You all seemed to like my breakdown of DeepSeek’s technical reports. Here’s another DeepSeek thread, this time on company culture.

Did China work harder than the US? Are quants better AI researchers than techbros? Has China surpassed the US in innovation? DeepSeek is a subsidiary of High-Flyer, a Chinese hedge fund. However, they are a very new hedge fund. They are disrupters in the Chinese market.

Okay. Thanks for the nerd snipe guys. I spent the day learning exactly how DeepSeek trained at 1/30 the price, instead of working on my pitch deck. The tl;dr to everything, according to their papers: Q: How did DeepSeek get around export restrictions?

A: They didn’t. They just tinkered around with their chips to make sure they handled memory as efficiently as possibly. They lucked out, and their perfectly optimized low-level code wasn’t actually held back by chip capacity.

Okay, “how did DeepSeek get around Nvidia’s export restrictions?” Here’s a philosophy major’s thoughts on the situation.

https://twitter.com/teortaxestex/status/1883345414415581577

To understand what DeepSeek pulled off, we first have to understand what exactly the export restrictions do. Are they actually even that bad? Are the GPUs we sell to the Chinese actually that much worse than the US GPUs?

It would be extremely funny if, after all resources are added up (GPUs and electricity for AI, food and water and education for humans), the cost of AGI is exactly the same as the cost of human intelligence This is actually something I mostly expect will happen. I don’t see any prima facie reason why “intelligence per unit of resource” would be higher for carbon than silicon, in completely optimal scenarios. And humans are fairly well-optimized