Why 20% Earnings Growth Suddenly Isn’t Good Enough

  • AI stocks are falling despite strong earnings growth because Wall Street is lowering the valuations it’s willing to pay for future profits.
  • AI earnings expectations remain stronger than the broader market, suggesting the recent selloff may be a valuation reset rather than the end of the AI boom.
  • Microsoft, Broadcom, and Alphabet offer three ways to evaluate AI investment opportunities, with profitability, cash flow, and capital spending now critical to stock selection.
AI stocks - Why 20% Earnings Growth Suddenly Isn’t Good Enough

Listen to the audio version of this article (generated by AI).

Here’s a quick test. 

A business you own was forecast to grow profits 21% next year. The forecast just got cut – to 20%.

Do you sell?

Of course not. You’d ask what changed and move on with your day. But Wall Street just answered the same question differently – and some investors are selling the company. 

According to Bloomberg Intelligence, expected S&P 500 earnings growth for 2027 has slipped from about 21.2% to 20.1%, down from the roughly 30% pace expected for 2026.

Twenty percent profit growth still sounds great. So why are stocks acting like somebody canceled the future?

Because investors already paid for a certain future. And now the market is deciding whether that future needs a small adjustment – or a much bigger rewrite.

Why AI Stocks Can Fall Even When Earnings Grow

We’d bet on the small adjustment. But to see why the market’s reaction has been so much bigger than the revision itself, you have to look at the math investors are actually doing.

Let’s put the forecasts into dollars and call this year’s earnings $100. The old forecast said $121 next year; the new one says $120. That’s a 1% haircut to expected profits – almost a negligible amount.

The adjustment was never the problem. The problem is the multiple.

Say a stock earns $5 per share and trades at 30 times earnings – a $150 stock. Next year, earnings jump 20% to $6, but nervous investors decide the business only deserves a multiple of 25.

Run the new math: $6 times 25 is… $150. Right back where it started.

The company grew profits 20%. The shareholder went nowhere.

That’s the whole AI selloff in one paragraph. Earnings haven’t broken – not even close. What’s broken, temporarily, is the market’s willingness to pay yesterday’s price for tomorrow’s growth. We’ve watched it happen in real time for a month: Taiwan Semi, Meta, and Lam Research all posted strong reports and sold off anyway. Our Peak Watch Checklist calls that signal “good news gets sold,” and it’s been flashing since October.

Here’s the flip side, though. Multiple compression is a one-time event. Once the market stops paying 30x and starts paying 20x, the repricing is done, and returns reconnect to earnings growth. The S&P 500’s tech sector now trades near 20 times forward earnings, close to its cheapest valuation since ChatGPT launched. 

A lot of this adjustment is behind us.

AI Earnings Forecasts Reveal a Growing Divide

The selloff headlines all quote that topline revision. Dig one layer into the Bloomberg Intelligence data, though, and the story flips.

While the broader 2027 forecast softened, information technology’s profit estimates actually improved. The same data points to the Magnificent Seven reclaiming earnings leadership in 2027 as growth slows everywhere else.

In other words, analysts are cutting numbers for the broad economy and raising them for the AI complex – the exact opposite of what this selloff implies. 

This is the AI Bifurcation. When growth gets scarce, reliable growth gets more valuable, and the companies with real AI earnings pull away from everything that was merely along for the ride.

Of course, a better forecast for the tech sector doesn’t bless every chipmaker, software company, and data-center operator inside it. The sector numbers tell us where to hunt. Individual results tell us what to own.

Is the AI Spending Boom Starting to Slow?

Which brings us to the other number making the rounds: SemiAnalysis has trimmed its longer-term AI spending outlook.

Don’t reflexively explain this one away – but don’t panic over it either, because why a spending forecast falls matters more than the fact that it falls. There are three very different possibilities:

  • Fewer machines. That directly pressures everyone who expected to sell them.
  • Lower prices. Customers get more computing power per dollar while suppliers face thinner revenue. Bad for the sellers, great for the buyers.
  • Later deliveries. A project delayed by power shortages, permits, or financing pushes equipment sales into a different year. The revenue moves; it doesn’t disappear.

Our own work says the slowdown is mostly the third flavor, with some of the second mixed in. We’ve estimated for weeks that new U.S. compute capacity comes in a few percent below plan in 2027 and about 15% to 18% below in 2028, with most delayed capacity shifting into 2029 – slower, not smaller. 

The real casualties are the business models that need expensive debt and years of spending before customers start paying. The AI opportunity can stay enormous while those particular models get much harder to finance. That’s a reason for selectivity, and it’s exactly why we keep sorting this market by balance sheet.

What Nvidia Needs to Show Us Tonight

All of this gets its biggest test today, Nov. 17. Nvidia (NVDA) reports earnings at about 4:20 p.m. Eastern, with the call at 5 – the company’s first report since Jensen Huang’s resignation, with CFO Colette Kress running the show as interim CEO.

We’ll be listening for more than a strong quarter. A company can post excellent sales from orders placed months ago while its next round of business gets shakier. 

The questions that matter most here: Are customers changing order volumes? Are delivery schedules slipping? Is pricing holding? And what does management see beyond the next few months – especially on the Megaspeed and Singapore revenue questions hanging over the stock?

One report won’t settle the outlook for the entire AI trade. But it should tell us whether the market’s worry has reached actual orders, prices, and delivery schedules – or whether it’s still living entirely in the forecasts.

3 AI Stocks That Could Outlast the Selloff

So how do we put all this to work? We start with businesses already converting AI demand into real earnings and cash flow; the ones that don’t need every rosy assumption to pan out to justify their price. Here are three names through that lens, and the question we’re asking of each.

Microsoft (MSFT): Is Growing Adoption Producing Enough Profit? Microsoft Cloud revenue hit $63.9 billion last quarter, up 30%, with companywide operating income up 21%. That’s real, compounding AI monetization. The question from here is whether customer growth keeps pace with the infrastructure bill – so watch cash generation after capex and whether customers keep paying up for more services. A massive backlog helps, but those contracts still have to turn into profits.

Broadcom (AVGO): Follow the Cash Alongside the Chip Orders. Broadcom posted $16.7 billion in AI semiconductor revenue last quarter while generating about $13.7 billion in free cash flow. That combination – hypergrowth and a cash machine – is rare, and it’s why we consider Broadcom a structural winner of the custom-chip shift. It isn’t immune to customer budgets, so watch program timing and customer concentration. But the cash gives you something concrete to weigh against the multiple.

Alphabet (GOOGL): Strong Demand Still Comes With a Huge Bill. Alphabet is both sides of this argument in one ticker. Google Cloud produced $11.1 billion in operating income last quarter – yet companywide free cash flow was negative $9.1 billion as capex swallowed everything operations generated and then some. The cloud business is wildly profitable; funding the buildout is wildly expensive. The question is how fast those investments start throwing off cash, and how much you’re paying today for that eventual payoff.

The Bottom Line: What the AI Stock Selloff Means for Investors Now

Twenty percent earnings growth can make you a lot of money. It can also disappoint someone who paid for 30%.

That tension is the entire market right now. Expectations have been marked down, the multiple has already compressed hard, and the earnings themselves keep coming in fine – which is why we think this looks like a reset within the boom, not yet the end of it.

But the reset rewards a different kind of stock than the melt-up did. Before buying any dip, ask the simple questions: If growth comes in lower, does this company still mint money? Can it fund its own plans without begging the bond market? Does today’s price leave room to be a little wrong?

The next great AI investment should still work when you turn the forecast down. The good news is that after this selloff, more stocks pass that test every week.

If you want a systematic way to run that test, my friend and colleague Louis Navellier built one. His Stock Grader system has spent nearly 50 years’ worth of refinement grading stocks on the numbers that survive a slowdown – and last Thursday, he revealed a brand-new way he’s putting it to work for 2027, including two free stock recommendations: one to buy, and one popular name he’d avoid.

These days, the stocks that fail the test matter just as much as the ones that pass it.

Watch Louis’ presentation – for free – right here.


Article printed from InvestorPlace Media, https://investorplace.com/hypergrowthinvesting/2026/11/why-20-earnings-growth-suddenly-isnt-good-enough/.

©2026 InvestorPlace Media, LLC