Pablo Sanchez

Apple built 19 to 21 million iPhone Air units. Most of them are still in the box. CIRP data says 6% of US buyers chose one. Suppliers cut production over 80% within two months of launch. Foxconn dismantled all but one and a half assembly lines by November. In China, Apple slashed the price 30%, dropping 2,500 yuan off a phone that launched at 7,999, just to move inventory. The BOM on a comparable iPhone runs around $485 per unit. Scale that across millions of unactivated devices and Apple is carrying more dead inventory on a single SKU than most hardware companies generate in total annual revenue. Samsung saw the same data. The S25 Edge sold 1.31 million units while the S25 Ultra sold 12.18 million. Samsung canceled the S26 Edge before production started. The entire thin phone category died in one product cycle. The Air was expected to capture 10 to 15% of the iPhone 17 lineup and prove Apple could build at 5.6mm before shipping a foldable. Instead it proved that consumers will pay $999 for the best camera or $599 for the cheapest iPhone, and almost nothing for what sits between them. Every sealed box in this photo is a $485 engineering project that answered a question nobody asked.

Excelente servicio Brubank. Sábado, todos los bancos cerrados para comprar dolares, menos el gran Bru. Hice una pruebita y funcionó perfecto. Al que le sirva el dato, feriados y fines de semana, ahí lo tiene. Otra cosa, anda muy bien en el exterior también (débito y crédito).

craziest story you'll read today. they turned human cells into an LLM. > scientists grew 200,000 human cells on an AI chip and taught them how to play DOOM on a computer. demo below is 100% real. > the cells learn, adapt and self-organize with (almost) ZERO energy required (human brain runs on 20 watts) > why do this? because cells adapt WAY QUICKER (18X) than silicon and use less energy. > it gets better - team has started growing cells on AI models, cells choose which word the AI generates next > the device that fuses the cells to silicon costs $35K and leverages a multi-modal array if we can figure out how to scale this then we unlock a completely diff chip design architecture and MUCH cheaper way to use LLMs.

Introducing the Claude Marketplace, a way for enterprises to simplify their procurement of AI tools. Now in limited preview.