Actually, the meaning of this post is huge!: Deepseek is the singularity it predicts (the cost and threshold of high-performance LLM models are greatly reduced), which will enable terminal devices/edge devices to carry AI and step onto the historical stage of daily life! I'm $appl, and I'm considering $amd, my personal thoughts are not investment advice! PS: @nake13 Panlaoshi's use of the word 'singularity' is amazing, YYDS! twitter.com/alexocheema/status...

As for the priority choice $appl, it is because Apple has many terminal devices. After the emergence of the singularity, the direction of LLM and LLM distillation is to reduce the hardware requirements, rather than to work miracles with force! Maybe in the future, iPhone, Mac, iPad and even Watch can run the distilled DeepSeek model.

$amd This wave may flip $intc (Intel): 1. The emergence of deepseek makes it possible to use NVIDIA directly for civilian unified integrated solutions. 2. In addition to the unified integrated solution of Apple's m series, AMD and Intel integrated CPUs including NPU + integrated memory in CPU + super graphics core. 3. In the unified solution of thin and light notebooks in 2025, $amd's AI 300 solution directly kills Intel

The r1 70b distilled version requires 43GB of space. If a 64G Mac can run it, as a small home device, the running score is between the full r1 (671b) and 30b, isn't it equivalent to @OpenAI's O1 commercial version? Isn't this terrifying! Whether you believe it or not, I believe the singularity has arrived!

www.youtube.com/watch?v=IYbMyv... Treasure video, re-watching it a year ago, now I know how awesome Apple's unified architecture is. Low power consumption not only allows for longer standby time, but also makes local reasoning for home private AI right in front of us.