pre mkt thoughts 22 Jun 26

US yields relatively unchanged in Asia. Meanwhile big gyrations in asian markets. Lots to unpack here! While KOSPI and HSTECH was unch, NKY is up 1.5% from the open. Underneath this mask of index calmness, single stocks are carrying the entire story. In the memory complex, $285A is up ALMOST 10% from last week, even though the move is only 1% today, this masks the almost 10% move done on friday when US markets were closed. $SK Hynix is also up 4% for the day. As i have reiterated many times, both $285A and $SKHYNIX looks set to get a US ADR listing soon - this will be a core catalyst for further rerating.
The star of today's show is $2513HK and $100HK, both up 17% - 20%. This are Zhipu and Minimax respectively. They are the Chinese model providers. For context, the chinese models have focused on being open sourced and relatively light weight. So why the rerating now? Because Zhi pu's GLM model 5.2 is about the same level as Opus 4.7-4.8, a convo with elon over twitter estimates them to reach Mythos level capability between Nov and Dec 2026!
There are MANY ramifications here. First of all, the HK markets may finally ignite. While $BABA is dead at the moment, as the chinese AI rerating continues, people will realise that $BABA owns more than 10% in many of the chinese frontier labs AND also supplies the T head chips.
I would also like to introduce readers to Kimi - (not listed, ANOTHER Chinese AI lab). Instead of trying to be good at everything, Kimi focuses on agentic coding workflows - in this aspect they are comparable to Opus as well, if not better.
Bear in mind that at this point, the chinese AI models are at least 4 times cheaper (if not more) than those by Claude or OAI....
Also, the competition is not just the chinese labs with Sakana Fugu - with the Fugu Ultra model claiming to "match the performance of Fable and Mythos - delivering frontier capability without the risk of export controls."
What does all this mean?
First we are in a future where AI frontier labs are in deep competition, this is directly bullish for the neocloud thesis - more compute is needed. Second, models will become more powerful and the common theme of agentic workflows will require even more memory - this is directly bullish for memory (tho rerated already) and CXL solutions ($ALAB, $CRDO, $PENG)...
On a more technical front, i would also look to reduce down leverage going into the end of June. Whilst the AI trade is strong fundamentally, flows wise this current period will be met with headwinds. Personally i would try to see if i could add leverage on a dip during this week and next.
Good luck!