Dhamar on Nostr: AI companies now designing own chips to cut costs, and tightly optimize hardware for ...

AI companies now designing own chips to cut costs, and tightly optimize hardware for their specific models and inference workloads, OpenAI’s Jalapeño is the latest high profile example

meanwhele, biggest pure chip company, Nvidia all in on releasing strong open weight models, such as Nemotron & others bla bla bla

eaaaaa! its classic full stack and ecosystem strategy in both directions, AI labs want to own more stack, less hostage to GPU pricing and supply, chips makers want competitive open models that run best on their hardware, keeping developers and data centers locked into their platforms even as closed frontier models multiply

everyone is moving up or down the stack 🤝 we’re friend in a multimodal bubble

quoting
nevent1q…xeqk

so prompt today: ok bots, let’s creat a chips🧅🥔