The AI Earnings Boom or AI Bubble? Stock Talk Update October 16, 2026

Is Wall Street Counting on Too Much? 

The AI earnings boom is real. But is it also becoming an AI bubble? We have discussed this topic for over 16 months now, almost exclusively in a positive light. Wall Street expects corporate profits to soar. Technology companies are spending hundreds of billions of dollars building the future of artificial intelligence. And investors are buying stocks today for profits they expect tomorrow. 

But there’s a problem. Earnings are growing. Yet for some of the biggest AI investors, free cash flow is under pressure because spending is growing even faster. Are we watching another productivity boom, or repeating one of the biggest investment mistakes of the Dotcom era? 

Now here’s the twist. Even with these risks, our team still believes the S&P 500 could reach 8,000 before the end of 2026, no change in that outlook for over 12 months, and potentially 8,500 during the first quarter of 2027. 

Why? Because earnings growth, election-year patterns and seasonal buying may be lining up at the same time. But that outlook comes with one important condition: nothing major can break in the earnings, credit or interest-rate markets. I’ve spent 35 years managing stock portfolios. I remember the excitement around the internet in 1999. The technology changed the world. But not every technology stock was a good investment. Today, we’ll compare that period with the AI boom, look at the warning signs, and explain why the next few months could still favor investors. 

Let’s start with something that makes the bubble argument harder. Corporate earnings are strong. And the overall S&P500 P/E is under 20x. According to early October estimates, Wall Street expects S&P 500 earnings to grow rapidly in the third quarter. And the Citadel Securities chart shows growth is expected to remain strong into early 2027. 

Source: Charlie Bilello, Creative Planning

Look at the trend. The chart shows reported earnings growth rising from roughly 11% in the second quarter of 2025 to nearly 20% in the first quarter of 2026. For the second quarter of 2026, reported growth is shown above 30%.Analysts expect approximately 27% growth in the third quarter, followed by roughly 25% in the fourth quarter and 18% in the first quarter of 2027. Those later figures are estimates, not reported results. 

That’s a powerful earnings backdrop. It also explains why I believe third-quarter earnings and 2027 guidance may matter more than the next Federal Reserve meeting. But strong earnings today don’t guarantee strong stock returns tomorrow. A company can grow profits by 25% and still see its stock fall if investors expected 35%. The question isn’t simply whether profits are growing. It’s whether companies can keep beating expectations—and whether those profits eventually turn into cash. 

Take yourself back to late 1999. The internet was changing how people shopped, worked and communicated. Companies were building fiber-optic networks, buying servers and spending enormous amounts of money on technology. Investors believed this spending would lead to years of strong growth. They were right about the technology. But many were too early about the profits. 

Some companies built more capacity than customers needed. Others borrowed heavily to finance growth that didn’t arrive fast enough.And this wasn’t happening in a world of falling interest rates. In 1999, bond yields rose across the United States and Europe. Yet technology stocks continued to climb into early 2000 because investors believed future earnings growth would more than offset the rising cost of money. 

Sound familiar? Today, AI companies are building data centers instead of fiber-optic networks. They’re buying advanced chips instead of the servers of 1999. And they’re betting future demand will justify today’s spending.There’s a major difference: today’s leading technology companies already have huge, profitable businesses. Microsoft, Alphabet, Amazon and Meta have established customers, strong balance sheets and major sources of cash. 

That makes this cycle different from many speculative internet companies of 1999. But it doesn’t remove the risk of spending too much. How much AI demand is real, and how much is still being built on expectations? 

There’s no question people and businesses are using more AI. One way to measure that growth is token consumption. A token is a small piece of information processed by an AI model.Rising token use tells us demand for computing is growing. It doesn’t tell us how much profit companies are making. 

Suppose an AI company doubles the number of tokens it processes. But what if the price it charges per token falls by half? Revenue might not grow at all. And what if the company had to spend billions on new computers and electricity to handle that demand? Its costs could rise faster than sales.We need to separate three things: usage, revenue and profit. They are not the same. 

Here’s the key distinction: earnings per share can rise even while free cash flow weakens. Accounting earnings spread the cost of expensive equipment over several years. Cash flow reflects the money actually being spent today.If a company buys billions of dollars of chips, servers and buildings, much of that cash leaves immediately. But the cost doesn’t necessarily hit reported earnings all at once. 

Goldman Sachs estimates that major U.S. hyperscalers would need about $1.42 trillion in combined revenue over 2028 through 2030 attributable to the 2026–2027 buildout to clear its illustrative 15% return hurdle. That is a model, not a guaranteed forecast. The demand may be real. The spending may be justified. But the timing of the cash returns is still uncertain. That’s where the earnings test begins. 

So is this an AI earnings boom or an AI bubble? I believe we need to watch three numbers. Number one: earnings revisions. Are analysts raising their forecasts for 2027, or lowering them? Great third-quarter results won’t help much if companies then cut next year’s guidance. 

Number two: free cash flow. If earnings rise 25% but free cash flow falls, is the company making a smart investment that will pay off later? Or spending more and more just to maintain its position? 

Number three: capital spending compared with customer demand. Are customers paying more for AI services? Are businesses using AI to lower costs? Are the companies selling AI tools seeing stronger margins? The next phase of the AI boom needs more than new data centers. It needs economic returns. But here’s the surprise: even if some AI stocks disappoint, the broader stock market could continue higher. And the next two quarters have a historical pattern worth watching. 

Look at this chart from RBC Wealth Management. 

The red line represents the second year of the presidential cycle. That’s 2026. Historically, this has often been the most difficult year of the cycle, with a correction during the middle of the year followed by a recovery into year-end. But the most interesting part is what happens next. The green line represents year three: 2027. Historically, the third year has often been one of the strongest. Averages don’t guarantee anything, but they show why I wouldn’t automatically assume recent gains must end with the calendar year. 

According to the Citadel chart, the S&P 500 has historically gained an average of about 5.6% in the fourth quarter of midterm years, versus about 2.9% for fourth quarters across all years. That’s nearly twice the average gain. 

The monthly study also highlights October and November as historically favorable months in midterm years. The averages are around 2% to 3% per month for the S&P 500, depending on the month and sample.None of this means stocks have to rise. But when you combine favorable seasonality with strong earnings growth, improving investor positioning and potential corporate buybacks, you have a meaningful tailwind. 

Seasonality describes what investors have tended to do. Earnings describe what businesses are producing. Buybacks are one source of demand for shares. When all three align, markets can advance even when the news isn’t perfect. 

Now let’s put numbers around that outlook. 

Using an illustrative starting level of 7,812, reaching 8,000 before year-end would require a gain of about 2.4%. Reaching 8,500 during the first quarter of 2027 would require a gain of roughly 8.8% from that same level. Neither target requires the kind of explosive advance we saw in the final months of the Dotcom bubble. 

We still believe 8,000 before year-end is achievable, and 8,500 during the first quarter is a legitimate bullish scenario—if earnings keep growing, the market broadens and interest rates don’t cause a major valuation reset. But something cannot break. We cannot have a major disappointment in AI spending returns, a serious credit event that threatens financing, or another sharp rise in real interest rates that overwhelms earnings growth.That’s not a guarantee. It’s a conditional outlook built on earnings, market history and the level of the index. 

In 1999, the internet changed the world. But that didn’t save investors who paid too much for companies that couldn’t deliver profits.Today, AI may also change the world. But a great technology doesn’t automatically make every stock a great investment. 

And the reverse is true. A disappointment in a few expensive AI stocks doesn’t necessarily mean the entire market has to fall.Banks, industrial companies, healthcare, consumer businesses and other sectors can benefit from economic growth, stronger productivity and broader earnings gains. 

That’s why market breadth matters. A bull market supported by more industries is generally healthier than one carried by a handful of companies.For retirees, you don’t need every AI stock to succeed. You need a portfolio that can participate in growth without depending on one investment theme. 

So let’s return to our opening question. Is this an AI earnings boom or an AI bubble?The earnings boom is real. The spending risk is real. And the next few quarters will help determine whether the profits justify the investment. 

But we also have strong expected earnings growth, a historically favorable fourth-quarter election pattern and the potential for market leadership to broaden.My working outlook: 8,000 on the S&P 500 before year-end is achievable, with 8,500 a legitimate first-quarter 2027 target—provided earnings hold up and nothing important breaks. 

Watch earnings revisions. Watch free cash flow. Watch interest rates and credit markets. And watch whether more stocks begin to participate.Because the next leg of this bull market may depend on something bigger than AI spending alone. It may depend on whether the earnings boom spreads beyond AI.