We spent a lot of time discussing the structural shortage of compute (through an investor’s lens, not a technological lens) on the recent TCAF episode we did with Ankur Crawford. She explained how, at this point, there simply isn’t enough token capacity for all of the AI workloads, present or future. This lack of capacity has effectively been the only restraint on AI revenue growth for the OpenAIs and Anthropics and Geminis of the world. At least, this is now what their leaders are effectively saying.

Dario Amodei, at a fireside chat last week, had jaws dropping around the world as he casually mentioned Anthropic is now on a $30 billion annual revenue run rate as of the current quarter (we wouldn’t know otherwise, it’s a private company) and that $50 billion is in sight. This is up from approximately zero dollars just a couple of years ago. It’s up from an annual run rate of $9 billion as of the end of last year.

As Dario explained, the company was planning for a 10x growth rate and instead the reality is looking more like 80x. These are growth rates that traditional securities analysts simply cannot process. It’s literally stunning and no one knows what to even say in response to this.

But they do know what to do. Upon hearing this little tidbit, the Nasdaq and its chip stocks took off to the upside. The buying seemed limitless, with gigantic stocks making 7%, 12% and 15% daily moves like they were microcaps and not worth trillions already.

And as bullish as we all are about what growth rates like these will mean for revenue and earnings growth for the stocks associated with the AI capex theme, surely we’ve got to be hitting some sort of a limit - at least in the short-term. I have a few charts to show you below that I believe represent unsustainable things…

Above, you’re looking at the net income expectations for Nvidia (I am long) and in the right pane you’re seeing the stock market’s investors refuse to re-rate the company’s valuation. Nvidia sells at a 24 PE despite a growth rate significantly higher than any company at its size has ever experienced. This is because Nvidia actually trading at a multiple that befits its growth rate would make its market cap larger than almost every other stock in the market combined. The sheer size of this company has now effectively pinned the potential multiple it will sell at to a number far below what the company probably “deserves” if we’re being consistent. It’s absolutely bizarre to see the largest companies in the stock market also have the highest growth rates and best profitability, but that’s the modern Nasdaq in a nutshell. We’re living through things that shouldn’t be possible, and yet, here we are.

Kai Wu (of Sparkline Capital) was one of our guests on The Compound and Friends this week. He brought a bunch of great charts including this one - another aspect of the current moment that seems unsustainable. This is the percentage of market cap for the whole S&P 500 currently residing with the top seven index constituents. We have literally left earth and are orbiting in space on this metric alone, as you can see. The betting man would say “surely this must reverse at some point” but the problem is he may have been saying that most of the last three years and he’d have been wrong. During the 2022 tech crash, this did temporarily reverse as Meta and Amazon fell into 50% drawdowns butyou can see on the chart how big of a head-fake that was.

Above I’m showing you the share of S&P 500 overall profits currently captured within the top ten component stocks. We’re at 34% and that number is rising. This is 50% higher than a decade ago and a full double from 1996, before the internet. This concentration of profits among the largest companies is obviously (I can say “obviously” at this point) among the great secular trends of our era. And while the pace of this might be unsustainable, directionally - I would guess - it’s unlikely to reverse. The rich get richer, the big get bigger, something pretty major would have to come along to put a stop to it.

The furious rally in semiconductors is not happening in a vacuum. In the chart above you’re looking at the profit outlook for the stocks that comprise the SOX index of chip companies. As you can see, it’s an upside shock for the ages, with the earnings expectations for next year rising by nearly 100% over the past 12 months. The rate of this upside shock is obviously unsustainable. Eventually, the analysts catch up with their forecasts. Will 2028 be as big of a surprise as 2027 is? Doubtful.

In the final chart of this run I am showing you the semiconductor index market cap as a percent of the S&P 500 via our friends at Bespoke. Can this go to 30%? 40%? It could. Is it likely to do so at the current rate of speed? Semis were 5% of the stock market as recently as 2019 and now they’re almost a quarter of it. A lot of this is Nvidia and Broadcom but then there’s also KLA and LRCX and AMD and MU and MRVL and QCOM and TXN and INTC and AMAT and ASML and TSM and they’re all just going and going. At this pace, they’ll be half the stock market by this time next year and, well, I don’t f***ing think so.

These charts are all totally unsustainable - and yet completely understandable given what’s been going on. We’ll have to check back in on some of this stuff later in the year to see whether or not this was a moment in time for us all to take a deep breath and check ourselves. It’s not a given as extreme things can always become more extreme.

In any case, we’re witnessing (and making money from) one of the greatest bull markets in history. Let’s not gloss over that fact. I was on the floor of the NYSE this week as Cerebras (CBRS) began trading as the largest IPO since Snowflake (SNOW) back in the fall of 2020. One of its lead investors and board directors, Brad Gerstner, told me it’s just the beginning for what the company does - chips for AI inferencing. He’d been invested in the company for nine years before it’s debut on the Nasdaq on Thursday.

When he wrote his first check, it was still five years before the world would be introduced to ChatGPT and the first LLMs. As recently as a year and a half ago, CBRS had to pull its S-1 due to lack of institutional investor interest in owning a semi with a concentrated customer list. The orders for the IPO this week were 20-to-1 buyers versus sellers.

It’s insane to think about how quickly attitudes have changed.

For a wide-ranging discussion on the charts above and some of the biggest events of the week, you’re going to want to check out this week’s all new episode of The Compound and Friends.

Ben Carlson, Director of Institutional Asset Management here at Ritholtz Wealth, and Kai Wu, founder and CIO of Sparkline Capital sat in for the conversation and we covered a lot of ground: The Nvidia situation. What to make of the Anthropic valuation. Whether software disruption is still in its early innings or already priced into everything. The acceleration of market cycles and what it means for how long any thesis actually has to play out. And Kai's framework for intangible assets.

We also got into Ben's new book, Risk and Reward, which he says is the best thing he’s ever written (imagine how high a bar that is!).

You can watch and listen at the links below!

THE COMPOUND & FRIENDS

Michael Batnick⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Downtown Josh Brown⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ are joined by Ben Carlson and Kai Wu⁠ to discuss: Nvidia, Anthropic, software disruption, intangible assets, faster market cycles, and Ben’s new book Risk and Reward and much more!