‘Our partners have a role in helping our customers … navigate AI transformation successfully, coming out the other end, driving growth versus just efficiencies. So we get very excited about that,’ he tells CRN.
Microsoft Commercial CEO Judson Althoff has a bold goal for Microsoft’s 500,000-member partner ecosystem over the next 12 months.
Althoff hopes to see more “impactful frontier transformation scenarios come to life across our customer base” through applied AI enabled by partners. Althoff clarified that “transformation” in the AI era isn’t only about efficiency and reducing costs.
Frontier AI transformation can mean more innovation and democratized intelligence throughout an organization. “And these frontier transformation scenarios come from a place of actually understanding the business flows and the business opportunity,” he said. “I want to see AI unlock growth for our customers, powered by our partner ecosystem. And I think we have a great opportunity to make that happen.”
Read on for more of what Althoff had to say to CRN about the opportunity for solution providers to unlock AI transformation for customers.
What’s your message to Microsoft’s solution providers?
At Microsoft, we’re a company that’s been founded on the shoulders of partners and have worked with partners throughout our entire 50-year existence.
Partners are always going to be a critical part of how we operate here. And given what we’re going through right now in terms of AI transformation and its impact on society, its impact on our customers, it’s never been more important to have a strong partner ecosystem.
I really genuinely believe there’s never been a better time to be a Microsoft partner than right now because if you just follow the landscape of everything that we’ve been through over the last couple of years, and the excitement around the opportunity as well as the concerns over the impact and the risk associated with AI, I think partners are staring at contemplation around what is the future of the services business. What is the real opportunity out there in AI? Will AI eat software? Will AI eat the world’s GDP? And how do you find some real grounding and all of that relative to building a business around it and carrying it forward?
Over the last [seven] months, I have really regained my conviction on the role of the partner in today’s society. Because the reality of the situation is that the two most important things for AI success right now are not models, and they’re not silicon. But they are, rather, intelligence and trust.
And we have pivoted our entire portfolio of capabilities around delivering an intelligence platform and a trust platform.
At the end of the day, customers need to know that their IQ, their intelligence, is being amplified by AI and not somehow harvested by AI—that what they do to provide unique value in their companies, in their industry, to provide competitive differentiation, that their IQ is being protected and amplified through AI, and that there’s this trust platform that allows them to govern and manage the same.
And so what that means for partners, very broadly, is customers need partners to come in and build out those AI capabilities.
Our products, like Copilot, provide a great gateway or entry into AI from the end-user perspective. But in order to get real value, we have to build agents that line back up to the business processes that are most important for our customers.
And our partners have a massive opportunity to go build practices to build out those agents by line of business, by industry, connect them back into Copilot so you get these end-to-end agentic business processes, and then have the opportunity to govern and manage those scenarios on behalf of our customers.
How is Microsoft investing more into that governance layer for AI users?
We recently launched in November our Agent 365 product, which is an observability product. It allows you to see agents operating across your environment of all kinds, whether they’re built on the Microsoft platform or any other.
It’s an open agent registry. You can give agents identities. You can observe their behaviors. You can track ROI associated with the use of those agents in the environment.
And as the proud owner of IT inside of Microsoft, I made my team turn this solution on inside of Microsoft back in November before we exposed it into preview. And we discovered we had, at that time, [500,000 AI agents generating 65,000 responses per day over a 28-day period].
And so as a technology company, we were even overwhelmed by the amount of agents happening and being used in our environment. So you can only imagine what our customers are going to go through.
And the opportunity for partners around delivering a managed service to help customers get value out of their agents, tune their agents, secure their agents, and govern and manage them, is massive because I think most of the commercial customer base out there is not going to be equipped to be able to handle managing these agentic processes. So [it’s] both from a build opportunity and a manage opportunity.
It’s why I’m so convinced that the opportunity for our partners has never been greater, and we’re really excited about investing in that ecosystem to really help bring those two aspects to life here. It’s such a critical body of work for us right now.
What is Microsoft doing to help solution providers with their AI practices?
[There has] been a lot that we’ve learned as a company over the last couple of years since generative AI really hit the scenes.
And if you take our Copilot product, for example, it’s built on top of Microsoft 365, which has a broad install base across our commercial ecosystem, largely thanks to our partners.
So many of our CSP [Cloud Solution Provider] partners and SI [systems integrator] partners out there manage those solutions on behalf of our customers and were a part of implementing those solutions for our customers. So Copilot itself really builds upon that.
We’ve been working really hard to actually open up Copilot. We’ve made Copilot open with respect to model diversity. At the beginning of this month, we announced that you can use [Anthropic] Claude models inside of Copilot in addition to OpenAI [the maker of ChatGPT] models.
And you connect to agents built with any model that sits on top of our Foundry platform. And our Foundry platform has over 11,000 different models in it.
[Another example is meeting customers] regardless of what cloud platform they sit on. If they sit on Azure, that’s, of course, great. We love that.
But if they build agents on the Google platform or on the AWS platform, it’s very easy to connect them back into Copilot now and then manage them with Agent 365.
We have seen this push for openness, for model diversity, coming from our customers and sort of engineered that into our products.
And now, effectively, we’ll be going out to the partner base and training and educating them and building the skills out there to really have core practices around this build-and-manage paradigm. And so we’re very excited about that.
We think that for partners, the biggest imperative is to learn the adjacent skills. The good news is they’ll already come from a place where they understand Microsoft 365 quite well.
And so amassing the AI skills around that to then implement Copilot and build out the agent flows around it is a natural progression for them. And then I think the hardening of the managed services opportunities around the business model is something that we would encourage them to do as well.
So it’s about skills. It’s about being structured around the business model. And frankly, leveraging what you know. So much of what we built in this intelligent platform lies on the rails of what our partners have implemented for our customers for a long time.
So this idea that our customers protect their IQ using the Microsoft IQ platform is something that our partners should really leverage out there in the market.
What do you want to see from the partner ecosystem in the next 12 months?
What we really want to see is the frontier transformation of our customers. And right now, I think there is a real high degree of standard deviation in terms of how our customers are even adopting AI.
And some of it’s industry bias. Some of it is segment bias. But a lot of it is skills bias.
And I think partners can bring the pragmatic, applied use of AI to so many of our customers, regardless of what industry they serve or what segment they serve. And so what I really want to see are more of the impactful frontier transformation scenarios come to life across our customer base.
What is an impactful frontier transformation?
‘Transformation’ is probably the most overused word in our industry right now. Everybody talks about it, unfortunately.
Most people have used the word over the last couple of years to mean efficiency. How can I use AI to reduce my cost structure in my business? Which then, of course, has the negative connotation of talking about job loss.
The reality of the situation is AI has so much power to unlock creativity, unlock innovation, and really push and pivot toward growth. And so we use the term ‘frontier transformation’ and to talk about democratizing intelligence.
Finding the cure for the incurable. Discovering that world’s greatest filmmaker that doesn’t know they’re a filmmaker yet. And these frontier transformation scenarios come from a place of actually understanding the business flows and the business opportunity.
Of course, with applied use of technology and applied use of AI—but if you don’t understand, for example, retail commerce and hyper personalization for that consumer-packaged goods segment and you’re going to try to go in and implement AI on behalf of that customer, I don’t give you high odds of success.
But the good news for our partner base is we’ve been very focused on industry skills. We’ve been very focused on business process transformation. It’s the lifeblood of that SI ecosystem that we have.
And so understanding how you’re going to reinvent the supply chain, understanding how to even use the concepts of Lean [a business management approach focused on maximizing customer value while minimizing waste] for information work and do the hard yards of process redefinition and then apply AI—that I think yields to great success.
I want to see AI unlock growth for our customers, powered by our partner ecosystem. And I think we have a great opportunity to make that happen.
What are the biggest lessons you’ve learned as AI has grown in adoption?
AI success really is not about the best model. It’s not about the best silicon. It is about intelligence and trust and unlocking that for our customers. And our partner ecosystem is really the only way we’re going to actually get out there and achieve it. I view it actually as a competitive advantage for Microsoft to have the partner ecosystem that we have.
Why do I think that it’s not about the model? Well, if we would have had this interview back in November, you would have probably said, ‘Well, gosh, Judson, [OpenAI’s] GPT-5 is state of the art. What do you think of GPT-5?’
And then, if we had it again a month later, you would say, ‘What about [Google’s] Gemini 3? Don’t you think Gemini 3 is the best model?’ Today, you’d be asking me about [Anthropic’s] Claude. ‘Gosh. Claude seems amazing. Like, isn’t that state of the art? Isn’t it going to be all about Anthropic?’
And I would say [to you], ‘Wait a month. Because I can’t tell you which one it’ll be, but you’re going to be asking me about a different one again in a month.’
And so do we really expect that customers are going to replatform how they think about information work? How they think about DevOps? How they think about security? How they think about supply chain management? How they think about finance? How they think about legal?
Do you think they’re going to rethink that every time a new model drops? Well, no.
So what have we done about that? We’ve created this IQ platform.
We have three pillars in the IQ platform: Work IQ, Foundry IQ and Fabric IQ.
Work IQ is basically the brain inside of Microsoft 365 Copilot. It understands all of the semantic context of all of the Microsoft 365 implementations that our partners have done with our customers.
So it knows how you work. With whom you work. The content over which you collaborate. Your most important business flows.
And it will continue to make M365 Copilot faster and more accurate and trusted while being model-diverse and open and heterogeneous. Because what we’ve done with Work IQ is we’ve also now made it an agentic building block. And so if people want to build agentic flows on top of Work IQ, our partners can very easily—either using Copilot Studio or GitHub or their favorite IDE [integrated development environment]—on top of Work IQ.
We’ve done the same thing on the data side with Fabric IQ, where basically we took the brain out of Power BI. Think of it that way.
All of the semantic models inside of Power BI that are multi-cloud, multi-data service. So if you have data in Snowflake, [Google’s] BigQuery, Databricks, doesn’t matter—those semantic models are harnessed inside of Fabric IQ and you have one intelligence layer.
And then finally, in the middle, we have Foundry IQ, which is where you build your knowledge bases that are custom to your business. Any external index data that you use to build agents.
How do these help Microsoft solution providers build intelligence and trust with their customers?
When I say the most important things are intelligence and trust, helping our customers build out that intelligence platform so that whatever new model drops can amplify that intelligence means that you’re preserving your unique value. What you do differently in your industry.
And our partners have a role in helping our customers harness all of that and navigate AI transformation successfully, coming out the other end, driving growth versus just efficiencies. So we get very excited about that.
And the trust equation matters just as much because you have got to be able to observe, govern and manage the outcomes and make sure you’re getting the ROI from AI while at the same time protecting your enterprise.
All the talk is about the models and the silicon. At the end of the day, intelligence and trust are what makes AI successful for our customers. And our partners are key to delivering all of that. So very excited about that.
What have you learned watching the channel evolve over the decades?
I feel like we have been telegraphing for a decade that where this is all headed is in service delivery.
And we come from a basic, deep perspective, being the company we are today because of the success of our channel—but our best partners are the ones that have gone out and built real, bona fide service practices or brought companies in to build in organically those kinds of capabilities.
And I feel like that industry knowledge— that core expertise and development and the core long-term service ability and managed services—are what are going to prevail in this new AI world.
What’s your take on this conversation around whether AI will kill Software as a Service?
I think some SaaS will die. It’ll largely be dependent, though, on the importance of the semantic model that the SaaS in question harnesses.
There are some SaaS companies who built the business off of pretty lightweight workflows that are, frankly, easy to replace with agents and agentic business flows.
But the core system of record, the idea that somehow the general ledger goes away or your main supply chain inventory database goes away? No, probably not happening.
They will be modernized, of course. And they will all need to become headless so that agents can operate over them.
But we do believe, at the end of the day, AI has to serve human ambition, which means that for a long time, you’re going to be dealing with human beings and AI agents working together across business flows.
So for that, you’ll need a multitude of UIs from folks who love to live in a command line to, if you’re more like me, you appreciate things in graphical form.
I always say to those who push on the command line interface that 1987 called. It wants its DOS prompt back.
People are going to want to work the way they want to work. Some thrive in a command line. Some thrive in a GUI.
Agents are going to want to speak to agents. And so [key will be] decomposing these systems of record into the core data that matters the most. The data estate will, of course, matter. The semantic context that I talked about with our IQ platform–—it’s great to have a data estate that’s well-curated. But if you don’t understand the context and that semantic layer for how it’s being utilized across your people and your agents, it’s not terribly valuable.
So we’ll see strong monetization, I think, in that context layer and the agent harnesses that run on top of it. So SaaS is not dead. But SaaS must evolve in order to accommodate human ambition and how AI agents will work across the same.
What’s your advice to Microsoft partners navigating rising memory prices?
The way partners should think about this is that we are entering into a phase of this AI wave that is going to pressure the world’s capacity because AI models are needing more power, more memory, of course, highly specialized silicon.
And so efficiency will start to matter a lot. And so I think the good news for partners, particularly those who invested in deep services practices, [is they] will have an opportunity to take a look at an agentic flow.
And this is where we love the Agent 365 platform for this because you’ll be able to see the supply chain flow that you’ve built on behalf of a customer and say, ‘Wow. Where are the hotspots?’ And pick, perhaps, the out-of-stock agent and say, ‘Wow. This agent is using a frontier model. And we can probably use an open-source model that we fine-tune on behalf of the customer, save them money, consume less resources.’
There will be this need to drive more efficient use of all of the assets that are out there because the world, I think, is going to see a bit of a supply crunch in multiple places. I think partners have an awesome role in consulting with our customers to help drive those efficiencies.
