The demand for AI is on an exponential, we need to continue investing at every layer of the stack if we are going to be able to service the demand for AI compute efficiently and at humanity scale. TPU's are so wonderful : )
HARDWARE
-
Open-Source Quadruped Robot: From CGI to Garage Build
By
–
wow this is soooo fun!
— Thomas Wolf (@Thom_Wolf) 9 avril 2025
CGI for now but really not far from current robot capabilities –though battery life is a challenge with 150 lb sitting on it–
now we need this as open-source hardware + software so I can build one in my garage pic.twitter.com/QzOMZkBLf3wow this is soooo fun! CGI for now but really not far from current robot capabilities –though battery life is a challenge with 150 lb sitting on it– now we need this as open-source hardware + software so I can build one in my garage
-
China Trade Impact: iPhone Prices Expected to Double Globally
By
–
You'll be fine? iPhones will 2x in prices, so will much likely 95% of all material good on your shelves. Since China basically provides the world with most made goods.
-
US tariffs make indie hardware manufacturing 104% more expensive
By
–
Hard time being a indie maker in the US now. PCBway, Banggoods etc. Everything just got 104% more expensive.
-
Single GPU Inference Performance vs. Unified Memory for AI
By
–
Gemini n’arrête pas d’exploser tous les benchmarks, et ça continue une semaine après la sortie. Sans aucun doute, la meilleure IA aujourd’hui pour le code.
-
People are finally buying GPUs
By
–
J'ai vu ce tableau aussi. Je vais en faire une vidéo demain, beaucoup de choses ne vont pas avec llama 4
-
Cerebras Technology Simplifies Large-Scale AI Model Training at EPCC
By
–
Cerebras technology will eliminate the need for complex parallel programming, making it easier and more accessible for scientists and ML practitioners from every discipline to train and use powerful AI models. EPCC can now: Train models from 240B to 1T parameters
-
SambaNova Launches Llama 4 Scout at Record Inference Speed
By
–
Herding in a new era of the FASTEST AI inference As an official @AIatMeta launch partner, we're bringing #Llama4 models to life on SambaNova Cloud, powered by RDUs! Llama 4 Scout is available now, running at 697 t/s Llama 4 Maverick coming soon Read more
-
Blocks New Bitcoin Miners Impact US Supply Chain
By
–
Also makes the case for @blocks new miners for US miners. I do wonder how their chip and hashboard supply chain will be affected.