Oracle: The First Crack in the AI Trade

AI investment isn’t the problem.

The assumptions behind it are.

For the past three years, the entire AI trade has leaned on one simple bet:

OpenAI and Anthropic would eventually generate enough revenue to justify trillions of dollars in infrastructure spending.

That bet is now breaking.

Not because AI stopped growing or improving, but because the economics behind it changed entirely.

In fact:

Most investors still see this as an OpenAI problem only.

It isn’t.

It’s a problem for everyone who built their businesses and growth around OpenAI eventually paying the bill…

This is why you see companies like Google, GE Vernova, Intel, etc.

All beating earnings expectations only to lose double-digit percentage points shortly after.

The reason?

If OpenAI and Anthropic aren’t able to make good on over $5 trillion of commitments, then the players in this value chain won’t see that money back either.

This affects everyone from cloud computing, chipmakers, and infrastructure spenders.

These industries have become dependent on AI models, whose economics have worsened through:

Naturally,

The bond market started asking a different question:

Can these commitments actually be repaid?

That’s when these credit default swaps (CDS) began pricing in a greater chance of debt defaults.

The truth is, Oracle wasn’t downgraded because it suddenly became a bad business.

It was downgraded because a growing share of its future cash flows - and ability to repay mounting debts - now depends on OpenAI’s profitability.

The bond market just voted, and it voted against AI.

That’s why Oracle is the warning.

Oracle: The First Crack in the AI Trade.

Oracle’s contracted obligations, now 5x larger than last year at $650 billion.

Under other circumstances, I would tell you this is massively bullish for a stock that is now trading at 37% of its 52-week highs.

There’s one huge problem with Oracle though…

Over 60% of all contracts are concentrated into one single company.

Meaning,

Close to $400 billion of these backlogs could go up in smoke if the contracted company isn’t able to pay these obligations.

Given Oracle’s market cap is roughly $330 billion, I can say with confidence that the company has bet the ranch on these contracts.

Ordinarily, I would look for the bull case in such an extraordinary setup like this one.

But, there isn’t one.

Oracle has signed up for over $125 billion in debt, while burning through more than $40 billion worth of free cash flow (which is now negative.)

They deployed over $160 billion into AI infrastructure projects, designed to accommodate the capacity needed to fulfill services behind that $650 billion backlog.

In other words,

Oracle has run out of cash, spent 50% of its balance sheet on capacity for one single client, and now their fate is up to that client making good on contracted commitments.

Is there a chance this will turn around?

Considering their debt is now five times larger than earnings, I would say a proper restructuring of the company is the only path to a turnaround.

Possibly a management shift, and bailouts from customers that do rely on Oracle lest they find a replacement soon.

Otherwise this could become the Bear Stearns of the 2008 housing bubble.

It’s not just me saying this by the way, a trillion-dollar market has made it known they’ve lost confidence on Oracle’s ability to dig itself out of this hole.

Again, not without outside intervention and a complete restructuring.

Bond traders have forced Oracle’s debt into junk status.

Credit default swaps (CDS) have spiked to levels beyond the 2008 crisis and COVID put together.

These spreads essentially price in a higher probability of Oracle defaulting on its debts, debts which now represent a larger sum than the company’s entire balance sheet.

Remember Credit Suisse not that long ago? Archegos? This is no different.

The outcome in Oracle is not limited to its balance sheet.

You see, after Oracle lands the commitment deals from OpenAI, and they take on massive amounts of debt to build the capacity needed to fulfill their services…

Someone else receives that capacity investment, and it goes like this:

Which is where the whole circular financing “value” chain connects between several other companies involved in similar products.

Whose bottlenecks and skyrocketing margins have incentivized competitors to join and help ease that bottleneck.

All while falling into a similar situation that got Oracle in trouble in the first place.

That’s why you see announcements like this morning’s:

Upon realization that Oracle won’t get paid, and therefore won’t buy more NVIDIA equipment considering OpenAI’s failure to fulfill its obligations…

NVIDIA has taken it upon itself to inject OpenAI with an additional $500 billion in backstop capital, to keep the money flowing through the scheme again.

Oracle is up over 5% today on the news, while NVIDIA is down.

Confirming my read on market preference going toward companies receiving the spending versus the ones doing the spending.

At this point,

It’s clear the spenders won’t get paid any time soon, and might as well kiss investments like that $500 billion goodbye.

Oracle will probably announce something along the lines of new orders from OpenAI, or the first (of many) payments to come as they fulfill obligations.

However,

The market’s reaction should tell you all you need to know…

Several companies, not just Oracle, have bet a dangerous amount of their balance sheets on the belief that OpenAI and Anthropic will eventually repay them through obligation fulfillment.

Which is why I suspect we’ll begin to see more headlines like these…

Trillions riding behind these AI projects will be valued at much lower levels than they were initially underwritten for.

Why?

Because the one source of customer funding has dried up to the point where backstops need to be made.

And there are only two companies able to enact these backstops:

Anything outside of that and we’re looking at a full out government bailout.

Let’s put it together now:

Whenever you see charts like this one, be very skeptical.

Oracle justified all of the bond issuance and free cash flow destruction on this chart:

However,

As OpenAI misses on 90% of its revenue targets, and new projects carry a negative present value as we saw above…

Nobody is buying the story behind these promises, the math just stopped making sense at this point.

The Oracle story serves as a warning for all else involved, and you should use it as such to spot similarities around the promises being made around massive debt issuance.

The SMH semiconductor ETF has fulfilled on a very typical distribution pattern.

I flagged the fact that hedge funds and prime broker data was selling out of tech stocks at the most aggressive pace since COVID.

Which is why you saw the following:

I’m not ballsy enough to short the companies I know are misleading you, since the hype may burn me (and you) in the end.

What I’ll say is this:

I find it very interesting that all our Offside Portfolio stocks are up over 5% the same day all AI-related names are bleeding.

So I will double-down on my thesis because it is working.

If you haven’t been with me for that long…

Then I will relay you to my mid-caps thesis that’s tied to a long TLT bonds and Chinese tech idea as well.

When one of them pays, they all pay.

Click here to read my report.

Now you have the broader idea, time to get access to the individual trade structures and execution.

The same ones that have generated over two years’ worth of membership costs for members.

See you inside Offside Premium.

In the meantime, here’s a take on AI exposure from Goldman Sachs, showing you how hedge funds really feel about these stocks.

Keep that in mind as you watch the hottest names in the market bleed today:

Until next time,

OFFSIDE RESEARCH

Against the Tape, Ahead of the Curve.