Sourkraut

equities. certified public tax advisor (Germany). master of international taxation (MiTax). treaties for the avoidance of double taxation and more such shit.

[Sourkraut @AccountantReal]

Das ist schlecht! Wer redet von einvernehmlicher Untreue🤡
https://t.co/1bMCzLPkuF

[will depue @willdepue]

I have a major result to share.

I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.

This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.

Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.

I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.

I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.

It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.

These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.

I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.

There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

[Fireside Alpha @firesidealpha]

Dylan Patel says Nvidia’s next CEO has to hold off OpenAI, Broadcom, Google, Anthropic, Meta, Amazon, and AMD chip teams, who are all 'fucking good'

“It’s not like you’re fighting, no knocks against Palo Alto Networks, but their competitors are not that amazing, versus Nvidia’s fighting an OpenAI chip team, a Broadcom chip team, a Google chip team, an Anthropic chip team, a Meta chip team, Amazon chip team. These guys are fucking good. AMD. They’re all really good, and all these startups.”

“They’re all good. It’s not like they’re incompetent. I think it’s very different. As great as Lisa Su is, she took down Intel, who’s incompetent.”

“What is the person who defined a category and created it and kept all the invaders at bay? It’s very different than being second place or making smart tactical business decisions, innovating on something once there’s already a category established. It’s just very different.”

[Sourkraut @AccountantReal]

@WealthCoachMak RennTech sold 99,7% in Q3. Bought AMD, being the 2nd largest position for latest 13F. However Renntech is run by fast acting machines...

[Sourkraut @AccountantReal]

@snicklink Da fehlt mindestens eine

[Sourkraut @AccountantReal]

@mooreslawisdead Yes he might, but this would mean he was never Jensen The Moat Huang. He was just a very stupid littlw guy woho destroyed everything on purpose

[Sourkraut @AccountantReal]

@TichysEinblick Nvidia hat einen visionären Gründeringenieur durch einen Finanzchef ersetzt. Kauft AMD.

[Jesse Cohen @JesseCohenInv]

Jensen Huang walking in to Nvidia offices this morning and addressing employees

[Sourkraut @AccountantReal]

@TDrygala @MooreDividends Er ist so wichtig, dass sehr mächtige Personen, die Medien und vielleicht sogar das Board Meeting steuern können tätig wurden.

[Sourkraut @AccountantReal]

@FraukePetry Glaubt jemand der Kampf darum, wer als erstes die Allgemeine KI hat wird mit der Entfernung von Huang zuende sein.
Armer Nvidia CEO hätte er sich besser an Epstein gewendet, wie Gates und Clinton.

[Sourkraut @AccountantReal]

@goldi__of @mario_liccardi_

[Sourkraut @AccountantReal]

Vorsorgebeiträge orientieren sich häufig am Nettoeinkommen. Wenn der Steuerbescheid aber nicht rechtzeitig vorliegt droht eine Schätzung mit hohen Beiträgen.
Der Beitragszahler darf dann die gesetzlichen Krankenkassen um Hilfe bitten (toll!).
Was zwischen (ausländischen) Banken und FiAmt funktioniert, funktioniert natürlich nicht bei gesetzlichen Krankenkassen oder ähnlich.

[Coach Mak | Know Your Money @WealthCoachMak]

$TSLA $AAPL $MSFT $META $AMZN $GOOGL $AMD $NVDA $AVGO $MU

1 is not like the others

[Sourkraut @AccountantReal]

@WohlstandsWal

[Sourkraut @AccountantReal]

Gekauft

[Sourkraut @AccountantReal]

@sagjetztendlich Schlau

[Sourkraut @AccountantReal]

@WealthCoachMak Same Boat

[Not Jerome Powell @alifarhat79]

The internet is undefeated

[Sourkraut @AccountantReal]

@RZitelmann Mentale Selbstbefriedigung. Mindestens 1/3 aller Post fällt darunter. Einige hams gern hart andere lieber zart.

[TastefulLindy @LindyTasteful]

One thing you have to like about AMD.

You dont have to worry about a Chinese spy being sent to seduce their CEO