As AI gets more expensive, popularity of Chinese AI models like DeepSeek is growing fast, even though these models are not as good as ChatGPT, Claude and Gemini. Industry experts believe the cost pressure is real and most companies will move to a hybrid model for AI use.
Do lightweight models such as DeepSeek still require a server room filled with GPUs to run?
I’m speaking in the context of AI-assisted coding. A loop of feedbacks, hooks, and self-reflections to reduce errors and bugs generated by AIs.
This kind of automation is somewhat effective time wise but extremely costly in computation, memory, hardware, and energy.
I guess as long as we rely on GPUs, NPUs, TPUs, or any kind of silicon chips, the dream of democratizing (the means of production of) AI is still far fetched.
You can run Qwen 3.6 27b on a laptop with less than 32gb ram now, and it does a surprisingly decent job at coding tasks. There are also tools like this which dramatically improve local model capability https://github.com/itigges22/ATLAS