Dear Fellow Investor,
Good morning. Happy Saturday!
I hope you’re getting a little rest this weekend. After this week, we all earned it.
But first, let’s talk about the $18 billion that went missing on Wall Street.
On Monday, the S&P 500 closed above 7,800 for the first time ever.
On Tuesday, it hit another record. So did the Nasdaq.
Then on Wednesday, the Financial Times reported that OpenAI — the company behind ChatGPT — had told its investors something nobody expected.
Its annualized revenue at the end of September was about $50 billion.
Not the $68 billion that had been widely reported just weeks earlier.
By Thursday, the Nasdaq had its worst day since mid-August.
Intel fell more than 5%. Nvidia fell about 3%.
If you’re like most people, you’re probably wondering whether this is the moment the AI bubble finally pops.
Fair question. Let’s dig in.
What Actually Happened
First, let’s be clear about what this was — and what it wasn’t.
OpenAI didn’t lose $18 billion in sales.
According to CNBC, a person familiar with the matter said the bigger number included gross revenue from partners — money flowing through other companies that sell OpenAI’s products.
The $50 billion is closer to what OpenAI actually keeps.
That’s an accounting difference. Not a collapse.
And back in August, Bloomberg reported OpenAI’s run rate had just topped $40 billion.
Going from $40 billion to $50 billion in a couple of months is still remarkable growth.
So why did the market react like the house was on fire?
Because the $68 billion had become the yardstick.
Investors used it to compare OpenAI to its rivals.
They used it to justify the enormous sums being spent on chips, data centers and power.
When the yardstick shrinks, everything measured with it gets measured again.
The Tennis Match
I said this yesterday, and I’ll say it again.
This market is a tennis match between interest rates and earnings.
Right now, both players are hitting hard.
On the rates side, the 10-year Treasury yield touched its highest level since 2002 this week.
On the earnings side, Wall Street expects S&P 500 profits to jump about 30% this quarter, according to LSEG — with technology leading the way.
When rates are this high, investors have very little patience.
They’ll pay up for growth — but only as long as they believe the growth is real.
So the moment one number looks inflated, they sell first and ask questions later.
That’s what Thursday was.
And not everyone thinks it’s over.
Adam Crisafulli of Vital Knowledge said the AI-linked tech stocks are “unlikely to simply stage a sharp, V-shaped rebound.”
He could be right.
Bumpy is normal. Bumpy is not the same as broken.
Earnings season kicks off in earnest next week with the big banks.
That’s when we start to find out whether the profits back up the prices.
Watch the Orders, Not the Headlines
Here’s what I found far more interesting this week.
On Friday, the CEO of a company called Lumentum sat down with Bloomberg in Tokyo.
Lumentum makes optical components — the laser and light parts that move data around inside AI data centers.
Michael Hurlston, the CEO, said his company is “completely sold out” through early 2029.
He said Lumentum can’t meet about 70% of demand for some products through next year.
And here’s the line that jumped out at me.
“We are very, very far behind what customers want.”
Six months ago, he expected to be sold out through 2028.
Now it’s 2029.
Lumentum’s stock jumped on the news.
Think about that for a second.
On Thursday, the market panicked over a revenue estimate for a private company.
On Friday, a supplier said its customers are lined up for years.
One of those is a projection.
The other is an order book.
I’ve been doing this almost 35 years, and I’ve watched this movie too many times.
Headlines move stocks for a day.
Orders move them for years.
Follow the Bottleneck
Now, I’m not telling you the AI trade is risk-free.
It isn’t.
Some of these companies are priced for perfection, and perfection rarely shows up on schedule.
The speculative names — the ones with no consistent earnings — will get hit hardest when the mood sours.
We tend to avoid those anyway.
But the companies that make the things everybody has to buy?
That’s a different story.
When demand runs this far ahead of supply, the bottleneck gets paid first.
And nobody is buying faster than Elon Musk.
His AI company plans to switch on about 420,000 Nvidia GPUs at Colossus, his supercomputer in Memphis, in November.
On Monday, he confirmed early talks for Taiwan Semiconductor to join his Terafab chip venture.
He isn’t waiting on anybody’s revenue estimates.
He’s locking up everything he can get.
And from what I’ve found, the $1.75 trillion Colossus depends on one company Wall Street has overlooked.
I’ve put together the full story on who it is and why it matters so much.
Have a wonderful weekend.
I’ll see you tomorrow.
“The Buck Stops Here,”
Found this helpful? Share it with others.
Written by Dylan Jovine