ex frontier lab | macro, ai, tech investing thoughts | not financial advice.
singularity
Joined May 2025
- Tweets99
- Following51
- Followers19
- Likes272
the cascading effect from this is going to be crazy
tldr; the crash started on @binance
> @binance $SNDK printed its first large seller-initiated trade at 21:21:59.39 ($407K that second, $1.03M the next). @binance $MU started at 21:21:59 too
> @HyperliquidX first sweeps landed at 21:21:59.87 ($SNDK IOC, $SOXL trigger orders) and the main burst at 21:22:01.1 to 21:22:02.9: roughly ten addresses firing IOC sells across $SNDK, $SOXL, $DRAM, $SKHX, $SKHY and $MU within 1.8 seconds, then a second synchronized burst at 21:22:55.9. All IOC, none reduce-only, all opening shorts
> @binance carried the volume early: $SNDK $21.9M vs $3.1M on @tradexyz in the 21:22 minute, $MU $5.1M vs $0.3M. @HyperliquidX volume only exploded at 21:28 and 21:30 ($10.8M and $9.5M on $MU), which is the liquidation cascade
"I used to think that if there was reincarnation, I wanted to come back as the president or the pope or as a .400 baseball hitter. But now I would like to come back as the bond market. You can intimidate everybody."
— James Carville, 1993
Prediction: although their after-hours performance has been relatively weak/confused, $MU and $SNDK will skyrocket tomorrow.
A lot of the FUD around memory is around the cyclicality of the space and this ER call shows that there is going to remain demand until 2028 (and likely beyond) and that's a clear reason to own memory names. Also, they've gone ahead and locked down customers to longer-term contracts, which makes it less of a "one-time" necessity from the AI buildout and longer-term, which makes $MU look quite cheap.
$MU says memory and storage will be “much tighter” in fiscal 2027 and 2028 than in 2026 which is an escalation from last quarter’s “tight beyond 2027.”
Thats huge since a lot of the $40B+ capex is going into buildings and cleanrooms now while new supply needed to ease the shortage still takes years to arrive.
I’m long $NBIS here at ~$214:
This narrative around “pacing the frontier” seems like kayfabe & maybe a bit of warming up to the Democrats who have a higher chance of winning midterms.
But, the fact is: no lab can actually afford to slow down in terms of their usage of the underlying AI trade. Will they lose a bit of their margins? Likely, yes. But, does this nuke the underlying AI trade? Quite the opposite.
Putting more effort into alignment will ultimately require more compute and agents to monitor their RL runs & “secure” AI. If this is true, then ironically, AI Safety only increases the necessity for more compute & resources.
$NBIS just signed a deal with Palantir & if the thesis of “sovereign compute” takes place in markets, this could be a sharp V rebound.
Took a major loss today because of some terrible positioning. Unfortunate but we keep the head high and move on...
When it comes to making business decisions and proving the market wrong, there are only two goats: Jensen and Elon.
