AI Czar David Sacks Explains the DeepSeek Freak Out

Chamath's point on American VCs showering founders with money is very important. Anyone who's followed private markets for the past 4-5 years can recognize the trend and the necessity to correct it.

Chamath is absolutely 💯 spot on. Agree. We created our proprietary claims language model even before openAI came out with ChatGPT for costs similar to actually less than DeepSeek right here in Silicon Valley. But investors want the hype, brash entrepreneurs, and flashy startups to invest millions and billions. You’re giving too much money to startups that don’t have a sustainable business model, and a good product in much needed industries and spaces. VCs keep giving me this theory that it has to be horizontal but nothing about LLMs are proprietary so anyone can come up with various other models more cost effectively down the road. It’s in the small domain models and implementations and open source libraries for middleware and some software around LLms that has intrinsically more value.

There's a simple creativity maxim here: More limitations lead to more innovations.

The biggest impact is that you can now run a good AI model in a desktop computer, with no internet connection. That lets companies and people worldwide access AI that required US company involvement previously. That fact reduces American exceptionality in this area, and will end up driving investment money more worldwide.

Chamath ‘S point reminds me of Ernest Rutherford’s dicta to a British Science Policy committee after WW 1 “ We havent the money, so we will have to THINK”

Well said. When US failed to compete, National security and ideology are their ways to divert

So many important insights in this episode. Thanks for the help in breaking through some of the noise.

Excellent analysis.

Great insights!

Thank you.