Pre Market Thoughts 8 Jul 26

Yields inched 1-2bps higher across the curve. Overnight geopolitical risk is back with news of US and Iran love taps during market hours. While oil has rallied on the back of the that, it is also pricing in a "supposedly glut" for next year.
My thoughts on Oil after March 26? This is a market where even the so called experts get it wrong. Waddling here is probably deeply negative EV.
Next, asia markets took a huge beating. KOSPI is down 5.5%, NKY is down 1.6%. Today on NY close is the end of the subscription period of $SKHY. At this price point, the case is either you believe in the earnings and the non cyclical case for HBM and DRAM or you dont. $SKHY is now back to the friday/thursday lows. This has dragged down $MU and $SNDK in thin overnight markets. Again, levels where i believe there is good r/r is ($MU 800-850, $SNDK 1500s). The caveat to this is that while i expect the momentum puke to be over already, it is currently still on going. If you want to bid this, you must be ready to hold for a while.
In other news, semi analysis is writing about Anthropic and OAI not being able to serve their models because of a lack of compute. Also they estimate Anthropic's valuation to be at about 6trn at 20x 200b ARR in 2027. This is a gargantuan number and marks a massive payday for VCs. On the compute shortage, especially as OAI and Anthropic ramp up towards serving inference - this changes the risk in their business models. In AI usage, the power law distribution holds. AI is of tremendous value to some people (those with high agency - i strongly believe we will see the rise of billion dollar single man companies in the future) and of some value to most people.
Compute market currently still remains in short supply, especially with Anthropic and other players gobbling up supply. Even the short term rental markets have no supply for many of them. A100s (i just checked) are renting for $1/hr....this is something that was launched 6 years ago - This means that for the people renting out A100s, they would have made back everything in 2years (rental was ~2.5) Every year after that is pure profit.
Anthropic's 3Q26 profit was also leaked at 1B. The bear case for compute - that the frontier labs will collapse because of an inability to raise funding - that case is dead. The frontier labs are now profitable. To scale, they need more compute. Serving inference is a 70% gross margins business for them and the compute shortage will tighten even further as the labs jostle to get market share.
Personally, i have rotated all my plays into neoclouds. This remains the most mis understood business (2 year payback periods) and the core inflection point now is the frontier labs turning profitable.
NVDA providing credit support for neoclouds is extremely bullish for them because it reduces the interest cost in financing the build out.
TLDR - while this was meant to be a market update, i could not help but to put this down as the biggest play for 2H'26. IMO, we are still in the very early innings because people do not yet have a grasp of the economics of the trade. This market is like memory in October 2025 - some people are seeing the vast need for compute and the inflection point that the frontier labs being profitable mean - while others point to an outdated bear case that sounds sexy but fails to live up to scrutiny.
What neoclouds would i bid? $NBIS in my opinion remains the leader of the neoclouds in terms of execution. $CRWV is a strong contender, however 98% of its contracts are locked in and hence it will not be able to directly monetise on the spot shortage in compute. However, $CRWV has already been rumoured to hit its 30B ARR target exiting 2027. $IREN boasts power however, in terms of deal execution - this has been lacking.
For the super speculative play, i like $SHAZ. Again, this trades at ~2B market cap. It has 40,000 GB300s which when rented out will be able to fetch $6-11/hr which means an ARR of 2-4B. If we apply a 6x ARR multiple to this, this is 12b-24b. Also $SHAZ capital structure is relatively clean with a thick equity cushion and also NVDA credit support for its GPU purchases. I like this play for the massive risk reward it offers.
Now - does this mean that neoclouds will rerate upwards immediately? No. In a positioning wipe out, everything gets sold. However, cashflows and valuations will dominate in the longer run. For a generational trade like neoclouds (where i see at least a 3x rerating on $NBIS, $CRWV - because of Anthropic's profitability)
As always, i welcome thoughts on the matter, especially to see where i may have gotten it entirely wrong.
Good luck!