Live from IBM Think 2026 in Boston, RedMonk’s co-founders James Governor and Stephen O’Grady share their key takeaways from the event. They discuss the growing case for matching the right model to the right task, noting that enterprises won’t sustain the “token-maxxing” mentality of spending thousands per day on frontier models. They unpack IBM’s AI operating model as a useful framework for organizing the sprawl of projects, libraries, and components shaping enterprise AI. They also highlight quantum computing’s emerging real-world impact, including Cleveland Clinic’s work with IBM quantum systems to advance biomedicine.

This RedMonk Quick Take video is sponsored by IBM.

Transcript

James Governor (00:02)
This is James Governor, Stephen O’Grady from RedMonk. We are here at IBM Think 2026 in lovely Boston. And yeah, we’ve been here for a couple of days. This is a quick take about our reckons, what we think about the event, what are the key takeaways. So Steve, over to you. What are your key takeaways?

Stephen O’Grady (00:17)
So there’s a couple things I think. So one of the ones that jumps out to me is the sort of notion of models, right? So this idea that when we think about the industry today, so much of the time and attention from a press standpoint goes to these frontier models, The Claudes and the ChatGPTs of the world. And they’re great. They’re enormously capable. They can do sort of all sorts of incredible things. But one of the things that has also occurred over time is that the open model’s been getting better and better and better. And for some tasks, depending on what you want to do, you might not sort of be able to leverage that, or you might not want to.

to leverage larger model, because you want to train it to do something small and discreet. So I think one of the interesting conversations that has been happening here is this idea of, there’s room for lots of different models depending on what you’re trying to achieve.

James Governor (01:00)
I love that. I mean, I think the right model for the right task. Super important. I mean, as an industry, we sort of made a joke about like token maxing. You’ve got people that are like, you’re not going to make it unless you’re spending $2,000 a day on tokens. You, on the other hand, spend a lot of time waiting until your next lot.

Stephen O’Grady (01:16)
Yeah, as we joke internally, it’s the, I have to wait for my token refresh period. It’s the new, I have to wait for my code to compile.

James Governor (01:23)
can’t work until 4

p.m. or whatever it is. So basically, in an environment where enterprises are not going to be buying into this token-maxxing mentality of people spending thousands of dollars a day literally on tokens. That’s just not gonna fly if you work at a bank in the Midwest or a retailer or something like that. So from a token economics perspective, and I do think the industry’s gonna increasingly be focusing on this, the economics of tokens in terms of the models we’ve got, certainly in terms of software delivery, absolutely

that’s going to be key, right model for the right job and yeah that’s something they’ve been talking about for a bit. What else? AI operating model.

Stephen O’Grady (02:00)
Yeah, so obviously when we think about it, it’s sort of AI, right? We think about the landscape, and this is true again whether you’re using it as a developer or an enterprise, right? There’s a lot of moving pieces, right? There’s a lot of different projects, a lot of different frameworks, there’s a lot of different libraries. Everything has to come together to sort of deliver whatever it is you’re trying to deliver, right? Whether it’s something basic from code assist, you know, to AI agentic applications and so on. And, you know, one of the things that has been difficult, you know, sort of at an individual and at an enterprise level is trying to figure out, all right, what are these requisite pieces and where do they fit?

put them together and what does this look like and it’s not a solved problem by any stretch of the imagination but IBM here has been talking about the AI operational model right so they’re defining categories and they’re saying okay here’s the sort of different requisite pieces and here’s how they would serve in theory fit together and so that’s been useful I think it’s very much first steps

James Governor (02:53)
Enterprise need a lot of help. It’s

always a culture transformation. Yeah, you know this that’s one that IBM

Stephen O’Grady (02:58)
It gives you a framework in terms of how to think about it.

James Governor (03:00)
Absolutely. So that’s that’s sort of they’re saying the AI operating model needs to happen now. One thing that’s in the future is quantum computing. And I think it was really interesting is that they Cleveland Health come and they were talking about the fact that they were using quantum computing now in order to get better outcomes in terms of basically biomedicine. And you know anything that’s taken state of the art forward for medicine. I’m all about that. And you know it’s one of the first really clear use cases. And so Cleveland have worked with IBM. They’ve actually got

their own quantum computer and they’re doing that work, think that is fascinating. And I guess for me, those would be some of the key takeaways from Think 2026.

Stephen O’Grady (03:38)
Sounds good.