2026 Predictions: Cloud & Software

Are we approaching the end of traditional SaaS as AI-native architectures and agentic platforms take center stage?

TBR Cloud & Software Senior Analyst Alex Demeule discusses which SaaS incumbents are best positioned to execute a true hard pivot toward AI, highlighting how workforce restructuring, sales realignment and platform modernization at vendors like Salesforce, SAP and Microsoft reveal early signs of strategic transformation. Additionally, this conversation explores the growing tension between general-purpose large language models and specialized small language models, digging into the economics, power constraints, and architectural trade-offs that will define enterprise-grade AI applications.

Episode highlights:

• Which vendors are positioned well for the pivot from traditional SaaS to AI strategies

• The partial displacement of large language models

• Key expectations for SAP, Microsoft and Salesforce

“We’re seeing these partnerships form where you’re bringing in domain expertise from the outside and then working on building niche models. And these niche models are going to be specialized. And they’re not going to be as capable from a general-purpose standpoint, but when you focus the training data and you focus how you’re building these models around the workflow, you’re able to get use-case-level capability that is on par with the largest models out there,” said Demeule.

Listen and learn with TBR Talks!

Submit your Key Intelligence Questions for Patrick and his guests

Connect with Patrick on LinkedIn

Learn more about TBR at ⁠⁠⁠⁠⁠https://tbri.com/⁠⁠⁠⁠

TBR Talks is produced by Technology Business Research, Inc.

Edited by Haley Demers

Music by Burty Sounds via Pixabay

Art by Amanda Hamilton Sy

2026 Predictions: Cloud & Software

TBR Talks Host Patrick Heffernan: Welcome to TBR Talks: Decoding Strategies and Ecosystems of the Globe’s Top Tech Firms. Where we talk business model disruption in the broad technology ecosystem from management consultancies to systems integrators, hyperscalers to independent software vendors, telecom operators to network and infrastructure vendors, and chip manufacturers to value-added resellers. We’ll be answering some of the key intelligence questions we’ve heard from executives and business unit leaders among the leading professional IT services and telecom vendors. 

I’m Patrick Heffernan, Principal Analyst, and today we’ll be talking about 2026 SaaS predictions with Alex Demeule, Senior Analyst for TBR’s Cloud and Software Practice. 

Who is positioned well for a hard pivot from traditional SaaS to AI strategies

Alex, thanks very much for coming in today, and you have the predictions document out there. There’s a lot in it. Congratulations. You were able to do 3 full predictions. In the services side, we only came up with two. 

Alex Demeule, TBR Senior Analyst: *laughs*

Patrick: So, you’re ahead of us on that to start. But I want to pull out a couple of things, and I had to write them down when I was reading it because it’s really, you get deep very quickly. The overall concept was, are we at the death of software as a service. That was the overall concept. But you don’t have to answer that question necessarily, but one thing you did write was about a hard pivot from traditional SaaS to AI strategies. And that would be the companies that are traditionally been in that space making that hard pivot. So, it immediately made me curious, like which of the companies that you cover are well positioned to do a hard pivot because hard pivots are by definition hard.

Alex: Yeah, absolutely. And you know, the topic that we went after with this report is such a fun one. And it’s one that we’ve been asked so many times this year because everyone’s trying to get a lay of the land on where are we today? What does this look like? You know, we can see these sort of threads kind of coming through that, these threads that are disrupting the SaaS model and making the whole idea of a UI and an application seem like, you know, how many years in the future does that still exist? Like if we are able to continue on this path of innovation where agents get better, how does that role diminish, and what does that timeline look like? And that’s one of those questions that I think everybody is scratching their head wondering, because we live in this gray world right now where we’re kind of at the precipice of the possibility, but we’re not there yet. And we’re kind of watching what these innovative model developers and the software leaders and the technology leaders, rather, that are leading this charge, watching what they’re doing. And obviously, they’re always going to be marketing with their best foot forward. And so, getting it to terms of what that timeline looks like, it’s a real challenge. And I don’t think there’s a person out there that can give you the perfect answer. And so, with this document, we really just tried to be practical and try to look at, where are we today? What are the challenges that the AI market is facing? And what are the sort of solutions that are on the horizon for those solutions? And that’ll probably ultimately give us an idea of sort of where this whole thing is going. 

You asked about the hard pivot, and obviously the hard pivot, like you said, is something that not a lot of vendors can do. I think that when we look at the SaaS incumbents, it’s something that we’ve seen in a bunch of different places. We went through this whole workforce rationalization for a lot of the vendors that I cover, SAP, Salesforce, where we saw headcount declining, large 10% of workforce layoffs. And then on the tail end of that, we saw rehiring come in right away. And all that rehiring was going into AI sales, AI product. 

Patrick: Right.

Alex: And so, we really have seen a pretty big workforce transformation for a lot of these software incumbents. Salesforce and SAP are the two that jump off the page for me, given just they’re software pure plays. Microsoft, obviously, you have that massive infrastructure, so we didn’t really see that sort of headcount recalibration. 

Patrick: Right.

Alex: We still see some, but it wasn’t necessarily as pronounced as it was when you looked at the numbers for SAP and Salesforce. So, a lot of these big players are looking at their headcount and at their resources and saying, okay, this is what it looked like. We’re going towards AI and that has to be a strategic priority. What do we have to do within our resource base to be prepared to go after that? And that’s something that has been, you know, to different degrees, kind of ubiquitous. 

Patrick: Okay.

Alex: SAP and Salesforce, they really went all in on it. And Salesforce just had their earnings call last night. One of the things that they did was they declined their sales headcount dramatically just two years ago, going through an operational efficiency program, but now it’s up 20% year to year. And so, we’ve seen their sales force go from being cut dramatically. And at that time, some people were talking about like, oh, they’re going to try to lean on their own AI capabilities to sort of augment that massive decline in sales headcount. But that just was not what played out. What we really saw was them getting rid of a lot of sales headcount that were trained and sort of focused on the more traditional software lines, but then hiring back dramatically in the AI sphere and to focus on AI.

Patrick: Right, so when you talk about a hard pivot coming, in some ways it’s the companies that have already begun that pivot that are best positioned to be able to do that. And I understand the struggle, or I understand where the question’s coming from about when will the software as a service model sort of fade away is partly because you don’t want to be caught still using that commercial model, still using that business model when the rest of the world has moved on to something else. 

Alex: Mmhm

Patrick: And being able to predict that, of course, is, as you kind of noted, kind of impossible. 

Partial displacement of large language models

So, one thing I also want to bring up another thing from what you wrote, that was about proprietary models will become critical differentiators for enterprise-grade AI applications. I read that, and I think you’re talking about displacing large language models. If proprietary models will become critical differentiators for enterprise-grade AI apps, is that displacement?

Alex: So, in my eyes, it is partially displacement. I don’t want to undersell the role of LLMs in the long-term AI opportunity. But one of the things that I think has gotten lost too often in this conversation is the role of diversified architectures when it comes to language models and AI models. You need to look, and we all need to look at some serious constraints that the AI market has. The two that jump off the page to me, cost and power need. 

Patrick: Yeah.

Alex: And the cost and the power need is directly related to the number of parameter counts that a model has. And so, as we push general purpose large language models with hundreds and hundreds of billions of parameters, we’re setting ourselves up into a position where you have really a master of none that is extremely expensive to run. 

Patrick: Right.

Alex: And so, the way around that, and something that, this isn’t- in SLM strategy, it’s not new. Like, we’re talking about stuff that’s been sort of in the works for Salesforce, Microsoft, for years now. And it’s been sort of quieter. We hear so much about their partnerships with OpenAI and Anthropic because obviously those firms are sort of pushing the bleeding edge of general-purpose large language models. But Microsoft, they launched their Phi-3 family of SLMs over a year ago. And ever since that launch, they have been adding and adding new partners that they’re developing specific industry SLMs with on top of that. 

Patrick: Right.

Alex: Salesforce, they released their xLAM, which they’re calling large action models, over a year ago. They’re working with Workday to be able to sort of build around that foundation for an HR type model. And so, we’re seeing these partnerships form where you’re bringing in domain expertise from the outside and then working on building niche models. And these niche models are going to be specialized. And they’re not going to be as capable from a general-purpose standpoint, but when you focus the training data, and you focus how you’re building these models around the workflow, you’re able to get use case level capability that is on par with the largest models out there. So, it’s about focusing on specific use cases. The flip side of that is that it’s a much more fragmented development. Like each- instead of focusing on building one model, it’s just something that has to happen across a multitude of engagements, each getting that SLM to a production quality AI model. So, there is a time disadvantage in some ways. In other ways, it requires a lot of partnership from Microsoft and Salesforce to bring in data that they might not necessarily have. When I think of Salesforce, they were great and have access to CRM data, but they’re really positioning themselves to be more than just a CRM agentic platform. They’re kind of trying to butt up against Microsoft specifically as being sort of an overarching agentic platform where it’s capable of dealing and working with HR and ERP and other software areas in addition to where their domain expertise is. But that’s all areas that they have to go out and partner with. 

So, to me, in my eyes, the long term will incorporate large language models and small language models. And when I’ve talked to CTOs too, from software vendors, this is kind of the general architecture that they’re sort of looking for in the future, where an LLM is sort of sitting at the top of like the orchestration layer of an agentic platform, but that large language model is handing off specific tasks to SLM backed. So it’s the orchestration being handled by the powerful generalist model, but then the specific tasks, the work is being done at the SLM layer.

Patrick: Okay.

Alex: And so, the software vendors have been kind of the most forthcoming with that sort of SLM development, and it’s an area that I think that is going to create a lot of value, really be very important to lowering the cost of running these models, and a big part of them sort of cementing themselves within this AI opportunity.

Strong PaaS portfolios and credible SLM roadmaps: SAP, Microsoft and Salesforce

Patrick: And that’s a perfect segue to another question that I had, again, based on what you wrote. And this time I couldn’t even write it in my notebook, it’s too long. But I have to read it out loud because it’s super important. Platform services, data clouds, and integration layers have become the entry point for AI adoption. And this reorientation favors vendors with strong PaaS portfolios and credible SLM roadmaps, which you just touched on. So again, the first question that comes to my mind, because at TBR, we always come back to the specific companies, which are those vendors that have a strong PaaS portfolio and credible SLM roadmaps.

Alex: Well, I mentioned one already, Salesforce. You know, they’re a vendor that I would say that their past portfolio, it’s been improving over the years. I still think that they’re a little bit more immature relative to an SAP or a Microsoft, of course. Microsoft and what they’ve done with Fabric, I view as a pretty big move for them, and they’ve done a really great job of scaling that very quickly. Salesforce, we do see that traction early on. You’re foreseeing triple digit year-to-year increases for Data Cloud and Agentforce. And that’s great signs, but it’s kind of off of a small base. And so, Salesforce has done a lot of work in sort of improving that portfolio and that customer traction is starting. But I would still put them at a, sort of an immature positioning, but something that I think that they’ve shown early strength and that I think that they can get to that point. 

SAP is a vendor that I think has been doing this well for a really long time, going all the way back to BTP and the attach rates that they saw with BTP on RISE migrations and S/4HANA cloud migrations, and being able to leverage that success into Signavio and LeanIX into sort of building these process automation integration workflows that are bringing their SaaS applications together and enabling sort of the groundwork for an agentic system that you can layer on top of that. I think SAP has done a great job at the PaaS layer. I would say that on the AI side of things, SAP has been a little bit weaker than Salesforce. You know, we’ve seen a lot of talk around Joule, but when we talk to customers around how valuable it is, that’s still an area where the jury is still out. When we talk to the, sort of the channel on Salesforce, I tend to hear a lot more bullish sentiment around what they’re doing with Agentforce relative to what SAP is doing with Joule and their other AI capabilities.

Partnership positioning: Microsoft, Salesforce and SAP

Patrick: Yeah. And then, because I’m always interested in the ecosystem play on this, are there certain companies that you cover that you think are probably better positioned to take advantage of all the changes that you’ve been talking about because of their strengths in partnering and their ability to partner better across the ecosystem.

Alex: Yeah, I mean, we cover- when I look at my coverage, it’s the biggest software vendors in the market. And by virtue of being the biggest, you’re also going to have- you’re going to present the most opportunity to partners, and that opportunity is going to drive partner engagement. Microsoft, I mentioned before, they have been tacking on new partners for their Phi family of industry models very consistently ever since they announced that initiative. And so the engagement has been great there. Salesforce has seen great engagement around partners building on Agentforce. SAP obviously has great partner engagement, especially when it comes to migrating to the cloud. As far as partner engagement on AI, again, this kind of comes back to sort of what we’ve heard on the ground level. It’s just less pronounced than what I’ve been seeing from Salesforce and Microsoft. But it’s still kind of a case where they’re preoccupied with getting to the cloud, where making this shift towards AI almost feels harder because they’re focused on that broader modernization story still. I think that they’re gonna- they’re kind of forced based on the market into sort of promoting their AI a lot and we continue to hear more and see them doing stuff. And so, they’re certainly working towards having a mature AI strategy, but that’s still something that I think that relative to Microsoft and Salesforce, I would put them at a little bit of a weakness. But again, we’re talking about the largest vendors, and there’s going to be partner engagement for a lot of these, or all these agentic platforms that are coming out.

Patrick: Right. And I think you touched on something there, maybe inadvertently, but the boom in marketing budgets, because everybody has to talk about how great they are at AI now. So, if you’re not marketing yourself around AI, you’re out of a job. 

Looking at next year’s Cloud & Software portfolio

So last question, as we go into 2026, and you think about the companies that you cover, and you think about the portfolio, the Cloud and Software portfolio more broadly, are there certain issues you think you’re going to be tackling in the coming year, whether that’s new coverage or maybe more importantly, what are some of the- this is a question I want to really go sink my teeth into?

Alex: I mean, the big challenge for me right now is starting to quantify what’s happening within the finance- so right now we get loose estimates around revenue generation related to AI from all these vendors. 

Patrick: Right.

Alex: But it’s never a clear view. And so, being able to dig into, okay, what is the monetization success? That’s something that is going to be on my mind in 2026. You know, Salesforce, they just had a good quarter in terms of their guidance going forward, and so they’re kind of at that point where they’re saying AI revenue is on the horizon now, and so does that come through? Is that real? And being able to put numbers to that, that’s going to be a big challenge in 2026.

Patrick: And that’ll be super important because we’ve heard now for a couple of years, just the massive numbers around, we’re investing a billion, 2 billion, 3 billion, 5 billion in AI. So now we need to see where’s the revenue that’s going to come out the other side.

Alex: Yeah. And that’s a massive question. I mean, it’s a whole other topic to go into sort of the investment cycle and especially on the hyperscaler side, which we’re talking about cloud software today so, outside the scope. But yeah, we got to see the money now. It’s very important.

Patrick: All right. We got to see the money now. That’s a great way to put it. And we will come back in season five and we’ll talk about exactly that question. Alex, thank you so much.

Alex: Appreciate it. Thank you.

Final thoughts 

Patrick: Next week, I’ll be speaking with Allan Krans and Angela Lambert about our 2026 Alliances predictions. 

Don’t forget to send us your key intelligence questions on business strategy, ecosystems, and management consulting through the form in the show notes below. Visit tbri.com to learn how we help tech companies, large and small, answer these questions with the research, data, and analysis that my guests bring to this conversation every week. 

Once again, I’m your host, Patrick Heffernan, Principal Analyst at TBR. Thanks for joining us, and see you next week.

T

TBR Talks: Decoding Strategies and Ecosystems of the Globe’s Top Tech Firms

Join TBR Principal Analyst Patrick Heffernan weekly for conversations on disruptions in the broader technology ecosystem and answers to key intelligence questions TBR analysts hear from executives and business unit leaders among top IT professional services firms, IT vendors, and telecom vendors and operators.

“TBR Talks” is available on all major podcast platforms. Subscribe today!