To The Moon

It’s been a wild week for the stock market. The AI trade seems to be taking a breather, if not worse. Investors are starting to worry about growth, margins, and valuations for some of the most talked-about stocks on the market. Even the biggest companies in the world saw their stocks fall 10% earlier this week.

Longtime readers know I’ve been somewhat cautious about the market for most of the past year. I’ve sold part of my position in Robinhood and Alphabet on valuation concerns.

But that caution has also led to the Asymmetric Portfolio underperforming the market over the past year.

The reason was simple. I didn’t get in on the AI trade. I didn’t buy Neoclouds because I don’t think they have a sustainable business model. I’m not buying semiconductors or memory stocks that I think are approaching the peak of their cycle. And I’m not buying hype names as they pop up.

Short-term, that’s to my detriment.

Long-term, I think I’ll avoid being in front of a freight train as we saw on Thursday.

1. A Lot of AI’s Promise Is a Myth…For Now

As investors, we love to tell stories about the future. In fact, it’s often the story that has value more than the fundamentals.

Just look at $Tesla Motors(TSLA)$ , which has a compound annual growth rate slower than inflation over the past three years, but is worth 13x sales because investors buy a story about its future.

The myth has very real value.

Until no one believes the myth.

In the AI world, myths began forming the moment ChatGPT was launched. Chatbots were going to disrupt search, plug-ins (remember those?) were going to replace apps, and agents were going to do our work for us, leaving humans to “beach” all day. (I hope you get that Barbie movie reference)

None of those have really happened as prognosticated, and now very real questions are being asked about the buildout supporting disruption that may not be there.

What’s After Coding?

There’s no question that coding is both an amazing use case for AI and also had an adoption curve faster than any product I’ve ever seen. A year ago, AI coding was useful, but with $Anthropic(ANT)$ ’s improvements late in 2025, it became a requirement by spring 2026. And that coding adoption drove the revenue growth at AI labs.

By late summer 2026, coding had more competition from OpenAI and $Alphabet(GOOGL)$ , companies started to optimize spend, and the growth rate for Anthropic flatlined.

If AI coding has reached saturation, what’s next?

The search for the “next thing” brings hype, but hasn’t brought much real business or disruption.

Agents Aren’t Real…Yet

You’ve heard about AI agents and their amazing future, but have you used them? And have you used them consistently after a trial period?

Most people haven’t. $Uber(UBER)$   confirmed this in their conference call: out of 320 million trips during Halloween week, the company only had about 500 orders from agents.

If their posts were true, a few people on Twitter seem to have accounted for nearly all of those orders.

So, agents have been sold by everyone from OpenAI to $Meta Platforms(META)$ ( ▲ 0.84% ) as the future of AI, but are they?

I happen to think agents will be a valuable tool in the future.

I also happen to think the dispersion of agents will take A LOT longer than tech people think. And that rate of adoption has a big impact on capex ROI.

ROI Is Up In The Air

All of this brings up a timing question.

Spending happens now.

Revenue comes in…sometime in the future.

Right now, it seems like everyone is making money. GPU rental prices are high, growth in phenomenal, and margins are extraordinary.

But what happens when the exponential growth you planned for doesn’t arrive?

In the worst case, you go bankrupt.

Exponential Growth Squared

What happens when you get your exponential growth expectation wrong?

We’re not talking about one company exponentially increasing supply anymore. EVERYONE is spending on data centers, chips, new equipment, etc.

Supply is going to explode over the next two years.

But is demand?

And is that demand willing to pay for AI?

Ask this question at every layer of the stock. Is chip demand going to go up enough to please $NVIDIA(NVDA)$ $Advanced Micro Devices(AMD)$ $Intel(INTC)$, and the hyperscalers who are making their own chips?

Is chip demand going to drive memory demand?

And equipment?

And energy?

Bubbles don’t pop when we see growth slowing. They pop when capex growth flatlines. And even Dan Ives sees the future having less growth than previously expected.

That sound you hear is growth rates slowing, margins compressing, and multiples falling.

Not what we want as investors.

2. Valuations Aren’t Always What They Appear

I’m seeing a lot of “Stock X is cheap because the P/E is under Y” comments right now.

As always, valuation matters, but we need to understand what drives the valuation. I’m going to pick on NVIDIA because it’s one of the biggest companies in the world in this example.

NVIDIA looks cheap because the P/E ratio is just 24x (purple), and the company is extremely profitable and generating strong free cash flow.

The question we need to ask is: Should we value based on today’s operations looking backward or future operations?

And if we’re looking forward, what does the future look like?

Periodically, NVIDIA’s growth rate goes negative. Revenue growth isn’t going to be 80% forever, so when does it slow and maybe even go backward?

NVIDIA has also never had 65% operating margins. Historically, operating margins hover around 15%-20%. And sometimes they go negative too.

If growth stalls and margins go back to historic norms (all possible based on what I outlined above), NVIDIA isn’t cheap at 24x trailing earnings; it’s expensive because it’s trading for 75x future earnings.

We could do the same analysis for $Micron Technology(MU)$ or discuss objectively expensive stocks like AMD.

The point is, in a booming market like AI — which looks a lot like a commodity boom — valuations may not be what they appear. We have to figure out what’s sustainable and what’s not, and right now those questions are up in the air.

3. Balance Sheets Matter Again

Survival will be key if the AI bubble does burst.

Who has the cash to survive?

$CoreWeave, Inc.(CRWV)$ , for example, is expected to spend more on capex than it generates in revenue for the foreseeable future. And it already has $35 billion in debt. If this is a bubble and it pops, there’s trouble ahead.

I’ve focused a lot on companies with businesses I like, but I also spend more time on balance sheets and free cash flow than a lot of investors. Like $Duolingo, Inc.(DUOL)$ , which, for all its faults, isn’t going bankrupt with $1.4 billion in cash and $400 million in free cash flow over the past year.

I could show similar charts for most of the Asymmetric Portfolio. Few companies I own have a real risk of failure in the next few years.

That will be important because before you can thrive, you have to survive if this is a market downturn and investors are finally starting to pay attention to those balance sheets.

4. The Strong Will Survive, And Thrive

I’m sure I’ve sounded doom and gloom today, but I’m not.

I buy stocks once a month, and I’m happy to be buying opportunistically if/when the market falls.

What I’m focused on is the business models that can thrive long-term. Businesses that aren’t built on hype, but rather have strategic advantages like network effects or aggregation that will be durable for years, maybe decades.

But to thrive, these companies need to survive first.

If this is a new, more risk-conscious market, I think we’ll see more focus on the companies with great balance sheets, cash flow, and business models that can survive the upcoming turbulence.

That’s also where the opportunity lies for investors, not in catching falling knives.

And once we survive any upheaval, those left standing will be better positioned to thrive.

I think the Asymmetric Portfolio is set up to not only survive, but thrive, and I’m happy to add to the best companies if prices go even lower.

And with that backdrop, here are four things I think I know about the market today.