Microsoft Isn't Selling Software Anymore — It's Sending People

On Thursday, July 2, Microsoft made a pretty unusual announcement. Not a new product, not a new model — a new "company." It's called Microsoft Frontier Company, and it's backed by $2.5 billion and staffed with 6,000 employees. What do they actually do? They go inside customer organizations and design, deploy, and improve AI systems directly. This isn't about licensing software — it's about sending people.

Here's why that's interesting: it's a totally different playbook from what Microsoft has run so far. Selling Azure, selling Copilot, bolting AI features onto Office 365 — that's the traditional move. "Buy our stuff." This time it's different. The pitch isn't "buy our stuff," it's "our people will walk into your company and physically build the AI in for you." The industry has a name for this: forward-deployed engineering, or FDE.

This trend didn't originate with Microsoft, to be clear. Earlier this year, Amazon, Anthropic, and OpenAI each announced their own AI deployment groups. All three companies seem to have landed on the same diagnosis: the models themselves are good enough now — the real bottleneck is actually getting them embedded inside a real company. Microsoft is riding the same wave here, except the scale — $2.5 billion, 6,000 people — is what makes this one stand out.

The person running this new unit is worth a mention too. Rodrigo Kede Lima, who previously led Microsoft's Asia business, is now president of Frontier Company. Moving from a regional lead role to heading a brand-new global unit is a signal in itself — it suggests Microsoft's internal leadership is putting real weight behind this project.

The Players

Let's start with the guy running the show, Rodrigo Kede Lima. He previously led Microsoft's Asia business, and now he's president of a global unit built from scratch. His background running a specific region — and one that includes markets with very different enterprise IT maturity — likely shaped how this org is structured. It's worth noting Tokyo made the list of five regional hubs; that Asia-market instinct probably wasn't an accident.

Then there's the 6,000-person workforce actually doing the work on the ground. They break down into four distinct roles: 2,000 solution architects, 1,800 deployment engineers, 1,200 trainers, and 1,000 strategists. Four groups, each with a different job, but functioning as one coordinated team. What's notable is that all 6,000 go through a six-week AI induction program. This isn't just existing staff getting reassigned overnight — it's people being specifically retrained for this mission.

The customer-side cast matters just as much. The named early customers are LSEG (London Stock Exchange Group), Land O'Lakes, Unilever, and Novo Nordisk. Financial infrastructure, agricultural cooperative, consumer goods, and pharmaceuticals — four completely different industries signed on as early customers at the same time. That spread suggests Microsoft designed this model to be industry-agnostic rather than built for one vertical.

And you can't leave out the supporting cast of consulting partners. Accenture, Capgemini, EY, KPMG, and PwC — five major consulting firms — are tasked with extending this initiative globally. Microsoft's own 6,000 people can't possibly cover every customer worldwide, so the plan leans on consulting firms that already have deep roots in every industry and region.

What Happened

Here's the announcement boiled down to one sentence: Microsoft is committing $2.5 billion and 6,000 employees to a new operating unit that embeds engineers directly inside customer organizations to design, deploy, and improve AI systems. This went out via Microsoft's official blog on Thursday, July 2, and was quickly picked up by TechCrunch, CNBC, and GeekWire.

The physical structure of this org is worth a closer look too. It's organized around five regional hubs: Redmond (Microsoft HQ), London, Tokyo, São Paulo, and Sydney. That covers North America, Europe, Asia, South America, and Oceania — one hub per major region. On top of that, there are smaller satellite offices in 18 additional cities. So the five main hubs likely handle strategy and training, while the satellite offices manage the day-to-day relationship with local customers.

Here's the org structure laid out in a table for a quick overview.

Category Headcount/Scale Role
Solution Architects 2,000 AI system design
Deployment Engineers 1,800 Hands-on implementation and deployment
Trainers 1,200 Training customer's internal staff
Strategists 1,000 Adoption strategy
Main Hubs 5 (Redmond, London, Tokyo, São Paulo, Sydney) Regional command centers
Satellite Offices 18 cities Local customer touchpoints

The announcement also specifies that all 6,000 people go through a six-week AI onboarding program before deployment — meaning the entire workforce has to pass the same training curriculum before being sent into the field. Given the scale involved, this doesn't look like a small pilot program; it reads as a serious, organization-wide bet from Microsoft.

The named early customers are LSEG, Land O'Lakes, Unilever, and Novo Nordisk, and to scale this model globally faster, Microsoft brought in consulting heavyweights Accenture, Capgemini, EY, KPMG, and PwC as partners. Rather than trying to cover every customer worldwide with its own 6,000 people, Microsoft designed this so the consulting firms act as an extended talent pool.

What Each Side Gains

Let's start with what Microsoft gets out of this. A common complaint in the enterprise AI market has been "we bought the license, but we can't actually get it working inside our company." Copilot adoption often stalled because employees didn't know how to actually use it, and it just sat there unused. Frontier Company is Microsoft directly closing that gap themselves. It's not just selling software and walking away — it's a signal to the market that Microsoft will stay on the hook until the software actually works inside the customer's walls. That should translate into higher real-world usage of Azure and Copilot, and over time, better retention on those subscriptions.

What the customers get out of it is just as clear. Companies like LSEG and Novo Nordisk often want to adopt AI but lack the internal headcount to design and run it. Hiring AI engineers takes forever and they're expensive. If Microsoft can send in a fully trained team of solution architects and deployment engineers instead, the customer gets to skip the hiring risk and jump straight into a real project. And the fact that 1,200 trainers are allocated separately suggests real thought went into knowledge transfer — making sure the customer's own staff can run the system themselves once the engagement wraps up.

The consulting partners aren't getting a bad deal either. Firms like Accenture and KPMG have long-standing, deep relationships with customers across every industry. By partnering with Microsoft's Frontier Company, they get to layer Microsoft's AI technology and manpower on top of a client network they already own. That's a brand-new AI-related revenue stream for the consulting firms, and for Microsoft, it's a way to reach regions and industries that its own 6,000-person team simply can't cover alone. Each side fills a gap the other has.

For Rodrigo Kede Lima personally, this looks like a major career inflection point. Going from regional lead to president of a brand-new global unit signals a serious level of internal trust at Microsoft. If this project succeeds, it carries real weight both for his career and for Microsoft's broader AI strategy.

Precedents: Wins and Failures

The concept of "forward-deployed engineering" isn't something Microsoft invented. The practice of a software company embedding its own technical staff directly inside customer organizations has already been tested and proven at various companies across Silicon Valley. The reason it works is straightforward: no matter how good a piece of software is, fitting it into a real company's environment — legacy systems, messy data structures, internal politics — is a completely different challenge. The industry has learned through experience that closing that gap ultimately requires actual humans on the ground.

The fact that Amazon, Anthropic, and OpenAI all announced their own AI deployment groups this year points to the same underlying logic. All three seem to have arrived at the same conclusion: our models are already good enough, and the real bottleneck is the last mile of actually embedding them inside enterprise customers. Each came up with roughly the same solution — sending technical staff directly into customer organizations. That's not a coincidence; it looks more like the whole industry independently spotting the same chokepoint at the same time.

That said, it's worth being honest that this approach hasn't always worked out. Sending large teams of people into customer organizations has a fundamental scaling problem. Software, once built, can be copied and sold infinitely. People can't. There's a hard physical limit on how many customers one person can support, and the labor costs never stop accumulating. There are cases across the industry where companies tried this model and later had to scale it back or pivot due to profitability issues. Microsoft's decision to bring in consulting firms as partners is likely a direct response to this exact scaling problem — recognizing that 6,000 people alone can't solve it.

Ultimately, the real question is how efficiently these 6,000 people actually operate, and whether this model turns a profit over the long run. Since we're still at the announcement stage, there's no actual performance data yet — that's something only time will settle.

Rivals' Counterplay

As mentioned earlier, this announcement didn't happen in a vacuum. Amazon, Anthropic, and OpenAI had already announced their own AI deployment groups earlier this year. So Microsoft's move here isn't opening a new front — it's joining a fight that was already underway, just at a much bigger scale. $2.5 billion and 6,000 people is an aggressive bet compared to what competitors have put on the table, and it reads as Microsoft signaling to the market: we are not going to lose this fight.

This puts real pressure on OpenAI and Anthropic. Both companies have competed primarily on raw model performance, and now Microsoft is playing a different card: "we don't just give you the model, we give you the people who'll build it into your company too." That adds a whole new axis to the competition. For enterprise customers, "which model is smarter" might matter less than "who can actually get this built inside our company." And on that front, Microsoft has an edge it can lean on — its existing Azure cloud infrastructure and its long-established enterprise sales organization.

Amazon faces similar pressure. AWS is already Azure's biggest rival in cloud infrastructure, and now this announcement extends that head-to-head fight into the "who actually embeds AI inside enterprise customers" arena too. Amazon already has its own AI deployment group, so it's easy to picture the two companies competing for the same customers going forward — a race over who can embed AI faster and better.

What's especially interesting here is that Microsoft pulled in consulting firms as partners. Accenture, Capgemini, EY, KPMG, and PwC have traditionally worked across multiple vendors rather than being locked into one cloud or AI provider. Microsoft partnering with them also means competitors can approach the same firms. That means these consulting firms could end up playing kingmaker across multiple AI giants simultaneously — and that dynamic could end up shaping the next round of competition just as much as any single company's headcount or dollar figure.

So What Changes

For everyday consumers, there's no immediate change here. Frontier Company isn't a consumer product — it's built squarely for enterprise, big-company customers. That said, there could be indirect effects. If you use Unilever products or rely on services connected to Novo Nordisk's medications, you might eventually — indirectly — experience the results of AI-driven processes those companies build with Microsoft's engineers, things like optimized supply chains or accelerated drug development timelines.

For people working in IT, especially in enterprise AI, the impact is much more direct. Think about what this does to the hiring market first. Microsoft organizing 6,000 solution architects, deployment engineers, trainers, and strategists all at once is itself a signal of just how hot demand is for AI implementation talent right now. Since competitors Amazon, Anthropic, and OpenAI all have similar groups, the competition for talent in this space is likely to get even more intense going forward.

For enterprise decision-makers — CIOs, CTOs, that level — this announcement effectively adds a new option to the menu. Until now, adopting AI usually meant either (1) building an internal AI team from scratch, or (2) hiring an outside AI startup or consulting firm. Now there's a third option: a platform company like Microsoft sending in its own trained workforce wholesale. For companies already inside the Microsoft ecosystem — Azure, Office 365, and so on — like LSEG, Land O'Lakes, Unilever, and Novo Nordisk, this option is likely to look especially appealing.

For investors and market watchers, this reads as a signal that Microsoft's AI revenue strategy is evolving to a new stage. Up to now, Microsoft's AI-related revenue has come mostly from Azure cloud usage and Copilot subscriptions. Frontier Company adds a new revenue axis on top of that — something closer to "workforce-as-a-service." Given the initial $2.5 billion investment and 6,000-person scale, it'll take time before this meaningfully moves Microsoft's earnings. But at minimum, it's a clear indicator of just how seriously Microsoft intends to compete in the enterprise AI market.

🥄 Three Things You're Probably Wondering

— Where did Microsoft even find 6,000 people for this? New hires, or existing staff getting reshuffled? Hard to say for certain based on what's been announced. We know all 6,000 go through a six-week AI onboarding program, but whether these are brand-new hires or employees pulled from other parts of Microsoft isn't specified. Given the scale, it's probably a mix of both, but it's too early to say for sure.

— How does pricing work for this? Is it subscription-based like Azure or Copilot? This wasn't addressed in detail in the announcement either. Since this is a people-deployment model rather than a software license, it likely runs on something different from usage-based cloud billing or subscription fees — maybe project-based contracts or consulting-style billing. That's something we'll only be able to confirm once more specifics come out.

— After these first four customers (LSEG, Land O'Lakes, Unilever, Novo Nordisk), who joins next? Since the consulting partners — Accenture, Capgemini, EY, KPMG, and PwC — are tasked with extending this initiative globally, the next wave of customers will likely come through those firms' existing client networks. But which specific industries or regions get targeted next hasn't been announced yet.

References

Numbers are as of announcement and may change.