Microsoft's $2.5bn bet

Mike Belsito hosts Now Shipping, Mind the Product's weekly AI news briefing for product practitioners. This episode brings together three stories that share a single throughline: the model era is giving way to the deployment era. Belceto unpacks Microsoft's $2.5bn bet on embedded AI delivery, the same company's simultaneous 4,800-person headcount reduction, and an enterprise benchmark study revealing that 71% of executives at billion-dollar companies say their own organisation is the biggest barrier to AI performance.


Chapters:
(0:00) Introduction
(0:23) Three stories, one thread
(1:31) Microsoft Frontier Company
(5:17) The AI layoff wave
(11:39) The organisational readiness gap
(15:10) Wrap-up

Transcript

SPEAKER_00 0:00

I'm Mike Belcito, and this is Now Shipping, the weekly AI news show for product people. Every week I bring you three AI news stories that matter to you, the people actually building software products for a living. Not the hype, not the noise, just the stuff you need to keep up with. Brought to you by the team at Mind the Product, this is Now Shipping. This week we have three stories that are all sort of a part of the same thread. I'll give you a little preview of all three. The first one is about Microsoft committing $2.5 billion and 6,000 of its own team members to a new unit whose only job is deploying AI inside enterprise customers. And there's been a few companies doing this lately, so it seems to be the start of a trend. Next up, four days later, Microsoft cut 4,800 jobs. Same company, same week. And look, they haven't been the only ones making layoffs. TechCrunch actually published a running list of every major tech layoff in 2026, especially those explicitly naming AI as the reason. And the list is pretty long, but who's getting cut first might be the real story. And finally, story three is about a new study that found 71% of executives at billion-dollar companies. They said that the biggest barrier to AI performance isn't technology, it's their own organization. Maybe there's a product opportunity hiding in here. Okay, that's a preview of this week's three stories. Let's get right into it. All right, story number one, all about Microsoft. On July 2nd, Microsoft bet $2.5 billion that the hard part of AI isn't actually building the models. It's everything that happens afterwards. Microsoft announced a new operating unit called Microsoft Frontier Company. 6,000 engineers, trainers, industry specialists, and sales experts whose entire job is to embed directly inside enterprise customers and deploy AI. Not to sell them a license, but actually embed AI. Literally go into the client company and run the implementation with them. Early named partners include big, big companies like Unilever, Novo, Nordisk. I mean, these aren't small pilot programs either. These are full-scale enterprise engagements. Now, here's the thing: Microsoft isn't the only company going down this path. Meta announced a near-identical unit recently calling it Enterprise Solutions, placing Meta's own engineers and product managers inside large corporate clients to deploy its tools. Amazon's followed with a similar commitment. We're hearing about consulting companies spinning up units, and they've been doing this for quite some time. It seems like every major AI provider has pretty much reached the same conclusion. The model is no longer the hard part. Getting AI to actually work inside a real company, a big company with messy data, legacy systems, teams that honestly weren't built for this, and leadership that bought the vision, but maybe they haven't quite figured out what comes next. That's all where it stalls, right? And the industry has pretty much collectively decided it's worth billions and billions of dollars to fix. So what does this mean for you? Well, first, if you're building software for enterprise customers, I'd say the competitive environment just got a little more competitive. I mean, the biggest AI vendors are now selling deployment outcomes, not just access to the model. That means before these customers sign anything, enterprise buyers are going to start asking, okay, who can actually make this all work? So if your product can accelerate time to value through things like smarter onboarding, tighter integrations, adoption tooling built into the product, it becomes a real differentiator. It becomes a lot more meaningful than maybe what you thought those things would be, say several months back. Now, if your product hands customers uh log in and just sort of, you know, wishes them luck, like it's giving them the tech, but now it's up to them to sort of figure it all out, you might want to rethink that strategy. That could end up becoming a very big barrier. Also, I would say watch what this does to pricing expectations. You know, when deployment's part of the product, when the value prop is transformation, not just the capability, not just access. The economics end up looking a lot different for products. I mean, we're at the beginning of what seems like maybe a split in an enterprise AI. There's the commodity model access on one side, then there's the high-touch transformation on the other side. If you're somewhere in the middle, you might not be seen as differentiated enough. You might be seen as one of these companies that's sort of not really making their bet one way or the other. If you zoom out for a second, I think it's interesting that Microsoft is committing 6,000 people to building this deployment layer. I mean, this is a massive org redesign decision. They're saying the constraint on AI value creation isn't actually the entire story, right? It's the human and process layer around it. And by the way, in story three, we'll talk about how they're not the only ones that feel this way. So, anyway, this is story one. We'll move on to story two, and Microsoft is a part of that story as well. All right, story number two. It's about the continual wave of layoffs that keep happening in the name of AI. And we'll start with Microsoft. On July 6th, four days after Microsoft announced 6,000 new AI deployment jobs, Microsoft cut 4,800 employees, about 2% of their global workforce. Now, this ended up hitting the Xbox division hardest, also commercial sales and roles around the coordination layer of the organization. Microsoft's chief people officer Amy Coleman said the roles eliminated today are not being replaced by AI. I do think it's interesting, though, that a company that just committed $2.5 billion to an AI deployment unit said that their own layoffs aren't really about AI. I don't know, maybe it's technically true. Maybe the Xbox restructuring is about gaming market dynamics. Maybe the commercial sales cuts are about go-to-market efficiency. But maybe, just maybe, since Microsoft is doing this in the same week, where they are investing in AI deployment. And look, the entire industry has been doing this over the past year where they invest heavily in the new layer that AI creates while trimming the old layer. Maybe that's actually what it's all about. Um, look, they're not alone. On July 6th, TechCrunch published an updated tracker, a running list of every major tech layoff in 2026 where the company explicitly named AI as the reason. Microsoft, of course, not necessarily on that list since they didn't name it, but that list is pretty long. There are public companies, well-funded startups, category leaders, and the companies aren't being vague, like maybe Microsoft was in their announcement. They're actually taking the opposite approach. They're naming AI in the announcements. It of course all started with Block, Jack Dorsey's company. They cut 4,000 jobs, nearly half the company. And Jack Dorsey ended up writing on X after they made that layoff. I believe the majority of companies will reach the same conclusion and make similar structural changes. And soon they did. Soon after that, there were all sorts of cuts at multiple companies again and again with efficiencies and AI as a big part of that. Meta cut thousands of people, 8,000 people, 10% of their total workforce. They also moved 7,000 people into AI focus roles, which from what I've heard, folks working within them don't exactly love the roles. Oracle's cut 20,000 roles. Google, Inuit, Cisco, Coinbase, they've all made big cuts all in the name of AI. Now, most people read a list like this as hey, jobs are disappearing because of AI. Like that's what the story's all about. I actually think the interesting question is who's actually being cut here? When you look at the roles, they're roles like middle managers, dedicated QA roles, program managers, certain PM functions, specifically the ones where the work is centered on information brokerage, you know, gathering updates from different teams, synthesizing those updates into something readable for leadership, writing the status doc, tracking what everything's doing. It's what I would call the mechanics of product work. Not the actual product work, but the mechanics of product work. And when you think about it, it does kind of make sense. I mean, that kind of work, pulling context from different places, surfacing it clearly, keeping people aligned, that's the stuff AI is pretty good at right now. Now, I'm not saying it does it perfectly, but it does it well enough that companies are making this bet. Now, ideally, companies would realize AI can do that kind of stuff well and they wouldn't actually make cuts. They'd automate that part of the work away and empower their product people to actually focus on the things that matter: product sense, taste, judgment, staying close with customers, real product work. The reason why most of us got into product in the first place. But we also know right now, for better or for worse, layoffs in the name of AI, it's sort of a signal. It's like a company saying, look how efficient we are. Like we're so efficient. We don't even need all these people because we're amazingly efficient. I'm not saying that's right, but it is a weird signal right now. So how do you wrestle with all this? Well, remember the PM roles most at risk right now are focused on the mechanics of product work. My bet is that, again, you didn't get into product to coordinate the weekly status update, to write the ticket description, um, to synthesize the research doc. If that's a meaningful chunk of your work in a given week, it may be time to proactively automate those tasks away so you can focus on more of the important stuff, the strategy, customer insights, you know, making calls in ambiguous situations. Don't think of this as like, well, yeah, someday I'll automate the mechanics away and I'll focus on that. No, do that right now because if you don't, somebody might make that decision for you. Also, remember all of this, it's really org redesign. It's not necessarily cost cutting. Okay, maybe in some situations, maybe in a lot of situations, some of these companies have overhired in the past, and this is a good reason to make these cuts. But a lot of cases, this is really org redesign. And what ends up, you know, being redesigned after the fact is a more autonomous uh organization, one with fewer layers. Um, we're seeing the player coach model, one where the person does hands-on work and leads, like that's showing up in a lot of these announcements. And I think it's worth thinking about. So I look, I know stories like this are depressing. I get it, but I would read all this as an opportunity. The roles that are being rebuilt after the cuts, and make no mistake, a lot of these companies are hiring after they make these cuts. They are going for people that, you know, I'd call AI native, people who have genuinely integrated AI into how they think and operate every day. Not just the ones that use ChatGPT a few times, but people who use agentic AI, the way they use email. They just sort of do it. It's second nature, and it's not too late to get to that point. There's so many amazing resources to learn how to get good at that. So the time is now to do it. All right, anyway, that's all story two. One more story to go. Okay, finally, story three. It starts with a number, 71%. And that number is because 71% of executives at companies with over a billion dollars in revenue say that organizational readiness is the biggest barrier to AI performance at their company. Not the technology, not the models, it's their own organization. Only 11% said technology was actually the problem. Now, this is from the Payments Intelligence Enterprise AI benchmark report, which was just published this week. Major enterprises across industries were asked plainly what's actually blocking your AI performance. And almost three-quarters pointed inward. Unclear ownership, misaligned incentives, teams that weren't built to work with AI, processes that haven't changed. Leadership bought the vision, but they haven't figured out what comes next. Look, we heard these basic sentiments earlier on in this episode, right? I mean, Microsoft just spent two and a half billion dollars and committed 6,000 people to try to fix this exact problem for enterprises at scale. They're literally doing this inside of their enterprise customers. They looked at a market where the majority of buyers are stuck, not because the AI doesn't work, but because their organization just doesn't know how to use it. And even Microsoft is committed to building a business around solving that. And then, of course, you go back to story two the roles that are being cut the fastest. It's all in the coordination layer, the layer that AI is starting to replace. Um, what's being rebuilt in its place? Well, roles that help organizations actually change. Deployment specialists, implementation engineers, people who can sit inside a company and help it transform. The threat across all three of the stories today is pretty much the same. The model era is giving way to the deployment era. Capability is no longer the differentiator, it's what you do with the capability and how your organization changes around it. That's what's actually making a difference. So, what do you take away from all this? Well, again, if you're building enterprise AI products, organizational readiness, this is now a part of your product problem. This isn't just a customer success issue. It's a product issue because if the thing stopping your customers from getting value out of what you built is their own organization, not your features, well, your product has to help them get organized. So things like adoption design, onboarding that changes behavior. Again, we talked about this earlier in the episode, but the teams that crack these things will close deals that technically identical products just can't close. And if you're inside a company trying to adopt AI, that stat should feel validating. You may feel like you and your entire organization are way behind. The reality is you're probably not far behind the majority of companies. You're actually probably in the same place. But even if you are behind, no, it's probably not because the wrong model choice. You might be behind because nobody's clearly answered the three questions. Who owns AI adoption here? What does success look like? And what has to change about how we actually work? And if nobody's answering these questions in your organization, maybe it's a gap where you as a product person can actually fill that gap. Maybe you can be the one to lead the charge within your organization. I mean, somebody has to answer these questions. If it's you, you might have found a way to future proof yourself inside of your organization. So, anyway, that's a wrap for this week's episode. If you found this valuable, please subscribe, tell a friend. If you have any suggestions, definitely leave a comment below. I promise you, I will read every comment you leave, and I will take those comments and try to make these episodes even better in the future. So, with all of that, once again, my name is Mike Belsito, and brought to you by the team at Mind the Product, this is now shipping.