For all of my work, I use a Dell XPS 17 9700 with an Intel i9-10885H 2.4 GHz CPU, an NVIDIA GeForce RTX 2060 GPU, 64 GB of DDR4-2933 memory, and a Samsung 981a NVMe 2TB SSD. It’s a great laptop for both general computing tasks as well as software development, data science, AI, and ML.
It’s definitely more powerful than you would need for most basic AI/ML tasks. However, a laptop like this with a powerful multi-core CPU and GPU is quite valuable when you start training deep learning and reinforcement learning models.
If I need something more powerful, then I use a data science virtual machine on Microsoft Azure. I can scale it up or scale it down as necessary. It essentially gives me low-cost access to a GPU supercomputer if/when I need it.
I only have to pay for the actual compute power when it’s running. Then, once my model has been trained, I spin down the virtual machine and it doesn’t cost me anything when I’m not using it. Just be sure you don’t accidentally leave it running — the cost can add up quickly!
This combination of physical and virtual hardware has worked well for every project I’ve worked on so far.