Building the Next Generation of AI Infrastructure

Chapter 2. Compliance, Security, and Sustainability Are Not Optional Anymore

One of the main concerns of AI adopters and learners is the difficulty of keeping up with all the constant news and technology advancements. This includes the constant feed of AI models reaching the first positions of the technical benchmarks, new startups being created every week, the endless release of promising AI platforms, and the myriad protocols and frameworks that take over the social media feeds on a weekly basis.

With that evolving context, it is easy to feel confused and even overwhelmed. However, there is some good news you may want to hear, as it will impact how you implement and evolve your AI infrastructure:

• Your AI technology stack and the overall strategy don’t rely on specific models or product announcements. They should and will be solid enough to observe all these developments and add relevant pieces when—or if—they are necessary. For example, if there is a new framework (e.g., A2A, MCP, Microsoft Agent Framework) and your team wants to adopt it, you may choose to test it first and then decide if you want to incorporate it into the development and platform layers.

• Some pieces of the AI puzzle are already there. Some years ago, you may have had to implement your AI infrastructure and related projects without certainty of what AI compliance or even proper security meant. That has changed a lot during the last few years. And that is good news for enterprise-level AI adoption. You ...