Essentials for AI Infrastructure—the AI in Business Podcast with AB Periasamy and Matthew DeMello

Essentials for AI Infrastructure—the AI in Business Podcast with AB Periasamy and Matthew DeMello

MinIO’s co-founder and CEO AB Periasamy was recently featured on the AI in Business Podcast where he had a rich conversation with Matthew DeMello—Senior Editor at Emerj—about AI infrastructure and object storage for enterprises. 

In this blog post, we take you through an abridged version of what was discussed. Let’s get into it. 

AB and Matthew begin with a conversation about the importance of education when it comes to helping businesses scale their AI infrastructure. AB explains that the heart of the latest AI breakthrough is what our human brains have done from very early on: capturing thoughts as multidimensional vectors and communicating and analyzing very effectively. The AI boom is what AB describes as the “greatest breakthrough of our lifetime.” AI will impact everyone, which is why education around it is critical. 

Generative AI has been a major point of conversation lately. Up until this point, we have been building up our so-called ‘first generation’ capabilities, which are comprised of machine learning, predictive analytics, etc. Now, we are seeing enterprises looking to transition to generative AI, the ‘second generation.’ AB explains that these changes are not going to happen overnight. We are talking about three to five years with a clear five-year plan. AB states, “The tech part, I wouldn’t worry so much about… if you have the need, the use case, the plan. Everyone will have equal access to technology… what they will not have, are the people and the data.”

This is a critical point. Data is the most important aspect of your business; it’s the asset that will differentiate them from everyone else, and the more you have the better off you are. The more you capture, the better off you are. The greater the diversity of that data (which feeds improved generalization, enhanced robustness, comprehensive learning, adaptability and applicability) the better off you are. 

But data requires people too. And to have the right people you need the right culture. People have a massive role to play in the AI revolution. 

Transitioning from a legacy culture to a data-based culture requires empowering the right people and transforming them into leaders. At the end of the day, the culture has to come from the top down. Look for the accountable people, the problem solvers, the clear thinkers; these are the folks that will really move the needle. It doesn’t matter how much someone already knows, but how much they are willing to learn. 

The conversation then transitioned to the topic de jure—what about the GPU scarcity challenge? AB’s take is thoughtful and far-sighted—time is solving this problem on the availability front and competition will solve it on the pricing front. Why? We understand the need for computing, and we have the use cases. The value that GPUs are generating is well understood, and markets are efficient and will reduce the excess rent that Nvidia is earning today. That doesn’t mean that GPUs are not the long-term answer—they are. In two to three years, GPUs will be the heart of computing. Nvidia is there today and thus they are succeeding in ways that few companies ever have. It is a new compute category and they own it, but over time AMD and Intel will have a role to play too.

AB and Matthew closed with a discussion about MinIO, AWS and how the world of infrastructure has changed. Amazon simplified everything by removing antiquated software and modernizing data storage with S3—creating a standard for modern data infrastructure. MinIO believed this technology was so important and couldn’t solely be in the hands of one provider, which led to MinIO’s founding. Now, through MinIO, the cloud operating model is available everywhere from the public cloud to the private cloud and all the way to the edge. Ironically, AWS uses MinIO as an example of what to run when they can’t.  

In the words of AB, “Anyone who is still in the old school enterprise infrastructure… they are in trouble.”

Previous Post Next Post