for our inaugural weekly letter, this song is too timely (released June 26) & too relevant & too good to pass up.

LINE GO UP LINE GO DOWN.

that’s what it’s all about. (unfortunately??)

sorry, Chad Matheny (Emperor X), but we cover markets a lot here. you’d probably hate that. but also art & stuff, too.

it will be difficult to not share an Emperor X song each week. but we will try our best to not do that.

we share updates on Wednesdays, because Wednesdays are generally slower.

with Wednesday’s calm waters, we may gaze out at the horizon & quickly grok how Monday, Friday, & the weekend impacted ships that dared oppose Mother Nature Market.

we perch—coffee in hand—& calmly observe which of the haughtiest specimens from past weeks have somehow(?) vanished. (they’re sipping coffee with Davy.) some ships emerge from storms with additional strength & gusto. many just get by.

we’re not sure. markets, news, ???

this entire project needs legs, even if they’re shaky. it just needs to get going. we’ll chisel it into something beautiful later.

it’s late and we want to sleep, so here’s a bunch of opinions on various things that we will probably (but perhaps not) come back to later.

saw something big happened with Neuralink but didn’t look into it further. we think it reinforces my hefty bias towards the future being “wet” and society increasingly shifting focus to biology, longevity, healthcare, drugs, &c.

this mirrors the shift from “soft” wares to “hard” wares like magic rocks, magic rock production, & magic rock containment buildings (data centers). and obviously a lot more beyond that.

tangible goods and a global re-evaluation of what “value” is and why certain things matter or not.

$F ( ▲ 0.07% ) has re-hired a bunch of experts to train both young people & AI.

that’s a nice twist on the (fake) AI layoffs narrative that’s driven primarily by prioritization (by bloated, shareholder-appeasing companies of silicon magic rocks) over human lives. tech companies would rather insert crumpled dollars into Zoltan AI than invest in (re)training human beings.

human beings that not only have an exponentially higher capability/creativity ceiling, but also human beings who buy products that fund their Eternal Enshittification Engines!

there was also a study of >20,000 companies that found the companies using AI the most (and not just testing the waters) hired more entry-level workers. we think it was something like 10% more. they also fared better as businesses.

another nice data point that contrasts with Zuck’s bed of AI-washed lies.

speaking of Zuck…

Zuck wants to make $META ( ▼ 2.46% ) into a neocloud now (selling “compute” power to others that need it).

if that isn’t a giant admission that your AI models suck and you don’t know what to do with the compute power yourself, we don’t know what is.

add to that:

should we take bets how long it takes Meta to offer the latest GLM & Kimi models for its customers? (no, because betting is stupid.)

markets should’ve nuked META from orbit for this. instead it rallied, while companies offering legitimate, expert “neocloud-plus” offerings dumped.

$NBIS ( ▼ 13.9% ), the next hyperscaler, dropped because Zuck erroneously let a thought bubble turn into a speech bubble? Really?? (we are big Nebius fans, especially because they’re fans of Nous Research—the best open-source research lab. we’ve seen Nebius sponsor Nous Research events & provide free model access in the past through their cloud, which is nice of them. and for the record, Nvidia/Jensen is also a big fan of Nous Research. Jensen also loves Nebius.

anyway, that brings us to…

$CRWV ( ▼ 5.46% ) & $CBRS ( ▼ 1.93% ) both brag about their Kimi inference. Databricks has been posting frequently about how far it can push the performance of GLM-5.2. Nebius is working with Eigen (which it acquired) to develop optimized versions of GLM and other open-source models.

Neoclouds: all-in on Chinese models.

then you have $COIN ( ▼ 4.02% )’s CEO saying companies should use Chinese models to work the most cost-efficiently. $UBER ( ▲ 1.89% ) capped its employees’ AI budgets after they blew through their token allocations too quickly. Meta apparently spends like $50k per employee per month on tokens. (they’re laying so many people off because it’s expensive to run a dumpster fire, by the way.)

OAI’s IPO delay was an IPO cancellation in disguise. sneaky, sneaky Sama!

A\ can still self-immolate frontier labs & markets later this year if it so chooses. it’s fearlessfull leader seems to not have a great grasp on reality or how the world works, otherwise his model wouldn’t be banned & his company hated by the US government. how’s that fearmongering going now? (apart from not materializing, repeatedly.)

the frontier labs shouldn’t IPO, unless progress slows. short-sighted markets won’t appreciate their long-horizon vision and will douse their soul flame in an instant the first time growth stops growing faster than growth is expected to grow. or something stupid like that.

“stay private, OAI.” -Johnny Cade (speaking to Ponyboy Sama)

then emerge transformed, in a million years. or just a few years.

this is getting long…

let’s leave the rest for next week. here’s a…

there’s news below all this if you’re interested. curated daily and sent out weekly.

see you next week.

_{AURUM NOSTRUM IN STERCORE SPIRITUALE EST ⚕ СЛАВА ЛЮДСТВУ}

here’s everything else we paid some attention to this week. one sentence each + categorized.