Explosive growth in AI, Martin Hosken (VMware) shares his insights

They discuss topics including Sovereign Cloud and Private AI. What exactly is a Sovereign Cloud? How does VMware distinguish itself in this area? And what is VMware doing in the field of AI? Larik-Jan and Martin discuss it in the video below.

Martin, what are the topics you are seeing most often at the moment?

‍Martin: “We see that there is a lot of talk about Sovereign Cloud and the impact it’s having on the market. VMware is more or less the leader in this entire space, and even Azure, Microsoft, Google Cloud, Oracle, and AWS are all trying to catch up now. But there are also challenges in this area, for example regarding the definition of Sovereign Cloud. There is no clear definition. As a result, we see that partners, as well as competitors, are all using it differently.”

What is the difference between VMware and those other parties when we talk about Sovereign Cloud?

‍Martin: “At VMware, it’s not just about data residency, but specifically about control. VMware is the only Cloud partner that helps customers and partners gain jurisdictional control over their data. This is because VMware does not own or manage its own Clouds. Our partners own and manage the Clouds within their jurisdiction and therefore have full jurisdictional control over their data, including for their customers. That is unique from a Sovereign Cloud perspective. We live in an era of hype, and that is currently the case with Sovereign Cloud; everyone wants to jump on the bandwagon. You can’t afford to be left behind, or you become irrelevant.” “The European Commission largely aligns with our thinking on what Sovereign Cloud is. But it would be good if they made a statement like: ‘this is how we see the Sovereign Cloud.’ That would also help us tremendously. They have been talking about it for a while, so perhaps it’s coming.”

What is VMware doing in the field of AI?

‍Martin: “In Las Vegas, we announced our ‘Private AI Initiative.’ This is about running your dataset iteration locally where the data resides. We have many companies and customers who collect data in their own data centers and then move this data to Public Clouds to perform analyses: AI-based analyses, machine learning, and so on. That is not efficient, both from a security and privacy standpoint, as well as from a time and cost perspective. The ability to perform AI analyses and machine learning on-prem or at the Edge is much more attractive. Thousands of VMware customers are already running these AI predictions; running these types of analyses based on VMware is not new. However, AI has really exploded in recent months, and this type of AI is very data-intensive. So, VMware is building Private AI, which is essentially VMware Cloud with a stack of various components you need, both open source and Nvidia proprietary drivers (or other GPU vendors). There isn’t just one architecture, but by giving customers a standard approach and a reference architecture, developers have a starting point.”

What are the challenges here?

‍Martin: “Obtaining the hardware/GPUs is a real challenge. At the moment, we are building it for customers who are already working on this. They already have the hardware and the skills, but they were just missing a standardized approach to get started.”

What developments are you seeing at the moment?

‍Martin: “We are seeing a growth in the number of workloads that were moved to the public Cloud in recent years being brought back. In some cases on-prem, but also to private infrastructure in colocation. At VMware, we have the right tooling to help customers with the migration back to VMware, but we are also building additional functionalities to further simplify that process. We call this ‘Cloud Smart’: adding intelligence to the decisions you make. I’ve said this before, but sometimes it’s just okay to change your mind and go in a different direction. If you decide to move something back because it isn’t working well, for example, you shouldn’t see that as a failure. You should see it as a learning process.” “That doesn’t change the fact that migrating between Clouds is still not easy. We have calculated that placing an application in the public Cloud costs around 600 to 800 dollars, and if you want to move your application out again, it costs around 3,600 dollars. This is because when you migrate to the public Cloud, it’s more of an infrastructure play; you are migrating from one infrastructure to another. But when you want to leave the public Cloud, you have to think about the application layout. In some cases, it is even faster to rebuild something than to migrate it.”