HOLY F*CK. a user reports a bug. by 3 AM Lauren Tan's agents have fixed it and shipped it. nobody on the team even noticed
SpaceXAI engineer:
"most people still test every line their agent writes by hand. that's exactly why they can't ship 4,000 updates a month"
in a 37-minute talk she shows how she taught her agents to run their own code like a real user and check their own work
Agent Writes It - Runs It Like A User - Catches Its Own Bugs - Fixes Them - Merges It Itself
better than any $5,000 coding bootcamp
skip netflix, watch it tonight, then tell me - who tests your agent's code right now?
Ian Buck on how Jensen hired him:
"I started a project at Stanford called Brook, which was to explore general computing on GPUs. We open-sourced it and got over 10k users, and then Jensen invited me to come to $NVDA and do it for real."
Brook later evolved into CUDA.
Vision.
@officialAroleap Wrong timing. If this had launched during COVID, it would’ve been a massive hit.
People don't just go to the gym to work out, they go for the environment and the social energy.
@lordmarme@kentakeuchi2003 In the long run, Nvidia could face a steeper cliff than Intel. Intel has operated without a founder-CEO since the Andy Grove era, leading to a slow erosion. Nvidia’s decline will take a few quarters to materialize, making it slower to spot—but potentially much worse.
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.