AI Needs Power: lots of it

I was reading a blog post, and this tiny bit stood out:

Kaushik Roy of Purdue University compared the power consumption of Deep Blue for chess (15kW), Watson for Jeopardy (200kW), and AlphaGo for Go (300kW) to show that matching human behavior in games does not come easy.

It would be interesting to see how the rise of AI/ML will impact the energy consumption at data centers and in general. I wonder if we are all thinking about the power needs of software-driven, silicon-optimized future deeply enough. Steve says that human brain needs about 20 watts of energy and it uses an interesting cooling system. It is also a good reason why we need to take breaks from activities — social media in specific — which cause us to use up too much energy.

Read article on ARM Community