Laptop recommendations for machine learning

Hello, I am currently pursuing my Bachelor’s in artificial intelligence and machine learning, the laptop I am using is pretty old, I am currently looking to buy a new laptop, I am very confused in whether to buy a normal laptop like a person would for software development or since I am doing bachelors in AI and also enrolled in the MLS course and looking to make my career in the field of AI should I look for those laptops which have graphic cards, namely gaming laptops?

I have googled and watched many youtube videos, half of them says you need a high performance gaming laptop and half of them says you don’t need an expensive laptop for machine learning since you won’t be training anything big on your laptop anyway. I’ve seen many of my colleagues being confused about this.

I’d love to hear your take on this, and whether I should get a normal laptop or an expensive gaming one. Especially if you are someone who works in this field, and uses laptop for work, what laptop do you use?

1 Like

Hi, @spignelon !

It really depends on your future plans.
Working with colab / kaggle remotely is good for introductory learning, so you can fully focus on code, good programming praxis, etc. Nonetheless, you will be absolutely dependent on this platforms, as local training on cpu is prohibitively time-consuming.

That being said, and from my experience, having an own machine that you can fully customize the way you want will give you a good insight into how DL/ML hardware-software works. Getting the grasp of the full stack will give you a good confidence for the future challenges you will deal with in your studies / work.

if you are someone who works in this field, and uses laptop for work, what laptop do you use?

I recommend going for a custom desktop setup. Having enough room for ventilation and customization is something your graphic card temperature sensor will thank you for (if you don’t need it to be portable). Anyway, for a laptop, some of the cheapest 3000 NVIDIA series are good enough.


In my opinion this is correct. Unless you are setting up a server farm at home, don’t plan on doing anything approaching production level amounts of data. That said, having a machine you can run at least some data through without having to pay for xPU is handy. The technology you want to use may influence platform choice; trax for NLP, for example, isn’t widely supported. Also, resign yourself to obsolescence; what works today or is the best bang for $ will change quickly. Being agile in your computing platform is just the cost of keeping up in this business.