AINEWS
Amazon's new Forward Deployed Engineering unit embeds AWS engineers in customer teams and bills on outcomes, not hours, taking on Accenture, Deloitte, and McKinsey for the AI services business.
Amazon is putting $1 billion behind a new bet: the hyperscaler who owns the way AI actually gets built inside enterprises will win the next platform race. The new unit, called Forward Deployed Engineering, embeds AWS engineers inside customer companies and prices engagements on outcomes, not hours. It is the clearest signal yet that Amazon has stepped up its challenge to the traditional consultancy model that has dominated enterprise AI rollouts to date.
AWS's $1 billion commitment is the largest publicly disclosed dollar figure from any hyperscaler on an outcome-priced AI services model. None of the traditional consultancies, Accenture, Deloitte, and McKinsey among them, are named in the press release. All of them are the most exposed, because AWS's own framing of the program as "different from traditional consultancy" and focused on "business results rather than billable hours" is itself a critique of the incumbent consulting model.
Instead of selling tools and walking away, AWS engineers sit inside the customer's organization, working alongside their business, engineering, and security teams. They use AI agents to build systems around the customer's data and governance, then oversee and verify the agents' output. AWS calls the pattern the "AI-Driven Development Lifecycle." Behind it sits a semantic layer, a translation tier that gives AI systems consistent access to a company's scattered data sources, deployed inside the customer's own AWS account, enriching metadata and creating a governed data foundation. The customer leaves with functioning AI systems and, AWS says, new engineering skills.
AWS claims the model can compress AI deployment timelines from "months to days." The phrasing is AWS marketing, not an external benchmark. The most concrete on-record example is the NFL: NFL CIO Gary Brantley has said the engagement helped ship NFL Fantasy AI and NFL IQ in weeks, per CNBC's reporting on the launch. Other named customers include the Allen Institute, Cox Automotive, the NBA, Ricoh, and Southwest Airlines, a roster that leans toward data-rich, engineering-deep organizations. AWS says it will pair FDE teams with outside technology partners for model and industry expertise, a hedge that concedes some of the work will not be done in-house.
Traditional enterprise AI services have been sold as per-seat SaaS (subscription software priced per user) or per-hour consultancy, where the bill scales with how long a team takes to deliver. AWS is repricing the unit of account, selling the outcome instead and absorbing the time risk. That is a direct attack on the consulting business model that dominates enterprise AI deployments today. If hyperscalers can deliver outcome-priced AI faster than Accenture or Deloitte can staff an engagement, the consultancies' AI services revenue, the part of their business that has been growing as enterprise AI adoption has accelerated, stops being defensible.
AWS has run a version of embedded delivery inside AWS Professional Services for years. Forward Deployed Engineering has to clear on delivery margin, not headline ARR (annual recurring revenue), and the unit has not disclosed any. The named customers, like the Allen Institute and Cox Automotive, are mostly organizations with internal data and engineering depth, the easier end of the market. The harder question is whether the model survives when AWS is dropping engineers into a mid-market retailer that has neither the data nor the engineering bench to absorb an embedded team. "Thousands" of engineers is also a vague, AWS-disclosed figure. It is not yet clear how many are net-new hires versus reallocated from existing AWS Professional Services, and that distinction matters for whether this is incremental capacity or a rebrand.
How many of the named customers (NFL, NBA, Allen Institute, Cox, Ricoh, Southwest) were already AWS shops before the FDE engagement is the first test. Whether the unit grows headcount or reallocates existing engineers is the second. The consulting industry's response will likely be the loudest signal. If Accenture, Deloitte, or McKinsey launch a parallel "outcome-priced AI services" offering by mid-2027, the threat is real. If they don't, this is another AWS marketing program with a bigger budget.