Jason T

Cool stuff

[Yuchen Jin @Yuchenj_UW]

“A single cluster currently serves 155 trillion tokens/day, generating $38 in revenue per GPU-hour (at off-peak pricing).”

Huge flex from DeepSeek. Not only did they open-source their inference engine, but they also open-sourced their revenue numbers.

They didn't disclose which GPUs they're using here, but DeepInfer supports both NVIDIA Blackwell and Huawei Ascend.

Even assuming they're renting B300s at $7 per GPU-hour, that's ~82% gross margin.

[Lagoon Labs @LagoonLabsMv]

Bill Gross (Idealab founder) built an AI-powered database of 1.2M startup ideas from patents, news, social media. He's launching https://t.co/7HY6MFEKY6 to match founders with ideas, collaborate, and tap $100M in funding.

[Joe McCann @joemccann]

Finally, a forward looking indicator for the markets.

Taiwan and South Korea Chip Exports Lead US Earnings by 6 Months; No Downturn Signal Yet

The average of Taiwan and South Korea export growth, pushed forward six months, closely tracks MSCI US forward 12-month EPS growth.

Both countries are "significant producers of the chips that are the foundation of the AI investment boom" and are released monthly, with South Korea publishing a 10-day estimate around the 21st of each month.

Chip exports are "likely to detect the pullback with about a 3-6 month advance."

Crucially, as of the October 15, 2026 publication, "South Korea and Taiwan's exports currently give little sign that the investment cycle is imminently about to turn down."

When they do turn, that's also a cue for yields to decline since chip exports lead nominal GDP as well.

[Archaeology & Art @archaeologyart]

John Martin - Satan Presiding at the Infernal Council (1824)

[Nathan C. Frey @nc_frey]

Where do you go next after reporting from the Strait of Hormuz? For the past three months, Citrini Research, an independent research firm, has been working with our team and an internal research model to see if they could use AI to create a new drug. What follows is an entertaining, candid, and humbling account of a group of motivated nonexperts (or “filthy casuals”, as Citrini would say) trying to tackle a real problem in human health with Claude.

We landed on scorpion neurotoxins, which kill over 3,000 people a year. Of the 211 designs we produced and tested, 17 met our criteria for a binder. None of the designs bound more than one toxin and none have been tested for neutralizing a toxin.

One of my favorite quotes from the piece is: “Our virtual biologist may have compressed a year's worth of work into a couple days, but biology works at the speed of life…Over two days, we watched Claude work through problems at a speed that was difficult to comprehend. For the next two months, we spent much of the time becoming familiar with laboratory turnaround times (This mostly involved sending memes in Slack).”

[Meltem Demirors @Melt_Dem]

first follower 💪

excited that Garrett is finally on this website to share his incredible plasma knowledge

[Trung Phan @TrungTPhan]

Eric Schmidt owns 1% of Anthropic. Dario’s wife — who Schmidt used to date — introduced him to Anthropic.

Schmidt’s peak ownership of Google was 5% (now ~1%). He got that Google CEO job after going to Burning Man with Larry Page and Sergey Brin.

Next-level networking skills.

[Garrett Hill @G01plasma]

Are we just studying cosmic physics, or are we holding mini-universes in our hands?
In the lab, a 1.5” diameter plasma discharge ⚡️elegantly mirrors the dynamics of a stellar event.
If the physics of a bench-top reactor and a galactic supercluster are scale-invariant, plasma is the universal bridge connecting all of us to the stars. #PlasmaPhysics #DeepTech #LaboratoryAstrophysics #Cosmology #RIMERE #UniversalPlasmaTechnologies

[Eric Weinstein @ericweinstein]

This is quite right. In the 1980s well into 2010s there was a crazy idea of employers, economists like @Noahpinion, and conservative ideologues like those at @wsj (e.g. Robert Bartley) that borders should be open because mass immigration was the rarest thing in all of economics: a complete free lunch.

I am not making this up. Actual adults. And conservatives at that.

And some people at the UN (e.g. Manolo Abella) and I tried to fight this mass immigration mass delusion using economics. But you have to have a high school education to understand the argument that we were attacking the @noahpinion like ideological zealots using their own economic theory and claimed idealism. The internet is struggling with this idea. An 80+ Iq is helpful but not essential here as well.

The point was to make the Free Market pay for immigration. And free marketeers hated the idea. Why? Because they weren't free marketeers. Ever. Duh.

They were thieves and fools.

You see the open borders movement was always psychotic and never about "the best and the brightest" nor "freedom" nor "markets". It was an anoying smug debating position which may yet lead to civilizational collapse in the West. Ask @Noahpinion et al.

Mark my words: It was always about theft at the expense of national suicides. Nation by nation.

[Frans Bakker @FransBakker9812]

Well ok then, I'll bite.
@SemiAnalysis_ is once again full of crap.

"And here’s the truth for you: onepicture, >24GW of datacenters in advanced stage of construction, all to be delivered in 2027."

No your "platform" does not have accurate tracking, nor does it have a right interpretation of such.

E.G. Crusoe - Texas.
According to SemiAnalsysis, 1GW of the Childress "Hyper" Campus for $META, will be delivered in 2027.

What does the more accurate tracking show us?
Pic 1. Satellite image from 10/18, yes, that was yesterday.
Pic 2. Aerial image from 10/9, 1.5 weeks ago.
Clearly this is in very early stages of construction, they have 6 transformers in their bulk substation, but there is hardly any step-down infrastructure in place.

Lookign at the actual building foundations, this is absolutely not a 2027 story.

But why would you trust me? My last name isn't Patel, I'm no famous headphones wearing AI quant?

Maybe listen to the actual company itself?

Crusoe has 3 TDLR filings for Project Hyper, 3 separate buildings, which 3 separate completion date. February, April, and May of 2028.

I added building 1 as the 3rd image, to make it clear this will not be a 2027 deployment.

So, a picture with many data centers may look very impressive, but SemiAnalysis uses free satellite imagery for its paid subscribers, to then mislead them with wrong intel that could be sourched for absolutely free with ChatGPT.

STOP PAYING MONEY TO THESE CROOKS, THEY ARE NOT RESEARCHERS, THEY FILL THEIR POCKETS WITH YOUR MONEY TO BUY NEW HEADSETS FOR EVERY NEXT INTERVIEW, STOP GETTING FLEECED BY THESE BASEMENT DWELLERS AND CANCEL YOUR SUBSCRIPTION TODAY.

[Atoms Not Bits @AtomsNotBits]

BREAKING: The world's first subsea geothermal well has been drilled into an underwater volcano nearly a mile below the Pacific Ocean's surface by American startup @endurancegeo.

300 miles off Oregon's coast, the company installed a 3-ton geothermal generator on the seafloor, operated it remotely from Seattle via fiber-optic cables, and briefly produced more electricity than it consumed using heat from a nearby volcanic vent.

[Allie ✞ @allie__voss]

Jia Jiang is famous for trying to get rejected as much as possible for 100 days, and he mostly learned it's harder than you think to get rejected

Things he tried:
-Asking a stranger to borrow $100
-Asking for a "burger refill" at Five Guys
-Asking to speak over the Costco intercom
-Asking for Olympic ring-shaped donuts at Dunkin
-Asking to plant flowers in a stranger's yard
-Asking to ride in a police car
-Asking to take a video of him playing soccer in someone's yard

In most cases even if people said no they still at least tried to help him in some way, because the "no" was mostly about concrete constraints

Rejection is rarer and typically far more gentle than you're expecting

(CC: @JiaJiang)

[pointsnfigures1 @pointsnfigures1]

Here is an example. When I was on the CME Board, we were working hard to try and bring a cash treasury contract and trade it inside the pit so traders could compete with the outside traders who were trading cash/futures. We worked pretty hard on it and finally were in presentation mode to small committees. Leo Melamed loved this idea for a lot of reasons. He presented it. Don was in the room.

We opened it up to questions. Don asked the first question. "If you are trading cash, there is usually a 5% haircut in and out of the market how are futures traders going to handle that?" Combine that with the amount of margin that had to be put up and it was dead in the water. One of the shortest meetings I was ever in. It's taken about 25 yrs, and CME is trying to trade that stuff because rules are different, players are different and it's all electronic.

[Beni @beniduboss]

I am never leaving this app LMFAO

>be Don Wilson
>predict compute markets will be > than oil within 10 years
>anons call you delusional
>lmao.exe
>fyi this is your track record
>1992: found legendary prop firm DRW
>name it after your own initials (omega based)
>pioneer quant and electronic trading
>your firm trades Bitcoin in 2010 (bruh)
>launch Cumberland in 2014
>become one of the biggest institutional LP in crypto
>co-found Compute Exchange and back Silicon Data to build GPU markets and pricing indices
>literally became a billionaire by being early for over 30 years

like bruh you’re literally arguing about markets with a man whose racing sailboats are called “Convexity” LMFAO

you can’t make this shit up

incredible stuff thank you for the laughs

[Hedge Fund Clips @HedgeFundClips]

Pabrai’s Citadel story: Ken Griffin hired a temp to protect a mathematician’s uninterrupted research time.

Mohnish Pabrai explains:

"I met Ken around 2000 or so. My wife went in as a consultant to Citadel."

"She'd come home every evening with a whole bunch of Ken stories. This guy is very unusual, and the place is very unusual."

"He had hired a whiz-bang Russian mathematician, PhD, postdoc, working on the algorithms."

"Everyone at Citadel would come to this guy with their problems, and Ken didn't want anyone coming to him. He just wanted him to crank."

"Ken told the temp: here's your desk, here's the mathematician. No one crosses."

"The temp says, “What do I do?” “Nothing.” Your whole job is to make sure no one crosses."

"The temp herself is in shock. Someone's paying me to file my nails."

"He found all the different nooks and crannies, built a tremendous business."

[Leo the Bayesian @Leothebayesian]

This evening I sat across from the CEO of a so-called trading neo lab with over $1B val. He told me their models—built on best-in-class RL environments and online learning algorithms—have produced exceptional pnl and sharpe ratios.

I later learned that the company is, in fact, a data broker LARPing as a frontier lab. Its actual edge is trading stocks and options on insider information of model releases and synthetic-data acquisitions.

It appears I have once again materially underestimated the absurdity of this industry.

[Tobias Carlisle @Greenbackd]

Each of the last several decades had a clear winner, an asset that rose so far and so persistently that it came to seem like the obvious place for money.
Each went on to disappoint in the decade that followed.

The pattern is consistent across very different assets:
* Gold rose more than thirteen-fold in the inflationary 1970s, then lost ground through the 1980s as inflation fell.
* Japanese stocks rose nearly six-fold in the 1980s, then lost half their value in the 1990s, beginning a slump that lasted decades.
* The Nasdaq rose roughly nine-fold in the 1990s, then fell 44 percent in the 2000s.
* Emerging market stocks nearly tripled in the 2000s, then barely moved in the 2010s.

In each case, the decade’s winner ended it with the widest following, the most confident forecasts, and the highest prices relative to its fundamentals.

[Michael Dell 🇺🇸 @MichaelDell]

Thanks very much @JensenHuang! My first trip to Taiwan was in 1985 when I was 20. Compeq (not Compaq 😂) made PCBs for us. This is the hand wired (Jay Bell, RIP) prototype of our first 286 based PC and the actual production motherboard. It was based on a chipset from Chips and Technologies, they were produced in Japan at that time before TSMCs rise. I explained a bunch of this history in #PlayNiceButWin.

https://t.co/FRDfrAMlu3

[Eva Fo𝕏 🦊 Claudius Nero's Legion @EvaFox]

When Elon Musk was a young guy living in Vancouver with almost no money, he walked into the employment office and asked for the best-paying job they had. It turned out to be $18 an hour to put on a hazmat suit, crawl through a narrow tunnel into a lumber mill boiler, and shovel out the hot, steaming sludge left inside.

He later called it "a Dickensian steampunk nightmare," and honestly, that sounds about right. Around 30 people started that week, and by the end of it only 3 were still showing up.

Elon was one of those 3.

When people look at him today, they see the billionaire, but I see someone who was never too proud for the hardest, dirtiest work, and I think that says more about him than any number ever could.

[AvantPrince_ @AvantPrince_]

"The Long Road: Men are invisible until they become successful and so the path to success is a lonely one, filled with setbacks and suffering he learns to thrive alone. This is why when a man finally arrives, and the world suddenly decides it wants to know him, because now and only now does he merit consideration - you can understand why he’d be cynical. And he’s right to be. Because the only people that can ever truly mean something to a man are the ones who will stick by him when he has nothing, not the ones who appeared when he became something. For most men like him that means a fairly solitary life, but he’s comfortable with that. You never wanted him, and now he doesn’t need you - he learned to thrive alone. When he needed you, you didn’t want him, but now you want him, he doesn’t need you. Some of you reading this will think I’m projecting or speaking for myself, but I’m not. I’m speaking for all the silent unheard men who had to struggle alone for years on end in the rain and in the heat, in the dark and in the cold, on gloomy days and on stressful ones, without encouragement, recognition or even as much as a compliment - the men who were ignored, passed over, berated and neglected but soldiered on anyway in their refusal to let a cold and unforgiving world that pretends to be more loving than it is beat them down, but instead cultivated an enduring spirit of good humour and aspiration that knows better than to expect something for nothing from anyone - care and compassion especially. So forgive those self-made successful men if they seem a little detached, a little ruthless, a little neglectful - but you haven’t made the journey they’ve made, and so suddenly showering them with the care and concern they were always denied because you finally see them as people isn’t going to make them forget. They’ve been conditioned not to need you, and so your noisy compassion is a luxury at best, at worst, an irritation. Never let it be said then that a man is not a success object, because that’s all he is. And be prudent in knowing the sentimental will deny this most deepest of truths because they have no stomach for it, and yet do not allow their foolish bleating to mislead you. The sooner you embrace this reality, the sooner you will make peace with it and start to thrive in spite of it, or perhaps somewhat poetically, because of it."

Anyone tell me who wrote this?