On Top of a Powder Keg

Its getting late in the game.  It feels like the bull market will last forever, as all corrections have been immediately bought up since October 2022.  Equity investors must feel invincible, seeing the SPX and NDX blast up to new highs despite the Strait of Hormuz still being closed, now coming up on nearly 3 months and counting.  The AI names have gone up so much, that they are dragging the index up with it, despite a bunch of lagging names.  It is what you saw in the later stages of the dotcom bubble.  

SPX component stock correlations to the index are at their lowest since 2000.  These dips lower in correlations over the past 10 years have been around market tops: mid 2018, late 2021, early 2025.  

Retail investors continue to pour in record amounts into both stocks and options.  The flows have been elevated since Trump got elected in November 2024. They have reached an even higher level in 2026.  Recently on CNBC, they said that retail investors are now the smart money.  Amber flashing lights going off in the background.  

Its not just retail that's very long.  Asset managers are near their all time highs in SPX net longs as a percentage of open interest.  

A big portion of semiconductor stocks in the SOX have inverted put/call skew (calls more expensive than puts).  Historically bearish for SOX over the next few months.

AI is becoming a huge part of the financial markets.  YTD net issuance of IG bonds and venture capital show how much is going to AI.  49% AI for IG bonds, 87% AI for VC.  

There are a lot of AI adjacent stocks out there.  AI is basically all that matters now.

It seems like OpenAI has speeded up their IPO plans from 2027 to September 2026, according to the Wall Street Journal.  Trying to feed the ducks while they are quacking.

So Space X will be doing the biggest IPO in history in June, with a accelerated lockup expiration that will release a big chunk in August after earnings, not the traditional 180 day lock up period. Open AI is trying to get public as soon as possible.  And Anthropic will be following soon after.  SpaceX estimated IPO valuation: $1.75 trillion.  Open AI private market valuation: $800B+.  Anthropic:  $900B+.  Usually private market funding rounds are at lower valuations than the public IPO valuations, so that is some serious supply coming.  If those valuations hold up after 180 days being public, that's releasing a potential $3.5 trillion of supply on to the US stock market by early to mid 2027.  

 The US and Asian stock markets are now just big bets on AI.  It will be interesting to see what happens when the AI capex tops out.  It is a big bear catalyst that is looming in the horizon.  But its a hidden bull catalyst for bonds which the market is ignoring due to the war.  

With the Strait being closed for nearly 3 months now, inflation is rising, and Fed is getting more hawkish.  

You are hearing more worries about the bond market in financial media, as the 10 year yield broke out above the 4.5% barrier.   For AI skeptics, this is setting up an opportunity to make a good risk/reward bet on the AI bubble popping pushing the US economy into a recession, like 2001.  

There is still the bull catalyst out there, the carrot in front of the donkey in the form of a Iran War deal.  It is what helped push oil prices lower on Thursday and Friday, helping the markets.  It is surprising that the market still reacts so strongly to these mostly bogus headlines.  One of these days, the headlines will be true and you could get a big squeeze higher in stocks and bonds on that day.  That is what keeps me from putting on index shorts.  And makes being long bonds a better bet than shorting stocks at this point.  

Small speculators have rapidly reduced their long exposure in 5 year Treasuries, and as of May 19, have become net short.

5 Year Treasury Note Futures Small Speculator Net Positions

We got a 3 day pullback from last Friday to Tuesday based on bond market fears.  As I said before, I don't believe you should short the index due to a weaker bond market.  That is a 2022 playbook.  I am going with the 2000-2001 playbook.   

While investors have fully bought in to the positive stock/bond correlation, I expect that correlation to go negative like it did in 2001.  Bond yields should top out before the SPX tops out.  For bears, buying Treasuries (short duration) is probably the safer play than shorting the index.  On the sidelines waiting.  Over the next few weeks, I see more opportunity in the bond market than the stock market.  

Its 2000 Not 1999

Big difference.  I hear a lot of talk about this being like 1999.  No, its not.  The breadth of the speculation was increasing as the year was progressing in 1999.  That is what you saw last year, not this year.  The breadth of the speculation is shrinking.  Last year, you had a lot of speculation in crypto, nuclear, quantum computing, meme stocks (TSLA, PLTR, etc.).  That has died down, and speculators are now laser focused on AI hardware:  GPUs, CPUs, semi equipment manufacturers, memory, storage, optical, and data center energy plays. 

In 1999, the market was enamored with companies that were internet based, that provided internet services, not the companies that provided the hardware and infrastructure for the internet.  Only in late 1999 and early 2000 did the market started narrowing and focusing on the picks and shovels of the internet, as they realized that the internet would require a lot of capex investment.  Semiconductors skyrocketed and went parabolic at the very end of the dotcom bubble.  They were the last companies to boom higher.  

This AI bubble has played out a bit differently as the biggest providers of AI are private companies, so they have skipped out on the early part of the bubble.  If OpenAI was public when ChatGPT came out, they would have been a trillion dollar company by 2024.  But due to easy capital available in private equity and less public scrutiny, they decided to stay private, probably because they were burning so much cash.  Burning tens of billions of dollars and maintaining a high valuation is much easier to do when you are a private company than a public company.  So with this bubble, its started with the hardware side but has expanded to peripheral hardware.  From GPUs and data center energy plays to DRAM, storage, and more recently CPUs and networking.  As you can see in the SOX performance in 2000 and its performance in 2026, there are some interesting parallels to the dotcom bubble.  

SOX 1999 to 2001

SOX 2025 to May 2026 

 

MU actually was doing nothing for much of 1999, until it suddenly went parabolic in 2000.  MU wasn't doing much for the first 8 months of 2025, and then it suddenly went parabolic starting in the fall of 2025.  CSCO was the favorite big cap tech stock of the dotcom bubble, like NVDA is today.  MU massively outperformed even CSCO at the end of the bubble.

MU (yellow) vs CSCO 2000

 

Expanding production capacity in memory takes years, so DRAM pricing is inelastic in the short to intermediate term.  This inelasticity works both ways, as a sudden surge of demand causes a squeeze higher in prices, which then collapses when the demand cycle peaks and new production capacity starts coming online.  A double whammy of higher production capacity and lower demand.  It will happen again, but investors are blinded by the huge growth rates, the extreme profitably that makes these stocks look cheap on peak earnings in a hyper cyclical industry.    

A new memory ETF, DRAM, which includes Samsung and SK Hynix, have received huge inflows over the past month.  There is serious FOMO chasing the most volatile big cap tech stocks.  It is rare to see a specialty ETF like DRAM attract as much inflows as SPY and QQQ over a week.  There are blowoff top vibes when the crowd goes crazy over a small group of stocks like this into a parabolic rally.

Even some of the favorites during the dotcom bubble like CSCO and QCOM are joining the party, even though it would be a real stretch to consider those stocks AI infrastructure plays.  

But is just the AI plays that are running.  The rest of the SPX are lagging badly, and a big portion are below their 50 day and 200 day moving averages.  Its a narrow market.  Most are interpreting that as bearish, which is wrong.  You had strong breadth in January, as Nasdaq lagged, and look what happened in February and March.  Nasdaq is what matters, and in particular, the biggest cap stocks in NDX.  People buy US stocks because of tech, not to get exposure to banks, health care, or consumer related stocks.  If tech isn't performing, there is no reason to buy US stocks.  You might as well buy cheaper European or Asian stocks that cover the same sectors.  

You will know when a bear market is imminent when the Nasdaq starts lagging the SPX and breadth improves.  That probably happens when the Strait opens and everyone gets complacent.  Right now, you still have that wall of worry, that positive catalyst in the front view, which is the Strait opening and oil prices going lower.  

There has been quite an increase in call speculation among small speculators (less than 10 options contract orders).  Almost as high as late October, which was a local top.  

 

Retail is also back with big inflows into tech.  Now they are no longer interested in catching the falling software knife, and are piling into momentum names in tech hardware.  It has just been one big week of inflows so far (as of May 6), so it may take a few more weeks for retail to get fully allocated and run out of dry powder.

We had some really bad inflation prints this week, with both CPI and PPI coming in hot.  There are consequences from running fiscal deficits above 6% of GDP in an expansion.  Add to that the $166B in tariff refunds going out this year, the lower tax rates from the OBBA going into effect, plus war spending.  All that fiscal largesse is chasing the same number of goods and services, so naturally prices go up.  Don't forget that the Fed is doing stealth QE as they buy tens of billions of T-bills every month to keep an ample reserve regime, and lessen the pressure in repo markets.  It would be a surprise not to see high inflation in such an environment.   US 10 year yields finally cracked above 4.50%, and JGB yields keep making new highs, now well above 2.5%.  Higher bond yields are a feature, not a bug of a banana republic fiscal policy with no desire for less government spending or higher taxes.  

Gold has effectively replaced US Treasuries for China, Russia, and a growing number of other emerging economies.  Gold is slowly replacing Treasuries as the central bank reserve asset of choice, as the US government shows no desire to reign in massive budget deficits.  

Despite the higher inflation and ever increasing fiscal deficits, I am not bearish on Treasuries at current levels.  I see a couple of positive catalysts for the bond market.  1.  Opening of the Strait.  2.  The popping of the AI bubble.  It may take some time for these catalysts to play out, but its a matter of when, not if.  If I were to buy Treasuries, I would focus on short duration, as I see a curve steepening once the positive catalysts play out.  Unlike the 2022 bear market, I expect the next equity bear market to be positive for bonds, as I expect the Fed to aggressively cut rates like Greenspan did in 2001 after the dotcom bubble burst.  With so much of global wealth tied to the US stock market, I expect significant negative wealth effects from the next bear market.  

The COT data for the week ending May 12 continues to show asset managers aggressively adding to their big net long SPX position.  This is a very early warning sign of turbulent times ahead.  Only after the asset manager positions have peaked and start going lower can you expect there to be a market top.  It usually takes a couple of months for the market to trend lower after a peak in asset manager net longs.

Was waiting to get short SPX on an extended move higher, and we got that on Thursday, as SPX burst through 7500.  Unfortunately, was waiting to short on Friday open and missed the entry due to the big gap down on bond yields shooting higher.  We got significant dips in all the high beta tech plays in AI hardware/semiconductors on Friday.  A lot of this pullback seems to be opex related, as the tickers with the biggest options volume and OI were all hit hard (NVDA, TSLA, AMD, MU).  

With the one day pullback on Friday, not much to do, but wait for the market to get back to all time highs or even higher to pull the trigger on the short.  Its still the higher probability play to buy dips in this market, as I expect dips to be shallow and bought quickly until you see the uptrend flatten out for a bit longer.  The AI hardware plays should continue to be the ones that go up the most when the market bounces from these dips.  Those with a bearish bias, who are adverse to playing bubbles from the long side should be patient, as I expect a grind higher for a few more weeks.  That 4%+ pullback is probably something to expect in June or July.  

AI Hardware Mania

AI is all that matters.  I see a lot of parallels between 1998-2000 and 2024-2026.  Back in 1998, it was the Asian contagion and LTCM.  In 2025 and 2026, its tariffs and the Iran War.  Those macro events were a distraction from the secular tech bubble that was growing.  

Iran is just a distraction.  For those who think the Strait of Hormuz will be closed for a few more months, then go long oil, not short stocks.  Don't try to fit a square peg into a round hole.  Oil-stock correlations are going back to where they were before the war, which is near zero.  Don't have a strong view on when the Strait opens. But I do know that with the Strait closed, you are probably not going to get investors all in on stocks.  And if investors aren't all in, then you are not going to get a long term top.

The Strait is closed and the SPX continues to go higher, making new all time highs.  Clearly there was pent up demand for AI stocks while investors were worried about the macro implications of the Iran War.  Even without the war ending, they just couldn't hold back the momentum.  The momentum for AI hardware:  CPUs, GPUs, DRAM, storage, optical networking, etc.  It is reminiscent of the dotcom bubble in late 1999/early 2000.  The sharpest up move in the Nasdaq happened at the end of the move in early 2000.  We are seeing sharp up moves in semiconductors and assorted tech hardware names.  Retail is starting to jump on the bandwagon again.  

 

The hottest tickers on Reddit Wall Street Bets are AMD, MU, and SNDK.  There wasn't so much retail buzz in January, when MU and SNDK were exploding higher.  Back in January, WSB was more enamored with gold and silver than AI hardware/semiconductor stocks.  Retail has a knack for getting excited around long term tops.  It happened in bitcoin in summer 2025.  It happened in gold/silver in January.  And its happening now in the AI hardware space.  When long term trends go parabolic, the end is near.  It happened in precious metals in January.  We are likely within 1-2 months of a blowoff top in these AI names.  

Timing the top of this blowoff move is not like trying to short the top of a short squeeze like CAR.  Short squeezes don't last long, because the narrative doesn't have staying power, there is no fundamental basis for the rise.  They collapse quickly.  But there is a fundamental basis for this parabolic move in AI hardware.  So that gives it more staying power, more fuel to keep going higher.  Rather than trying to time the exact top on the frontside of the move, it may be better to short these names on the backside of the move, after the top is clearly in place.  There is lots of room for downside even after these things go down 10% from the top.  

I remain an AI skeptic, as I see no way that there will be a positive return on investment on all this overpriced AI hardware.  

Equity inflows have been heavy for the past few weeks.  

Last week, BofA clients were heavy buyers of stocks.  

 

We are in the last stage of the AI bubble, as the best performers go parabolic, and retail investors get FOMO and pile in.  2026 = 2000.  The last stage is where the biggest gains are made.  So it is dangerous.  Shorting early in the final part of the bubble is deadly.  

How quickly we forget about tariffs.  But a lot of that tariff revenue that was generated over the past 12 months is going to be refunded, providing more government stimulus.  $166B is the estimate.  Importers get to collect the refund, plus the higher prices they passed on to the consumers as a result of the tariffs.  In the end, corporations win, consumers lose.  

 I have been on the sidelines for the past several days, waiting for a good spot to enter shorts.  With May opex coming up, it appears to be a good time to consider a short on any further rallies this week.  The SPX and Nasdaq are getting extended, and it would be normal to see a post opex hangover after the big run up in the tech names.  I don't expect much of a pullback, maybe 3-4% at the most, but it is worthy of at least a small trade.  

SPX Donkey and the Deal Carrot

Geopolitics and the stock market don’t hang around each other for long.  Even with the Strait of Hormuz still closed, the SPX closed at an all time high.  If you had told investors in March that the Strait would be shut into May, almost no one would have guessed all time highs, and definitely not more than 4% above pre war highs!  The half life for a war to affect the stock market seems to be about 2-4 weeks.  That’s about how long it takes to get weak hands to sell (I was one of those weak hands).   

You are now 2 months into the Iran war, with just a ceasefire and no deal, with the Strait still closed.  Yet here we are above 7200 on SPX.  The market has shown its hand.  It wants to go higher.  Its counterintuitive, but the stock market will be more vulnerable to a drop after the Strait is open than before.  Having a shut Strait keeps that carrot in front of the donkey.  That donkey will keep going trying to eat that carrot, just in front of it.  All these Iran deal pumps by Axios are the carrot.  The anticipation of an exciting event is often better than the event itself.  Once the donkey eats that carrot, there is nothing to look forward to.  It relaxes.  It rests.  

I don’t think this market will make a top until after the Strait is open.  The opening of the Strait is a bull catalyst.  Shorting in front of a bull catalyst is dangerous.  Even when the market looks overextended.

This war is a good way to filter out those on Fintwit and the financial media.  Those still talking about the Iran War, oil prices, and supply disruptions affecting the stock market are stuck in the rear view, always looking backwards.  These are likely the ones that you want to fade.  Those who have moved on and realize that the market is focused on other things, mainly the ongoing AI boom, are looking forward.   These are the ones that understand markets better, and are flexible and listen to the market instead of CNBC and Bloomberg.

Last week was filled with big tech earnings and the FOMC, events that could have stopped the uptrend, but ended up being a wall of worry that the market climbed over.  There was nothing new in these earnings reports.  The hyperscalers are still pouring massive amounts into AI capex.  But they are getting less bang for their buck, because of the ridiculously high prices for semiconductors.  At some point, big tech companies will want a return on that investment.  And while they are less price sensitive now, the closer they get to the end of their AI buildout, the more price sensitive they will become.  

This grace period for reckless spending on AI is going to end sooner than people think.  It looks like its already ended for META.  Who is the next Mag7 to join that list?  The SMH (semiconductor ETF) has outperformed IGV (software ETF) by over 60% in the past 6 months.  This is dotcom bubble territory for tech divergence.  

 

 

There is a generational short opportunity building up in semiconductors/hardware.  It is loved by the investment community.  The rally is getting parabolic.  Timing the top is the hard part.  If you have a long term view and lots of risk tolerance, you could short now, and profit handsomely a year later.  But if you short now, you may have to sit through a painful rally for the next few months.  The options market isn't showing signs of excessive pricing of calls vs puts, which led to reversals in previous rallies in the SOX index.  

 

You have now seen a return of the call buyers into this market for the past 2+ weeks.  The risk appetite is returning, but its been much less broad, as semiconductors and AI centric energy names have been the outperformers.  The ISEE index is now back towards late 2025/early 2026 levels, so investors are definitely feeling more comfortable speculating on further upside.  

 

Hedge funds have not fully embraced this rally in tech.  They were big sellers of Info Tech from April 17 to April 30.  Hedge funds definitely act differently than they used.  Now they seem to be less trend following and more mean reversion / trend fading in their behavior.  There is less dumb money in the hedge fund space these days, so their positioning information is not as useful as before.  

Bigger picture, you have very high allocations to stocks.  Not a great timing tool, as this level has been elevated for the past 2 years.  But it gives you an idea of how much potential weakness is possible when the bull market tops out.  

We are now 5 weeks into the rally off the March 30 bottom.  These rallies off a capitulation  bottom in a bull market usually last about 4 to 8 weeks, and flatten out and then have a pullback.  So we are now in the timing window where we could get a pullback which will likely be bought.  Both upside and downside are limited during the flattening out phase.  

 

As I write, news of Iranian missiles hitting a US warship near the Strait.  I stick with my view that shorting while the Strait is closed is dangerous.  I see little edge either long or short here, so will likely be on the sidelines for the next few days.