Microsoft, AWS deploy engineers to accelerate AI returns

Microsoft launched Microsoft Frontier Company this week, a $2.5 billion initiative that will place 6,000 of its own experts and engineers with customers to, in the words of Microsoft Commercial Business CEO Judson Althoff, "co-design, co-innovate, deploy and continuously improve" AI systems. AWS announced a comparable $1 billion investment in AWS Forward Deployed Engineering (FDE), embedding engineers into customers' business, engineering, and security teams. Both moves were reported by Dawn, Free Malaysia Today, and other outlets carrying AFP wire coverage, which framed the investments against a McKinsey finding that, as of the end of 2025, almost nine in ten companies had deployed AI in at least one business function while 94% reported no significant benefit from that spending.

Neither company is first to this model: Anthropic already runs its own FDE-style program, and the broader pattern echoes Palantir's earlier approach of embedding engineers to operationalize complex analytics deployments. CIO.com reporting notes Microsoft says Frontier Company is model-diverse, customers can run ChatGPT, Claude, Copilot, or open-source models, and will lean on systems-integrator partners including Accenture, Capgemini, EY, KPMG, and PwC to scale; early customers cited include the London Stock Exchange Group, Land O'Lakes, Unilever, and Novo Nordisk. AWS's Vasquez described FDE as building for the long term rather than treating each deployment as a standalone project, aiming to move customers from "observers to co-builders to autonomous operators."

Info-Tech Research Group's Thomas Randall told CIO.com that FDE-style programs work by compressing learning curves through vendor product knowledge and reusable processes, and that 77% of organizations still lack a corporate-wide AI strategy, the gap these programs are designed to exploit. The trade-off is architectural: FDE engagements accelerate specific, vendor-aligned builds, while broader multi-system integration across a messy enterprise still requires traditional systems integrators. Teams evaluating either offer should scrutinize data-access and IP terms, whether knowledge-transfer artifacts such as runbooks and architecture docs are contractually guaranteed, and exit provisions, since both vendors have an incentive to deepen platform lock-in even while marketing model choice and customer self-sufficiency.

Analyst Carmi Levy told CIO.com that "both Microsoft and Amazon are aggressively looking for ways to tighten customer lock-in," a tension worth weighing against Althoff's public claim that "customers shouldn't be locked into a single model any more than they should be locked into a single technology vendor." Whether these programs produce public, quantified ROI case studies, rather than the general "measurable outcomes" language used so far, will determine whether they close the gap McKinsey identified or simply shift how enterprises pay for AI integration work.