An Exclusive AI Party

As the bubble gets bigger, the speculators are getting more selective.  In 2020 and 2021, the government handed out trillions over 18 months and the result was a giant speculative frenzy.  During the Covid money spew, SPACs were popping up left and right. Most of them acquired sub-par companies and took a big slice off the top, making the after market SPAC buyers absorb all the risk.  Crypto was going crazy, as were meme stocks, bought up by a bunch of herd chasing, inexperienced investors.  Stocks like GME and AMC were the poster childs of that bubble.  You had NFTs, which was a ridiculous idea in the first place, going for huge amounts.  Sports cards, Pokemon cards were rocketing higher.  That all ended with the bond market rout in 2022.  

Then the second wave of this never-ending party started in late 2023 with AI.  The first beneficiaries of this AI boom was NVDA and the AI power infrastructure plays.  They were the most popular names in 2024 and 2025.  This speculative wave got another boost that was non AI related, with Trump getting elected in 2024.  Crypto pumpers and HODLers got all excited about a "crypto-friendly" government.  I remember hearing about how stablecoins were going to go mainstream, about how crypto would allow financial transactions to be tokenized, and you could buy parts of a share of stock, or even a house.  The idea is ridiculous, but speculators were coming up with any reason to buy bitcoin, ether, and alt coins.  The real reason they were buying was because it was going up, and they thought it would keep going up.  

In 2025, everyone was participating in the party.  At the core, you had AI and AI adjacent names going straight up.  Crypto was going crazy with bitcoin/ether treasury companies raising tens of billions to buy more coins.  People were talking about alt season.  Super speculative concept stocks in quantum, space, nuclear were up over 1000% in less than a year.  The speculation was broad, indiscriminate, and caused massive short squeezes.  That phase of the bubble topped out last October.  

2026 is a different animal.  On the surface, it may seem like more of the same animal spirits as 2025, with the Nasdaq going up almost every day, and SPX following along.  But this market is much narrower than 2025.  The breadth of the speculation is poor.  Not as many names are working as 2025.  This year, its been the previously overlooked AI beneficiaries that have squeezed higher.  MU, SNDK, WDC, STX, AMD, INTC, MRVL, DELL, etc.  Semiconductors of all types are working.  AI hardware companies are experiencing huge earnings growth as the AI capex boom continues.  But beyond that, buyers are notably absent.  

The 2 main barometers of US stock speculation are NVDA and TSLA.  When they are stronger than the market, you know that your average stock investor is doing well.  While NVDA is holding up relatively well, its not rallying with the other AI semiconductor names.  It trades heavy.  NVDA had great earnings a couple of weeks ago and it sold off on the news.  Having a market cap of over $5 trillion will do that to a stock.  

TSLA trades worse than NVDA.  It is another household name that is one of the staple tech holdings among your average retail investor.  It is lagging the Nasdaq badly since it topped out last October.   NVDA has been popular among retail since late 2023.  TSLA since early 2020.  Investors already have enough NVDA and TSLA in their portfolios.  There are not enough new investors coming into the market to bid up those names.  

TSLA vs NDX

Same can be said for bitcoin as for TSLA, but to a bigger degree.  Unlike TSLA, which gets a steady stream of buying from inflows into index funds, bitcoin is a pure spec play with no passive flows coming in.  In fact, bitcoin miners are a steady source of bitcoin supply.  In 2025, you had bitcoin/ether treasury companies, in particular MSTR and BMNR, buying up huge amounts of bitcoin and ether.  That caused a big rise, but it wasn't sustainable.  The bitcoin flash crash on October 10, 2025 changed everything.  That was the event that marked the end of the post Trump election crypto boom.  After that, the bitcoin treasury company stock prices got crushed, making it difficult for them to continue to sell stock to buy more coins.  That iterative cycle of investors buying the crypto treasury company stocks forcing more buying of the coins was frozen in its tracks.  At that point, investors who were going to buy crypto had already bought it.  There were not enough new buyers (suckers) coming in to prop up an inflated meme asset.  Bitcoin got crushed this week and went down to below $60K, as the Nasdaq is still up big on the year.  That is what happens to a pumped up meme asset that is saturated with bagholders.  

As the market is getting more selective on what tech stocks to pump, you had GOOG coming to the market this week to raise $80B of new equity.  And on Friday, you had META headlines about looking to issue tens of billions of equity.   

With all this liquidity getting drained from equity markets with equity issuance, you have a very different supply/demand situation than 2025.  From a supply/demand perspective, this AI bubble is rapidly catching up with the dotcom bubble.

This is the backdrop as the SPCX IPO is coming up next Friday.  They are looking to raise $80B.  Preliminary IPO pricing is at $135/share and a $1.8T market cap.  If that's the IPO price, there will be very little upside for retail investors buying into this IPO, as the valuation is already so high that its hard to imagine buyers bidding it up to even a bigger nosebleed valuation.  $1.8T is a bigger market cap than META.  The earnings gap between those 2 companies is enormous, and SPCX is the more expensive stock!  SPCX is a space play, which doesn't have the earnings growth that speculators want these days.  It is a pie in the sky stock with xAI attached to it, which is a glorified AI data center company.  xAI has essentially given up on competing vs. Anthropic and openAI, and are just renting out their data center capacity to others who have the demand to actually use it.  

The put/call ratios continue to show lots of call buying.  In particular, retail traders are buying a lot of calls.  

The COT data as of June 2 showed both asset managers and leveraged funds reducing their SPX net exposure.  After the index went up.  That is uncommon behavior.  

 

Dealers took the other side of that trade, as they significantly reduced a big short position. They are going back to similar sized positions of the 2nd half of 2025.  It just so happens that the put/call ratios are also down to 2nd half of 2025 levels.  Dealers are likely delta hedging their short call positions by buying futures.  This matches the recent retail options activity data which shows lots of bullish call positioning.  

SPX Dealer Net Position

This week, you finally saw the AI hardware/semiconductor trade get taken for a big hit.  

On Friday, MU, SNDK, AMD, INTC, DELL, INTC as a group were down more than double that of the NDX.  For the more speculative AI stocks, they were down more than triple that of the NDX, which was down more than 5%.  Realized volatility went through the roof this week, but VIX only closed at 21.5.  This as oil was going down, which is a new development.  Up until Friday, the SPX/NDX and bonds have almost always rallied with oil going down.  That changed in a big way on Friday.  A change of character for this market.  

With the strong NFP number, bonds took a hit, and that seemed to be the initial trigger for the selloff.  But SPX kept going down, an extra 150 points even after yields had stopped going up.  Short term Treasuries look interesting after the selloff on the stronger NFP.  I will consider slowly building up a long position as I believe the Fed will be late to try to hike even with high inflation.  As the Fed delays rate hikes, the AI trade likely goes bust in the meantime.  And there is no way the Fed hikes as the stock market is imploding, I don't expect Warsh to do what Powell did in late 2018 by hiking into a falling stock market.  

I added to non-AI tech shorts earlier this week, and remain short.  I am still not interested in index shorts, as I would rather target the weaker speculative tech stocks than the whole market.  I will look to hedge some of those shorts with AI longs next week.  While Friday's selloff looks scary, they usually don't last for long, especially after such a strong uptrend.  I expect the uptrend to flatten out and get choppy with big down days followed by big up days.  That could last for a few weeks before you get the big move lower.  By July, it will probably be time to go short NDX along with speculative non-AI tech.  Until then, I will look to put on more non-AI tech shorts after rallies, and hedge with AI longs on dips.