Ryan Orbuch
5K posts

Ryan Orbuch
@orbuch
unlearned helplessness | partner @lowercarbon, previously carbon removal @stripe





Congrats to @stripepress and @tamarawinter on selling 1M books!! That’s a lotta books!! 📚


My first interview with Ian Brooke (@ianbrooke), founder & CEO of @AstroMechanica. 0:31 Creating Long-Range Supersonic Planes 3:07 Private Jets vs Airliners 8:26 How Insanely Expensive Private Jets Actually Are 11:30 Building Great Products That Last Forever 20:39 Front Loading Work To Design Things Well 30:42 Making Contact With Reality 43:32 Following Your Intuition 48:38 Thinking In Shapes 53:54 Expressing Yourself Through Building Things 58:56 Starting With A Vague Idea & Sharpening The Vision Over Time 1:03:16 Trying To Tell A Story vs Just Building Something 1:17:44 Seed & Soil - Emmett Shear 1:27:57 Doing Things That Bring You Joy & Energy 1:32:35 Keeping The Product Vision In Your Head




Google publishes a paper showing that its AI models only use 0.26 mL of water in data centers per prompt. After, this article gets published: "Google says a typical AI prompt only uses 5 drops of water - experts say that's misleading." The reason the expert says this is misleading? They didn't include the water used in the nearby power plant to generate electricity. The expert, Shaolei Ren says: “They’re just hiding the critical information. This really spreads the wrong message to the world.” Each prompt uses about 0.3 Wh in the data center. To generate that much electricity, power plants need (at most) 2.50 mL of water. That raises the total water cost per prompt to 2.76 mL. 2.76 mL is 0.0001% of the average American lifestyle's daily consumptive use of fresh water and groundwater. It's nothing. Would you know this from the headline, or the quote? Why do so many reporters on this topic do this?


















