Brace for Impact (OpenAI)

OpenAI, the company behind ChatGPT, is running into trouble.

Its growth has slowed compared to Anthropic, huge losses are piling up, and over a dozen senior leaders have headed for the exits this year.

This is all happening ahead of OpenAI’s highly anticipated IPO, expected sometime before the end of next year.

Ordinarily, financial woes at a private tech company aren’t something you would need to worry about. But this is different.

Much of the AI trade rests on the idea that companies like OpenAI will continue spending tens of billions of dollars on chips, data centers, and power.

And all that money flows to companies that many investors currently own.

If OpenAI fails — or even just stumbles — it’ll have massive ripple effects across the stock market.

The Alarm Bells Are Getting Hard to Ignore

OpenAI generated $6.7 billion in revenue during the second quarter, up 18% from $5.7 billion in Q1.

That's impressive growth for almost any company… except this one.

During the same quarter, rival Anthropic's revenue more than doubled to $11.6 billion. OpenAI's operating loss, meanwhile, widened from $9.3 billion to an eye-popping $12.3 billion.

Think about those numbers as if OpenAI were already public.

Revenue grew 18%, but its closest rival grew more than 100%. And OpenAI lost almost $2 at the operating level for every $1 in revenue it generated.

How would Wall Street react to that earnings report? Probably not well.

That's important because OpenAI is preparing for an IPO that could come in 2027 and Anthropic is also preparing to go public around the same time.

So the two may soon compete for investor dollars as well as customers.

Then there are the questions about OpenAI's management team.

Recent departures include Chief Revenue Officer Denise Dresser, longtime COO Brad Lightcap and ethics chief Chloé Bakalar.

The Financial Times has also reported on repeated shake-ups, internal politics and worker burnout.

Dresser's departure stands out.

She became OpenAI's chief revenue officer less than a year ago, then left as the company's push into business customers became more important.

One or two senior departures at a fast-growing company isn’t abnormal by any means. But a dozen before one of history's most anticipated IPOs is suspect.

Investors don't just buy financial results. They also need to trust the people running the company.

And OpenAI is rebuilding key parts of its leadership team just as it prepares to face those investors.

The AI Money Machine

OpenAI's success has created an odd problem.

The bigger it gets, the more computing power it needs. That means more data centers, chips, and electricity. And all of that requires money.

OpenAI is expected to burn roughly $25 billion in cash this year. Its planned compute spending through 2030 has reportedly climbed to around $750 billion.

That money spreads across an enormous network.

OpenAI has major deals with Microsoft, Oracle, Amazon, Nvidia, CoreWeave and others.

Microsoft alone reported $24.1 billion in fiscal 2026 revenue from its relationship with OpenAI. Oracle reportedly has a roughly $300 billion cloud agreement with OpenAI. And CoreWeave has signed multibillion-dollar contracts with it.

In other words, OpenAI has become a central part of the AI money machine. Which brings me to one question…

Where is all that money going to come from?

OpenAI recently raised an amazing amount of private capital. There’s no evidence it will run out of cash tomorrow.

But given its losses and spending plans, I believe OpenAI will need even more capital before its IPO.

Just look at the numbers, then consider that OpenAI isn't the only one asking investors for money.

SpaceX raised $75 billion in June in the largest IPO ever. A month later, SK Hynix raised another $26.5 billion in its massive U.S. share sale.

Together, the two deals pulled in more than $100 billion in about a month.

And let’s not forget that OpenAI rival Anthropic is preparing an IPO of its own.

At the same time, other AI-related companies are also turning to debt markets. Oracle and Alphabet have raised billions.

And just last Thursday, it was reported that Broadcom filed paperwork that could allow it to raise more than $100 billion through debt and other securities.

That’s on top of a similar $39 billion financing deal Broadcom already completed in June.

Then, everyone has to compete with the biggest borrower of them all: Uncle Sam.

U.S. government debt has topped $40 trillion, which means Washington also needs a steady supply of buyers for bonds.

There’s plenty of money in the world, but not an endless amount available at any price.

SpaceX wants it. Anthropic wants it. OpenAI wants it. The companies building AI data centers want it. And the U.S. government needs it.

When everyone wants money at once, the cost of that money can rise. For OpenAI, that matters.

This is where the story comes back to you and me.

What Happens If OpenAI Stumbles?

Imagine OpenAI's growth keeps falling behind Anthropic. Its losses stay high, or investors won't finance the next $50 billion or $100 billion on good terms.

OpenAI doesn't have to disappear. It could cut spending, delay data centers, or change future plans.

But that would quickly affect partner companies that count on its spending, such as Oracle, CoreWeave, Microsoft, and Nvidia.

Then come memory chips, networking gear, power companies, utilities, construction firms, and data centers.

Markets don't wait for revenue to vanish. Stocks are priced on what investors expect companies to earn tomorrow.

If expected AI spending falls, stock prices across that chain could fall with it.

Suddenly, investors aren't just questioning OpenAI. They're questioning the entire AI spending boom.

We've seen versions of this before.

During the dot-com boom, venture money spread into advertising, telecom equipment, real estate, hiring, and other parts of the economy.

When that money stopped flowing, the damage spread far beyond failed dot-com companies.

That’s why the problems over at OpenAI are such a big deal. It sits at the center of the AI boom, and trouble there could quickly spread to companies you own.

It only needs to stumble badly enough for Wall Street to start questioning the hundreds of billions being spent on the idea that this boom will continue.

If that happens, the problem won't belong to OpenAI anymore — it will belong to all of us.


Dave’s last word: If you’d like access to Enrique’s insights on a regular basis, you should sign up for his e-letter Truth & Trends

Along with colleagues Greg Guenthner and Nick Riso, Enrique unpacks everything from technical indicators and strategies… to the tools that legendary investors use to build their empires.

Just go to the Truth & Trends website and sign up on the right side of the homepage to get daily delivery to your inbox.

As for the markets today… the major U.S. indexes are treading water ahead of Nvidia’s earnings release after the closing bell. The S&P 500 is just barely in the green at 7,682.

Treasury yields are inching back up, the 10-year note back to 4.66%. Gold is down but continues to hold the line on $4,600. Bitcoin is consolidating its recent gains, a little over $78,000.

U.S. oil futures are back over $81 in the absence of ridiculous market-manipulating headlines like the one yesterday about Washington offering sanctions relief to Iran if the Strait of Hormuz is reopened. 

The big economic number of the day is “core PCE” — the Federal Reserve’s preferred measure of inflation. At 3.3% it rings in unchanged from the previous month. 

You will notice this number is nowhere near the Fed’s 2% inflation target. 

Indeed this number has been higher than 2% every month since early 2021 — a phenomenon Fed chair Kevin Warsh will have to address one way or another during his inaugural “Jackson Hole speech” this coming Friday. 

More about Warsh’s conundrum when we get back to our usual multitopic format tomorrow…

FMF-Issue-082526-Featured

The Totally Fake Inflation Numbers

Going by the government’s official numbers, the inflation rate is 3.4%. But if the numbers were run the way the government ran them in 1980, it would be 11.4%.

FMF-Issue-082426-Featured

AI’s “Really Weird Math”

We pick up today where we left off with last Thursday’s Bullet No. 1 — the probability that ChatGPT maker OpenAI will blow up before it ever goes public.

FMF-Issue-082126-Featured

“Inflicting Maximum Damage and Pain”

This week, The New York Times spotlighted the power grid’s vulnerability to sabotage. Seeing as we’ve been on the case for over a decade, we have to ask: Why now?

FMF-Issue-082026-Featured

ChatGPF (The “F” Is for “Fiasco”)

“There is a very real, non-zero possibility that OpenAI runs out of money before they can go public,” says Paradigm trading pro Enrique Abeyta

FMF-Issue-081926-Featured

The Position of **** You

John Goodman’s financial advice in The Gambler has been undone by inflation.

FMF-Issue-081826-Featured

Gold and Great Nations

In August 1971, President Richard Nixon cut the dollar’s last tie to gold — and the dollar back then is worth only 12 cents now.

FMF-Issue-081426-Featured

Now She Tells Us (AOC)

History is clear: Economic hardship leads to societal chaos. There would have been no “woke” revolution without COVID lockdowns. And AOC just said so.

FMF-Issue-081326-Featured (1)

Investing Lessons from the Poker Table

Even if you’ve never played poker and never will, the game holds lessons that can pump up your portfolio.

FMF-issue-081226-Featured

The AI Bailout of 2027

Slowly, inexorably, the path is becoming clear: The AI industry will be asking for a taxpayer bailout next year. Or, at the latest, 2028.

FMF-Issue-081126-Featured

Turning Japanese

There’s a situation shaping up in markets that Paradigm macroeconomics authority Jim Rickards says could turn out to be worse than the 2008 financial crisis.