concern. When the 10-year yield keys off inflation rather than off growth, that tends to be detrimental to equities if inflation and yields are rising together.
There are some near-term benefits on the inflation front, thanks to favorable base effects and oil prices coming down. But the inflation problem is broader than just oil. AI itself is a component of the inflation story, and there is still tariff pass-through working through the system. For those reasons, we are not out of the woods yet, and there may still be upside surprises in inflation. The base assumption is that the Fed is in a position where it may have to consider rate hikes this year.
Markets are also much more sensitive to Treasury yields now than they were a year ago. That sensitivity ebbs and flows, but a forthcoming report — co-authored with Kevin Gordon — will take a more secular look at the relationship between the bond market and the stock market, specifically bond yields relative to stock prices. The context matters: the "great moderation" era ran for about 25 years up until roughly 2022, and during that stretch bond yields and stock prices moved in the same direction. In the 30 years prior to that, they did not. The coming analysis will place the recent data in that much longer historical frame.
Can Earnings Carry the Load if the Fed Tightens?
If the Fed is forced back to an explicit tightening bias, the question is whether earnings can keep carrying the heavy load for the S&P 500. There is not much direct interest-rate sensitivity where the earnings trajectory has been concentrated — in the broader AI space, those companies are relatively insulated. The trouble arises further down the market-cap spectrum: smaller companies with less in earnings, or no earnings at all, that struggle to make interest payments. So we should expect some differentiation at the cap level depending on what rates do.
The most important variable, however, is not simply that the Fed moves toward tighter policy — it is the speed of that tightening. If the Fed, once it shifts to a tighter stance, can take a "slow escalator up" and raise rates gradually, that is a far better backdrop for equities than if it has to take a "fast elevator up." History also offers a pattern in market leadership: in the lead-in to the first rate hike, cyclicals tend to do well, which is exactly what we are seeing now. But once the Fed actually starts hiking, the focus tends to shift toward defensive sectors. That rotation is a useful lens for thinking about market dynamics as a tighter Fed approaches.
The AI CapEx Question and the "Great Chip Dip"
AI capital expenditure is a huge driver right now. Amazon, Microsoft, Alphabet, and Meta each grew their CapEx by roughly 70% year-over-year in 2025. The question is whether the market is effectively handing these management teams a blank check for AI infrastructure, or whether investor patience is beginning to wear thin.
Concerns are indeed starting to creep in, and they help explain what can be called the "great chip dip" — a recent pullback in chip-related names. The worry stems from a confluence of factors. One is that the amount of capital consumed from a capital-spending standpoint is significantly higher than the profit contribution — the share of net income these efforts generate relative to the S&P 500. That gap between spending and profit is something to be mindful of. On top of that, an increasing number of corporate enterprises are saying they may have to start considering paring back some of the spend, because the concrete productivity benefits of the products — and, for the capital-spending giants, the return on invested capital — are not yet fully visible. This is not anything resembling a "look out below" moment, but the concerns are creeping in, and there is now a more discerning eye on these data points.
On whether CapEx will slow if monetization doesn't soon materialize: expect to hear more than just anecdotal stories of companies weighing the pros and cons of continued massive AI spending. Even with the price of tokens coming down, the total spend remains significant. One interesting reframing comes from conversations with corporate leaders, who are beginning to ask at what point the cost of human capital becomes relatively inexpensive by comparison. Given the concerns about AI's disruption to the labor force, there may be a silver lining there — human labor starting to look cheaper against the cost of AI infrastructure.
There is also a margin story hiding in the numbers. Consider the extremely high margins reported — for example, Micron's results, with margins in the 75–80% range. The benefit to the company is obvious, but somebody is eating the cost of that margin. That dynamic is entering the narrative, and it is why analysts — especially once earnings season arrives — are going to look much more closely at the profit-margin story, the productivity story, and the return-on-invested-capital story.
The Consumer and Household Equity Exposure
Turning from AI to the humans on the other side of it — the consumer — one striking data point is that household equity ownership has reached more than 45%, nearly tripling from the lows seen during the 2008 financial crisis. That raises two questions: does this make the market more vulnerable to corrections, and are households now systemically overexposed to an equity correction?
This is a chicken-and-egg story that works both ways. The rise in household exposure to equities, alongside a very healthy bull market since the 2022 bear market, has supported the economy through the wealth effect. Where it could shift is if a genuine corrective phase in equities — something deeper than the recent dips — feeds back into economic activity.
That is exactly what happened in the transition from the late 1990s into the early 2000s. The bear market began in March 2000, and by 2001 the economy was in recession. That recession likely came about because of the bear market in stocks: there were no major economic dislocations, it was not a financial-system problem, and it was not the product of massively tighter monetary policy. It was the wealth effect operating in reverse once stocks entered a bear market.
So this exposure has to be thought about in multiple directions — the chicken-and-egg dynamic. At this stage, elevated household equity ownership has been a tailwind, and the base case is that it remains a tailwind. But it is something to watch: the key open question is whether economic weakness, if it shows up first, works its way into a metric like household equity exposure, or whether a market correction works the other way and drags the economy down.
