Microsoft has decided that the hardest problem in enterprise artificial intelligence is no longer building the model. It is getting a Fortune 500 company to actually use one.
On July 2, the company unveiled Microsoft Frontier Company, a new operating business backed by a $2.5 billion commitment and staffed by roughly 6,000 industry and engineering experts whose entire job is to sit inside customer organizations and make AI stick. The move, first reported by TechCrunch and detailed on the official Microsoft blog, reframes the AI race around a decidedly unglamorous problem: deployment. Not who has the smartest model, but who can wire that model into a real company's data, workflows, compliance rules and impatient executives.
Inside the deployment corps
The new unit will be led by Rodrigo Kede Lima, a longtime Microsoft leader who previously ran the company's commercial business across the Americas and Asia. Its experts, a blend of engineers, industry specialists and change-management veterans, will embed directly with customers to build AI systems around Azure, Copilot, autonomous agents and each client's own data.
Judson Althoff, CEO of Microsoft's Commercial Business, described the group's mandate in unusually concrete terms. The 6,000 experts, he said, will work with customers "to co-design, co-innovate, deploy and continuously improve AI systems at scale based on measurable business outcomes." That last phrase, measurable business outcomes, is the tell. After three years of pilots and proofs-of-concept, enterprise buyers want evidence that AI spending shows up on the balance sheet.
Althoff was careful to distance the effort from the industry's fashionable label. "This goes beyond what has been labeled as Forward-Deployed Engineering," he wrote, "and will be the largest, most capable, outcome-driven engineering organization in the industry." Microsoft's existing footprint gives it a running start: the company has already placed engineers inside much of the Fortune 500. Its announcement cites early work with the London Stock Exchange Group, Unilever, Land O'Lakes and Novo Nordisk, and it plans to scale globally through system integrators including Accenture, Capgemini, EY, KPMG and PwC.
The last-mile services war
The timing is not a coincidence. Just two days before Microsoft's announcement, Amazon Web Services committed $1 billion to its own AI deployment venture, explicitly embracing the forward-deployed-engineer model. Earlier this year, OpenAI stood up a subsidiary known as DeployCo, backed by more than $4 billion and placing roughly 150 engineers on-site with customers, while Anthropic launched a comparable effort with Blackstone, Goldman Sachs and other investors aimed at mid-sized firms.
Every major AI vendor, in other words, has arrived at the same uncomfortable conclusion within a matter of weeks: the model is not the product. The deployment is. And that puts Microsoft on a collision course not only with fellow model makers but with the consulting and enterprise-software incumbents who have owned the corporate "last mile" for decades, chiefly Accenture, Salesforce and ServiceNow.
That is what makes the battleground so strange. Accenture is simultaneously a partner in Microsoft's rollout and a rival that has built its own Microsoft forward-deployed-engineering practice. Salesforce and ServiceNow, meanwhile, are racing to convince customers that their agentic AI belongs at the center of enterprise workflows, not Microsoft's. The prize is enormous: whoever controls deployment controls the integration layer, the data, and ultimately the customer relationship. Models are increasingly commoditized and cheap; the human expertise to operationalize them is scarce and expensive.
There is a strategic subtext, too. Microsoft is positioning Frontier Company as a platform-neutral integrator, an implicit contrast with OpenAI and Anthropic, which deploy only their own models through their own firms. That posture carries some irony, given Microsoft's own history and its deep, if cooling, partnership with OpenAI. But it signals where Microsoft thinks the durable margins live: not in renting tokens, but in owning the transformation.
Why deployment beat the model
The shift reflects a hard lesson from the past two years. Enterprises bought AI access in bulk and struggled to convert it into productivity gains that survived contact with the balance sheet. Ford recently rehired veteran engineers after AI fell short, and Deloitte has warned its own consultants that AI is coming for the billable hour. Against that backdrop, 6,000 people who can bridge model capability and messy business reality may prove more valuable than the next benchmark record.
What to watch next
Three questions will define whether Microsoft's bet pays off. First, can it actually hire and retain 6,000 elite experts in a market where that talent is ferociously contested? Second, how long can the partner-and-rival dance with Accenture and the Big Four last before channel conflict boils over? And third, will Frontier Company's "measurable business outcomes" pledge translate into published ROI numbers, or quietly soften into the same vague promises that defined the pilot era? The next few earnings calls, and the first customer case studies, should start to tell.
"This goes beyond what has been labeled as Forward-Deployed Engineering, and will be the largest, most capable, outcome-driven engineering organization in the industry."— Judson Althoff, CEO of Commercial Business, Microsoft