At IBM Think 2026 in Boston, RedMonk’s Stephen O’Grady shares key takeaways about emerging trends in enterprise AI strategy and infrastructure. Steve highlights the growing momentum of alternative AI models beyond Claude and ChatGPT. Both open-weight and open-source models are gaining traction, offering enterprises viable options at significantly lower cost points while addressing a critical concern for organizations that may not require or want to invest in frontier model pricing. A major focus of IBM’s message at the conference is the introduction of an AI operational model, which provides enterprises with a framework for understanding and structuring the complex landscape of AI components. Rather than treating libraries, frameworks, projects, and models as disconnected pieces, this approach helps organizations integrate these constituent elements into cohesive systems that meet their specific business needs. He also emphasizes IBM’s sovereign core concept, which addresses growing concerns around data sovereignty and compliance.
This RedMonk Quick Take video is sponsored by IBM.
Links
Transcript
Hi there, I’m Stephen O’Grady from RedMonk. I’m here at the IBM Think 2026 conference in Boston. A couple takeaways for me from being here for a couple days. So one of the interesting things that I’ve been talking about for a while, certainly IBM has talked about this quite a bit, is the sort of, essentially traction and progression and in some cases acceleration of alternative models, right? So when we think about AI,
So much of the time and attention in this industry goes to the frontier models, right? The Claudes and the ChatGPTs of the world, for good reason, they’re incredibly capable. But the flip side is that there are circumstances where you might not need or want a larger model. There are also circumstances where you don’t want to pay that kind of freight. And so it’s been interesting to talk to IBM here about, there are different open models, both open weight and in some cases open source.
that are able to do some of the things that you want to do at a much lower price point. So that’s interesting. One of the other pieces to me has been sort of the, I don’t know, the sort of conception and framing and sort of, I don’t know, sort of building a landscape, if you will, for AI generally. So as we all know, AI involves lots and lots of moving pieces. There’s libraries and frameworks and projects and models and all sorts of different constituent pieces we could talk about.
So the question for lot of individuals and enterprises is, right, well how do I put all this stuff together, right? How do I take these individual pieces and structure them into something that will do what I want it to do? And so IBM has begun to talk about that here as the AI operational model. Now it’s first steps, it’s not a sort of packaged, you know, of done thing, but at least it’s sort of offering enterprises in particular the opportunity to frame these and think of them as a, I don’t know, it’s sort of a,
branch and constituent pieces that together make a whole. So that’s been useful. And then the last thing is, we’ve had a couple of conversations about this here is the sovereign cloud, the sovereign core I should say. And this is sort of the notion of giving enterprises in particular and governmental institutions worldwide, the ability to run and have sovereignty and compliance and assurances around the sovereignty of a given workload. And typically we’ve had to do that sort of based on
where the actual sort of underlying infrastructure is placed geographically. And so the notion here of the sovereign core is essentially to offer that feature, but to do it in a way that is independent of the underlying platform and independent of the underlying GL. So it’s interesting. We’ll see sort of what the uptake is. But those are a couple of things that sort of I’ve noticed and taken away from, I Think, 2026.
