($) Do We Need Sovereign AI?
The more I think about it, the more I think we do.
All eyes are on this year’s Nvidia GTC conference – the AI juggernaut’s annual product and customer event. Typically, this event is already a big deal in the industry – drawing big crowds, Wall Street attention, flashy sponsorships, and lots of parties. But this year, it will receive more attention than ever before. The whole world, literally, will be watching and hanging on every word Jensen Huang utters.
That’s because of one term: Sovereign AI. (If you are playing “GTC keynote bingo” or a drinking game, I hope you have this term in your game.)
Sovereign AI is basically a made up term. The only comprehensive write-up of this concept is, unsurprisingly, a Nvidia blog post, published just a few weeks ago. Yet, since the ChatGPT “big bang” moment 15 months ago and Nvidia’s meteoric rise since, Huang has willed this made-up concept into existence.
I was initially skeptical of it all. It appears too nakedly self serving. If every sovereign entity in the world builds its own AI infrastructure, instead of leveraging the economies of scale of a global cloud computing network, the largest commercial beneficiary of this extra capacity would be Nvidia. How convenient!
Then, I started thinking deeper about the many issues that currently vexes generative AI, like hallucination and data governance. I began examining the global AI talent flow in the context of sovereign AI in a recent post.
The more I think about it, the more I’m coming around it. The world actually does need some sovereign AI. Here’s why.