Yes.

Hardware rarely stops you from doing machine learning.

Most traditional machine learning never touches a GPU

You can run it on more or less any modern device, from a phone to the beige office laptop your employer handed you.

Libraries like scikit-learn were built to be efficient on CPU architectures, and it shows. Classification, regression, clustering and basic feature engineering all run without a graphics card anywhere in sight.

Deep learning on consumer silicon

Deep learning really did need more compute.

Then Apple's M1 chips picked up TensorFlow support and a MacBook became a deep learning machine. Slow next to a datacentre, fast enough to learn on. The M1 iMac isn't a laptop, granted, but it's the same silicon, and it makes the point about what consumer hardware can now do.

I train on my own MacBook and it's fine for smaller models. The long runs still go to a dedicated cluster, because otherwise the laptop is locked up for hours.

Start on what you already own

Experiment on the machine in front of you. Learn the fundamentals without buying an expensive GPU first. Start with small datasets and small models, because that's where you find out whether your approach was ever going to work. Scale up only when something forces you to.

You won't train a state-of-the-art large language model on a laptop.

You also don't need to.