Markets Expect Too Much Out of AI

What if I told you today’s market is not your traditional “bubble”?

Bubbles typically form when valuation multiples climb to record-high levels, levels which create underlying expectations that become near impossible to meet.

Valuations across the AI trade do not justify calling the space a bubble, but everything about it feels like one.

So where’s the disconnect?

It’s in the earnings side of the equation, not the price. This is a bold call, but I’m not the only one showing you this view exists, it’s just a lot harder to understand since it takes some digging into accounting standards and financial statements.

Instead of trying to figure it all out on your own, and crash coursing through corporate finance in your free time, I have a better solution for you.

This Thursday, we will begin a multi-post deep dive into the economics of the AI trade:

— Where the earnings “fraud” bears may be right'
— Why valuations could collapse (and bring the S&P much lower)
— What we need to see in order to avoid this dreadful outcome

And most importantly, what the best way to invest through this is.

Come take a look at our coverage, our portfolio, and our Offside strategy around one of the biggest opportunities of our lifetime.

Become an Offside Premium member now.

I am not here to tell you that NVIDIA is going bust,

Nor will I say that Michael Burry is right about his claims on the company committing fraud.

Here’s what I want you to ask yourself:

— What must be true for NVIDIA’s price (a $5T valuation) to be justified?

Most finance professionals will answer this through the traditional discounted cash flow model.

But,

That model is only half the picture, and it serves Wall Street analysts more than it does individual investors like you and me.

The reason is that, as Buffett says, the model relies heavily on the user’s forecasting assumptions.

Therefore, the final valuation tells you more about the analyst than it does about the company itself.

So I will propose a different way of thinking, founded in expectations.

The market is a lot smarter than you and I can ever be, so let’s just assume that the market (in its infinite wisdom) is right about NVIDIA’s $212 share price as of this writing.

We begin by reverse-engineering this DCF model, using Excel’s “Goal Seek” function to solve for what must be true.

For the next decade, what must be true for NVIDIA to justify it’s $212 share price goes as such:

— A revenue compounded growth rate of 47.6%
— 70% EBIT margins all throughout the decade
— A 12% discount rate assuming no disruption to its market leadership position

In the world of expectations investing, these are very aggressive assumptions to say the least.

Were this a $50-$100 billion company, I would be more inclined to believe these numbers. But, at $5T, it just seems borderline impossible to get done.

Bulls will say these numbers are more than achievable, since compute demand (due to agentic AI) will more than 100x in the coming years.

That may or may not be true, but I would not recommend exposing your financial future to an assumption of that nature, which is highly uncertain.

In fact, this is where I may start to think Burry is right in his “fraud” claims.

NVIDIA, as of the latest quarterly release, chose to change the way it reports its revenue segmentation.

— Data center revenues now include both compute and networking

This is also the first quarter NVIDIA reported compute revenues BELOW expectations as highlighted above.

With a sudden shift like that, any shrewd investor would dig further into this situation, and that’s exactly what we did at Offside.

What we found should act as a warning for most NVIDIA investors, and all those with a retirement account.

As delusional as he may be at times, Burry might be right again this time.

We believe NVIDIA is cooking the books through legal accounting loopholes.

But,

Just as mortgage variable rates had an expiration date in 2007, so do these loopholes.

And we are closer to that expiration date than you think.

Join our Premium Research, we’ll break it all down to the last cent in there.

Just as expectations are placing an incredible amount of pressure on NVIDIA and other AI companies to perform beyond their current capabilities.

There is another sector of the market that demonstrates the complete opposite dynamic.

Actually, there are a few:

— Software
— Consumer discretionary
— To some extent, housing

Warren Buffett’s Berkshire Hathaway is set to buy Taylor Morrison in a $6.8 billion transaction, sending the stock higher over the weekend on the news.

Other players, which will be discussed inside our Offside Premium publication, have been pouring into some of these discretionary names as well.

For now, the trade remains attached to the massive capex projections in the AI ecosystem, though some risk management measures are advised.

Whether it is placing tighter momentum signals, options, or a volatility-adjusted stop loss.

You simply cannot expect these stocks to keep pumping by 30% or more on any given day, no matter what the underlying fundamentals may signal.

At some point, it all gets baked into the price.

We have reason to believe NVIDIA has priced it all in, others (like Micron and AMD) are only catching up now.

That gives you a couple of options as discussed above:

— Rotate into overly pessimistic implications in cheap sectors
— Continue to play the capex story around AI, though with tighter risk management

We have made our selection inside our portfolio, which you can get a front seat to inside our Offside Premium publication.

This Thursday, we begin our multi-post deep dive into the AI ecosystem, helping you recognize whether these companies are truly committing fraud or are really in as healthy an environment as they claim.

Here are some thoughts to take home from Goldman Sachs as they attempt to answer the question about the overextended tech rally:

Until next time,

OFFSIDE RESEARCH

Against the Tape, Ahead of the Curve.