
AI Gets a Performance Review
A couple of months ago, I wrote about how AI had become my newest employee.
It helps edit this newsletter, catches some of my spelling mistakes, works nights and weekends and, so far, hasn’t asked for a raise.
Pretty good hire. A five-star employee so far.
Apparently Wall Street has decided it’s time for AI’s performance review too.
For the past few years, investors have been willing to focus largely on the promise: smarter software, better productivity, enormous new markets and potentially transformational changes across the economy.
Now the questions are getting a little tougher.
How much is all of this going to cost?
And, more importantly:
When does all of that spending start making money?
Those questions helped make October a bumpier month for markets, particularly for technology stocks.
And I think they’re exactly the right questions to be asking.
The October Scoreboard

October was a mixed month.
The S&P 500 fell 0.9%, while the technology-heavy Nasdaq dropped 5.3%. The Dow went the other direction, gaining 1.7%.
Despite the monthly declines, all three remained positive for the year, with the S&P 500 up 10.7%, the Nasdaq up 9.4% and the Dow up 7.7% through October.
Bonds had a much better month.
The 10-year Treasury yield fell back below 5%, finishing October at 4.93%, while the Bloomberg U.S. Aggregate Bond Index gained 2.4%.
Oil moved sharply higher, with Brent crude finishing around $118 per barrel, while volatility picked up from September’s very calm levels.
In other words, October gave us a little bit of everything.
Stocks down.
Bonds up.
Oil up.
Volatility up.
And AI suddenly getting asked to explain the expense report.
The AI Buildout Is Getting Really Big
There’s a tendency to think about AI as software.
Chatbots. Coding assistants. Search tools. Image generators.
But underneath all of that software is an enormous amount of very physical infrastructure.
Data centers.
Semiconductors.
Servers.
Power generation.
Transmission lines.
Cooling systems.
Construction.
Land.
And a truly staggering amount of money.
Bank of America estimates that AI-related capital spending has increased by roughly 1.2% of U.S. GDP since the beginning of 2020.
That means this investment surge has already surpassed the comparable stages of the 1970s oil boom and the 2010s shale boom.
Only the 1990s technology buildout was larger at a similar stage, at roughly 1.5% of GDP.
So when people say the AI boom is big, they’re not exaggerating.
It is historically big.
But that doesn’t automatically make it a bubble.
And it certainly doesn’t automatically make every AI-related investment a good one.
The Easy Part May Be Over
The first phase of a new technology is often about possibility.
Can we build it?
Does it work?
Will people use it?
AI seems to have answered those questions pretty convincingly.
The next phase is harder:
Can we make enough money from it to justify everything we’re spending?
Bank of America’s research makes that distinction nicely. The AI infrastructure buildout still appears to have room to continue, but investors are becoming more selective about which projects get funded.
Instead of simply rewarding anything with “AI” attached to it, capital markets are increasingly looking at project economics, cash flows, access to power, construction risks and credible paths to profitability.
That sounds pretty healthy to me.
Early enthusiasm eventually has to become underwriting.
At some point somebody has to ask:
Does this thing actually make money?
Yes, We’re Talking About Earnings Again

You knew where this was going.
I made it several paragraphs this time.
Progress.
While investors are debating whether future AI spending will pay off, corporate America is already producing some impressive profits today.
With roughly one-quarter of S&P 500 companies having reported third-quarter results, FactSet estimates earnings are growing 29.5% year over year.
If that holds, it would mark the third consecutive quarter of earnings growth above 25% and the eighth straight quarter of double-digit earnings growth.
Ten of the 11 major sectors are reporting earnings growth.
Revenue growth is running at 12.7% and all 11 sectors are reporting higher revenues than a year ago.
And profit margins remain exceptionally strong.
The S&P 500’s blended third-quarter net profit margin currently stands at 14.9%. If that holds, it would be the second-highest margin since FactSet began tracking the data in 2009.
Broken record?
Absolutely.
But companies making more money continues to be a pretty useful thing for investors.
Prices Matter. Earnings Matter More.
Here’s another interesting wrinkle.
Despite stock prices being near historically high levels, FactSet currently puts the S&P 500’s forward price-to-earnings ratio at 18.7.
That’s actually below both its five-year average of 19.8 and its 10-year average of 19.1.
That doesn’t mean stocks are “cheap.”
It does illustrate something important, though.
Valuation isn’t just about what happens to the P.
It also depends on what happens to the E.
If stock prices rise while earnings rise even faster, valuations can actually improve.
That’s why staring at an index level by itself doesn’t tell us much about whether the market is expensive.
Once again:
Earnings.
I warned you.
A Great Technology Is Not Necessarily a Great Investment
This may be the investing lesson I find most interesting in the whole AI discussion.
A technology can completely change the world.
A company can grow incredibly fast.
And its stock can still turn out to be a mediocre investment if investors paid too much for those future expectations.
Those are three different questions.
Sam Ro recently used Anthropic as an example.
The AI company behind Claude has filed to go public, with reported expectations for a valuation somewhere between $1.5 trillion and $2 trillion.
Its first-half 2026 revenue reportedly reached $16.3 billion, up from $1.2 billion the prior year.
That is extraordinary growth.
It is also still an unprofitable company.
Sound familiar?
We talked about the same issue when SpaceX went public earlier this year.
The important question isn’t whether a company is impressive.
It’s what assumptions are already embedded in the price.
An enormous valuation tells you investors expect enormous future results.
Sometimes those expectations will be exceeded.
Sometimes they won’t.
And predicting which outcome occurs ahead of time is considerably harder than explaining it afterward.
IPO Excitement Versus Investment Results
There’s another useful reminder here.
Research from University of Florida professor Jay Ritter, cited by the aforementioned Sam Ro, shows that the average IPO has historically enjoyed roughly a 19% first-day pop.
That sounds great.
Over the following three years, though, the average IPO has underperformed the broader market by roughly 21%.
That doesn’t tell us what Anthropic, SpaceX or the next giant IPO will do.
It tells us something more useful:
Excitement and investment returns are not the same thing.
Price discovery can be messy.
Expectations can get carried away.
Fantastic businesses can disappoint investors.
And occasionally businesses that look ridiculously expensive can keep getting more expensive for much longer than skeptics expect.
The market gets to decide.
Meanwhile, the Economy Keeps Moving

For all the focus on AI, the broader economy remained pretty resilient.
Real GDP grew at a 3.4% annualized rate during the third quarter, the strongest quarterly growth in about a year.
Consumer spending increased 2.5% and business investment rose 3.3%.
The Fed, meanwhile, left rates unchanged at 3.75% to 4.00% in October.
But it wasn’t exactly an all-clear signal.
Two Fed officials preferred another quarter-point increase and inflation remains stubbornly above the Fed’s target. Core PCE inflation was running at 3.1% year over year in September.
So once again, several things can be true at the same time.
AI spending can create legitimate concerns.
Corporate profits can remain very strong.
Interest rates can stay high.
The economy can keep growing.
Technology stocks can have a rough month.
And long-term investors can still be doing just fine.
Funny how often markets work that way.
From Promise to Proof
For the last few years, AI has mostly been a story about what might be possible.
We’re entering the phase where investors are going to demand more evidence.
Show us the productivity.
Show us the revenue.
Show us the cash flow.
And eventually:
Show us the profits.
That doesn’t make me less optimistic about AI.
If anything, I think it makes the story more interesting.
Because the next stage won’t simply be about who can spend the most money or build the biggest data center.
It will be about who can turn all of that investment into something customers actually want and businesses can profitably sell.
That’s ultimately how technological revolutions become economic revolutions.
And for investors, it brings us back to a familiar place.
The headlines will change.
The technology will change.
The companies leading the market will change.
But eventually, somebody has to make money.
Broken record indeed.
Until next time, take good care!


