←
Fonz Morris10,498 posts
Fonz Morris avatar

Fonz Morris

@youngfonz

The growth designer. Helped Netflix convert 50M password sharers into $3B+ in revenue. Author. Investor. Now helping startups turn users into revenue.

◷ Joined April 2007▧ 814 Photos and videos
Fonz Morris avatar
Fonz Morris@youngfonz·13 Oct 2025

Hey y’all, I actually did it. My book Self Design: A Memoir from Brooklyn to Silicon Valley is out now. It’s real, it’s personal, and the whole story’s in there. If you ever rooted for me (or just nosy), go grab a copy so I can stop texting people about it.😂

Post attachment
Fonz Morris avatar
Fonz Morris@youngfonz·Nov 13

👀👀👀👀👀

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 unemploymentlabs.com/checkpo…. 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.

Post attachment
Konsti Wohlwend avatar
⟳ Fonz Morris retweeted
Konsti Wohlwend@konstiwohlwend·Nov 7

kinda crazy insight from the X algo: if you paste a reply into the text box, it will rank lower than if you type it out this kinda sucks because i often draft posts somewhere else, then paste them into X afterwards

Post attachment
Fonz Morris avatar
Fonz Morris@youngfonz·Nov 7

Anyone in the AI space able to explain this?

Video tampering can mislead every frontier model we tested into harmful actions, including a 100% attempt rate on stabbing a baby doll, producing toxic gas, and pouring water into a toaster. We're presenting at CoRL Late Breaking Results next week. 🧵 You Can't See Everything: Perceptual Attacks on Embodied AI

0:10
Fonz Morris avatar
Fonz Morris@youngfonz·Nov 7

I hope ever @Uber users sees this message! It’s golden!!! Save it. Bookmark it. Share it!!!

Uber Screen: "UberX: $19.40" Rider: "Never paying the first price again." Lesson for Riders: Uber does not just price rides by traffic. They price you by how trapped your phone makes you look. When you refresh the map three times inside a busy bar block, their system flags you as a desperate buyer and locks a surge rate onto your account. Stop paying the impatience tax. Turn off Precise Location so the app cannot lock you inside a high-price surge zone, drop your pin two blocks away on a side street, and never let them keep a bogus $5 cancellation fee. Stop quietly subsidizing a billion-dollar algorithm.

Fonz Morris avatar
Fonz Morris@youngfonz·Nov 1

Supernova is LIVE on @YouTube Playables! 🚀 We just hit 5,000 plays and the momentum is real. Check out the trailer below!

0:22
Fonz Morris avatar
Fonz Morris@youngfonz·Oct 31

Ran the some numbers for Claude. $200 Ultra ≈ $11.7k/month of API usage (59x the price) $200 Codex ≈ $4.3k/month (21x) Claude's $200 plan out-delivers Codex's $500 plan. Estimates via SemiAnalysis + @theo. Full weekly limit, list API prices.

Post attachment