Earnings Estimate Revisions Are Very Optimistic
Authored by Lance Roberts via RealInvestmentAdvice.com,
đź’° Earnings Estimate Revisions Are Very Optimistic
Last week, we discussed the S&P earnings record and why such record earnings could be a warning for the market. I want to continue that discussion by focusing not only on what has happened but also on what is expected to happen in the future. While the Q1 2026 earnings results are spectacular, so far, the earnings estimate revisions behind them are the real story.
The first-quarter 2026 earnings season is delivering results that Wall Street rarely sees. With roughly two-thirds of the S&P 500 having reported, the blended growth rate has climbed to 27.1% year-over-year, more than double the 13.2% that consensus modeled at the end of the quarter on March 31. If that figure holds, it will be the strongest year-over-year print since the post-COVID rebound quarter of Q4 2021. 84% of companies have beaten EPS, 81% have beaten revenue, and the average earnings surprise sits at 20.7%, nearly three times the 5-year average of 7.3%.
That’s the surface story. The more interesting question, and the one investors should be asking, is why analysts were so wrong heading in, and what it means that they’re now revising earnings estimates higher with a velocity that has almost no historical parallel.
Look at Morgan Stanley’s chart of consensus 2026 earnings estimate revisions versus history. In any normal year, by the time Q1 earnings season rolls around, analysts have been quietly walking earnings estimates down for six months. The historical median revision pattern drifts from 1.00 in January to roughly 0.92 by year-end. Two years of cuts. That’s the analyst playbook. Start the year too optimistic, get reset by reality, and end the year right.
This year is doing the opposite. The 2026 earnings estimate index cratered to 0.96 last summer during the Iran shock, then turned vertical. By May, it’s broken above 1.06. We’re looking at a roughly 14-point swing in earnings estimates relative to the historical pattern. That is what Morgan Stanley calls “fairly unprecedented,” and that’s analyst-speak for something they don’t have a clean comparison for.
The Mag 7 alone moved from a 22.4% expected growth rate at the end of March to a 61% blended print today. Four of the top five contributors to S&P 500 earnings growth this quarter are Alphabet, NVIDIA, Amazon, and Meta. The same four names driving index returns are now driving the earnings estimate revisions. That’s not a coincidence, and there is more to this story as noted by Sage Road Research:
“The AI distortion goes beyond stock prices to profits. Total S&P 500 earnings are on track to rocket 27% higher in the first quarter, FactSet estimates. But profits for the Mag-7 alone will be up 61%; for the other 493, just 16%, a figure itself inflated by semiconductor companies like Micron. This is skewing the division of the economic pie between capital and labor. As profits gallop ahead, labor compensation (wages and benefits) grew just 3.1% annualized in the first quarter, and actually shrank 0.5% after inflation, the Labor Department reported Thursday. Labor’s share of total business-sector output fell to 54.1%, the lowest since records began in 1947.” – @TrevorNoren
So, if it isn’t consumers’ and subsequently economic growth, driving earnings estimate revisions, then what is?
What’s Actually Driving the Upside
Three things are happening at once, and we have to separate them.
First, the AI capex cycle is finally showing up in the income statement. Hyperscalers have spent the better part of two years building out compute. The revenue is now landing. Communication Services is reporting +53% earnings growth, Tech is at +50%, and Consumer Discretionary is at +39%. Those aren’t soft beats. They’re the result of capex that was already locked in before the quarter started.
Second, margins are at a record. The blended Q1 net profit margin came in at 14.7%, the highest reading in over 15 years. That’s the real engine behind the surprise factor. Revenues grew 11.1%, which is solid but not extraordinary. The gap between 11% revenue growth and 27% earnings growth is operating leverage. A company that cut 8% of its workforce in 2023 and held headcount flat through 2025 is now monetizing every dollar of incremental revenue at a much higher incremental margin.
Third, breadth in Q1 results is finally improving. The Deutsche Bank charts make this point clearly. Earnings growth for the median S&P 500 company is now in the double digits, the highest reading in four years. All eleven sectors are tracking positive growth for the first time since 2022. Margins for “the rest” of the index, the 493 names outside the Mag 7, are turning higher after a steady three-year decline. Operating cash flow for non-financial corporates is running near 20% year-over-year. Q1 results are genuinely broader than they have been in years, and that deserves credit. But there’s an important asterisk on this point that I’ll address in the next section.
The Asterisk on “Broadening”
Now we have to separate two things that get confused in the headlines. Q1 reported earnings broadened, but the forward-year earnings estimate revisions did not. Those are different statements about different time horizons, and the difference matters.
Goldman Sachs published a chart in early May that quantifies the gap. The bank tracks a basket of AI infrastructure stocks (S&P 500 constituents in their AI Semiconductors, AI Data Centers, and Power Up America baskets) and compares it to the broader index on cumulative 2026 EPS revisions since December 2024. The numbers are striking.
AI infrastructure stocks have seen 2026 earnings estimates revised higher by 55% since December 2024. The full S&P 500 is up 7%. The S&P 500 ex-AI infrastructure is down 1%. Read that last figure twice. Strip out chip designers, hyperscaler infrastructure, AI data centers, and the power and grid names that feed them, and the remaining 470-odd companies in the index have collectively had their 2026 earnings estimates revised lower over the past 17 months. Not flat. Lower.
This is the cleanest picture of concentration risk you’ll see this cycle. The narrow-market critique, which has been valid for 2 years, isn’t going away, even after Q1 results came in. It’s hiding inside the index math. Mega-cap AI names have absolute earnings dollars so dominant that even modest forward growth in their numbers swamps the rest of the 500. When analysts publish their 2026 earnings estimate consensus forecast for the index, they’re effectively publishing a forecast for roughly 30 companies. The other 470 are a rounding error to the headline.
“Strip AI infrastructure out of the index, and 2026 estimates are actually lower than they were 17 months ago.”
The implication for portfolio construction is direct. If you own a market-cap-weighted S&P 500 index fund, you don’t own the diversified earnings stream the marketing material implies. You own a concentrated AI infrastructure bet wrapped in a passive vehicle. The two largest holdings in the SPY are Nvidia and Microsoft. Apple, Amazon, Meta, Alphabet, and Broadcom round out the top eight. Seven of the top eight names are direct AI infrastructure plays. That’s not diversification. That’s a thematic fund with 490 other names attached for legal reasons.
None of this is bearish on AI itself. The capex cycle is real, the earnings growth is landing, and the demand picture remains durable. The point is more subtle. The index’s strength masks the weakness of its components. If AI infrastructure names hit a single quarter of disappointment, whether from capex digestion, an export control surprise, or simple revenue deceleration, there’s no second engine in the index to absorb the impact. Equal-weighted measures of breadth being healthy on Q1 results don’t fix the forward-revision concentration problem. They are two different problems.
Here is What Nobody Wants to Talk About
Here’s where I have to put the brakes on. When earnings estimates are revised this hard, this fast, you have to ask whether the market is pricing the beat or the trend. Because historically, vertical earnings estimate revisions are a late-cycle phenomenon, not an early one.
Notice the long-term S&P 500 earnings growth estimate chart. The current reading sits near 19%, the highest print since 2000. The chart’s prior peaks tell a story. The “New Economy” peak in 2000. The “Tax Cuts” peak in 2018. The “COVID” rebound peak in 2021. Every one of those readings was followed by a meaningful drawdown in equities and a sharp downward revision cycle in earnings within twelve to twenty-four months. Forecasts above the long-term trend channel have a poor history.
“When everybody is revising higher, the marginal trade is no longer to buy the beats. It’s to fade the next miss.”
The other tell is the divergence between hard data and earnings. ISM Manufacturing is sitting in the low 50s, barely above the contraction line. The S&P 500 is up roughly 19% year over year. That gap historically closes one of two ways. Either ISM rallies into the high 50s as the cycle accelerates, or earnings get marked down to meet the macro. The latter has happened more often than the former at this point in a cycle.
This Summer is Where Headwinds Rise
There’s a calendar problem stacking up behind these numbers. The Q1 print benefited from easy year-over-year comparisons. Q2 won’t have that tailwind. By the time July prints arrive, the comparison base resets to 2025’s stronger second-quarter results, which means the same level of underlying earnings translates into a much smaller growth rate. That mechanical effect alone could pull the headline growth rate from 27% back into the low double digits, even if absolute earnings keep climbing. Markets don’t always distinguish between “growth slowing” and “earnings missing.” They tend to react to the headline number first and sort it out later.
Then there’s the bond market setup. The 10-year is still trading near 4.4%, the front end is pricing barely two cuts for the rest of the year, and core inflation has been sticky in the high 2s for six months. If the AI capex cycle keeps running hot, that’s incremental demand for chips, electricity, and skilled labor, all of which feed into the inputs the Fed watches. The risk isn’t a recession scare. It’s a “no cuts, maybe a hike” repricing that historically chops 5% to 8% off equity valuations in short order.
Positioning is the other variable. Sentiment surveys are stretched. Equity allocations among retail and institutional investors are at multi-year highs. CTAs are max long. When everyone is on the same side of a trade, and the data starts to disappoint, price discovery is brutal because there are no marginal buyers left to absorb the unwind.
Institutions Are On Risk Watch
The most useful way to gauge the risk landscape is to look at what institutional trading desks are actually doing, not just what they’re saying. The substance of the conversations across the buy-side and the dealer community is converging on a single posture: stay long, but explicitly hedge. The same desks publishing constructive twelve-month equity targets are simultaneously paying for downside protection in size. That’s the tell.
Five points are worth laying out.
First, positioning. The Nasdaq 100 just delivered its biggest monthly gain in over 23 years. A move of that magnitude has consequences for who is left to buy. Systematic strategies (CTAs, vol-target funds, risk parity) have completed their re-risking. The buy-the-dip retail bid has been engaged since the March lows. Discretionary trading desks are now running long exposure at around +6 on a -10 to +10 scale, up from -4 at the March lows. The marginal buyer in this tape has already shown up. From here, the question is who steps in if the data disappoints.
Second, the fundamental catalyst stack is largely behind us. The fiscal pulse that supported corporate margins is fading. The Q1 EPS print of 27% will not repeat against tougher comparisons. The operating leverage that drove the surprise factor cannot keep expanding indefinitely. The combination means the next four quarters of earnings reports face a higher bar with less wind at their backs.
Third, the Strait of Hormuz is still live. The tape has effectively forgotten last summer’s oil shock. That’s how markets work. We discount tail risk after the immediate catalyst passes, but the underlying geopolitical setup has not materially improved. A single headline can reprice oil 4% in a session, and equities are positioned for a benign energy backdrop that may not hold.
Fourth, the Fed is constrained. Last week’s hawkish hold told us where the committee sits when core inflation prints closer to 3% than 2%. The base case for cuts in September and December assumes labor market softening that has not yet arrived. If those cuts get pushed, the equity multiple has to absorb the disappointment, and historically that costs the index 5% to 8% in short order.
Fifth, narrow breadth is a real risk that history takes seriously. Most standard measures of S&P breadth are exceptionally thin right now. Nine of eleven sectors are positive on the year, which sounds healthy on the surface, but participation under the index headline is concentrated in a handful of mega-cap names. The strongest historical conclusion isn’t that narrow breadth is bearish (because, for two years, it hasn’t been), but that it raises the probability of a momentum rollover when the rotation eventually breaks. You don’t pick that fight. You do prepare for it.
Here’s the practical math that ties this back to portfolio action. One-month at-the-money puts on the S&P 500 are currently priced at less than 2% of the spot price. For investors carrying meaningful long exposure into a summer with the stack of risks described above, that’s compelling risk transfer. The same institutional desks publishing constructive twelve-month equity views are paying for that protection right now. They call it “the cost of a good night’s sleep.” That phrase belongs in every portfolio review this quarter.
🔑 Key Catalysts Next Week
After two weeks of Magnificent 7 earnings and payrolls data, the calendar pivots back to the macro gauntlet that will define the Fed’s June path. Tuesday’s April CPI, Wednesday’s April PPI, and Thursday’s April Retail Sales create a three-day inflation-consumer trifecta that will either confirm or break the “higher for longer” trade heading into the June 16 FOMC meeting. This week isn’t about individual stocks; it’s about the price level and the consumer’s willingness to pay it.
Tuesday’s April CPI is the week’s anchor. March ran hot with the headline at +0.4% MoM and the core reading rising +0.3%. That reflected the first full month of the Iranian oil shock and the broadened tariff regime. April is the second month of that regime, and the question is whether the acceleration was a one-month spike or the beginning of a new trend. Energy prices eased modestly in late April as Iran ceasefire talks gained traction, potentially providing a one-month offset. But core goods, where tariff passthrough lives, won’t have that benefit. Used car prices, which had been masking tariff pressure in prior months, are no longer declining. Shelter costs remain stubbornly elevated. If the headline comes in above +0.3% MoM or core reaccelerates, summer rate-cut expectations are dead.
Wednesday’s PPI doubles down. Producer prices feed directly into the PCE calculation that the Fed actually targets. March PPI printed a blistering +0.7%, the hottest monthly reading in over a year. April PPI tells us whether the upstream pipeline is still pressurized or whether oil’s modest pullback and easing supply chains provided relief. A PPI-to-CPI passthrough story is forming: if producers are absorbing cost increases now, margins will compress, and earnings will be revised down. If they’re passing them through, consumer inflation stays elevated, and the Fed stays on hold. Either outcome is negative for someone.
On the earnings side, this is the bridge week between Big Tech and the Nvidia event on May 20th. Cisco, on Wednesday after the close, is the enterprise IT capex bellwether. AI-driven switching demand, progress on Splunk integration, and the order backlog will tell us whether corporate technology spending is holding up or pulling back amid macro uncertainty. Alibaba Wednesday morning is the China read on cloud and AI revenue from the Qwen model, quick commerce investment, and tariff/trade war impact on cross-border commerce. Applied Materials on Thursday after the close is the semiconductor capital equipment signal ahead of Nvidia, its $5 billion EPIC platform bet, and wafer fab equipment orders are the leading indicator for chip manufacturing capacity expansion.
CPI will tell us where inflation is, while PPI tells us where it’s going. Retail Sales on Thursday will tell us whether the consumer breaks before the Fed blinks. Three data points, three days, one narrative. If all three run hot, the “higher for longer” trade hardens into “higher for the foreseeable future,” and risk assets may begin to reprice. This is why we continue to suggest maintaining portfolio management practices carefully.
What Should Investors Do Now
Q1 was a genuinely strong quarter. Margins are real. Cash flow is real. The broadening is real. None of that is in dispute. What’s worth disputing is the assumption baked into consensus. An 18.6% full-year forecast assumes the run rate from Q1 just delivered continues for three more quarters, with no margin compression, no demand weakness, and no AI capex digestion. That’s a stack of optimistic assumptions, and the historical record on stacks like that is unkind.
For investors, the playbook into summer is unchanged in direction but tighter in execution. Trim into strength rather than chase. Reduce concentration in the names that have done the most work, especially where position sizes have crept up from price appreciation rather than active accumulation. Add hedges, not insurance you’ll never use, but actual collars or put spreads on the largest exposures. Keep dry powder for the first material disappointment, because it always comes, and the names worth owning rarely go on sale during euphoria.
The setup that worries me isn’t that earnings are bad. It’s that they’re so good the bar has been raised to a level that historically marks a peak, not a launching pad. When everybody is revising higher, the marginal trade is no longer to buy the beats. It’s to fade the next miss. That moment usually arrives without warning, and the pattern has held in every prior cycle that produced a chart like the one in front of us today.
Stay long, but stay hedged. The asymmetry has shifted.