NEW:
Forget Alice Huang's role at Megaspeed, that's almost a distraction.
The real story is her role at China-based Shanghai Century Litong, which became a "preferred" Nvidia cloud partner in mid-2023 and facilitated >$7.5b in Nvidia chip shipments, incl securing BANNED chips.
Saaspocalypse update: Anthropic's launch of Claude for Office hit MSFT stock earlier this year, but internal data shows it's not gaining on Copilot in Office (at least for now).
Plus, new scoopy details on what MSFT is announcing at Ignite tomorrow: https://t.co/xTGt3Eosji
Scoop: SpaceX has axed the team (roughly 60 people) behind AI video generation tool Grok Imagine. It's still live for now, but the company is shifting resources elsewhere
My first story on the Elon beat, with @claudiackm and @JuroOsawa.
https://t.co/vfjJJeuTdd
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.
fmr Nvidia manager told me that Jensen told them & others that he planned to run the company for another 60 years, so no succession planning needed.
hence why there is nothing resembling a No. 2 exec at what was the world's second most valuable company...
👀👀Jensen departure is terrible news for daughter Madison Huang, a rising star at Nvidia who has clashed with CFO Colette Kress (who is now interim CEO)
Jensen ruled with an iron fist but Kress frequently pushed back on Jensen's most aggressive dealmaking
more new facts about what happened re Nvidia:
Chinese companies with ties to Jensen mistress Alice Huang ~~invested in DeepSeek~~ with a plan to supply DeepSeek with access to (banned) Nvidia chips
just a reminder
that a key potential smoking gun is Alice Huang's connection to a network of Chinese firms, including Shanghai Century Litong, that facilitated shipment of $7.5b of advanced Nvidia chips to... China.
NEW:
Forget Alice Huang's role at Megaspeed, that's almost a distraction.
The real story is her role at China-based Shanghai Century Litong, which became a "preferred" Nvidia cloud partner in mid-2023 and facilitated >$7.5b in Nvidia chip shipments, incl securing BANNED chips.
Nvidia-related news:
U.S. commerce dept set to close a longtime loophole that allowed Chinese AI companies to train AI using the most advanced U.S. chips.
Nvidia fought against closing the loophole...