Palantir invented the forward-deployed engineer model more than a decade ago. The firm’s engineers became known for living inside government agencies and Fortune 500 companies for months at a time, not to train employees on software, but to build systems alongside them from the inside.
On June 30, Amazon decided to take that model and run it at cloud scale.
AWS announced a $1 billion investment in a new Forward Deployed Engineering organization, seeded with thousands of engineers who will embed in small pods directly inside client companies for intensive engagements designed to get AI into production fast.
What AWS’s $1B FDE unit actually involves
Francessca Vasquez, AWS VP of Frontier AI Engineering and Services, announced the new unit and walked through how it works.
Engineers go into a client company in pods of five or six people. They stay for roughly 45 days, working inside the client’s own environment on the client’s own data, alongside the client’s business, engineering, and security teams.
Vasquez said the unit is not trying to create ongoing dependency. When the team leaves, the customer owns everything: the code, the AI agents, the workflows, and the internal knowledge to keep running without AWS staff on-site.
Related: Amazon challenges Costco with July 4 gas savings deal
“The currency that the customers are always talking about right now is speed. We do see FDE being a choice for customers who are looking for accelerated value back to their stakeholders, their customers, their executive teams,” Vasquez told CNBC.
The method AWS is using is called the AI-Driven Development Lifecycle. Human engineers oversee AI agents that handle the software writing and system deployment work. The idea is to shrink what normally takes a company months to do down to a matter of days.
The six organizations already working with FDE teams include the Allen Institute, Cox Automotive, the NBA, the NFL, Ricoh, and Southwest Airlines.
Why the FDE model is spreading across enterprise AI right now
The FDE idea has existed in tech for years but it is having a particular moment in 2026 because of where the AI market stands.
Companies spent 2023 and 2024 running pilots and proof-of-concept projects. Most of those experiments sat on servers and never touched real operations.
The gap between “AI project” and “AI in production” turned out to be much wider than most executive teams expected.
What FDE teams address is precisely that gap. A dedicated external engineering team, working inside the client’s actual infrastructure with access to real data and real constraints, can get things moving in ways that consulting decks and software demos cannot.
The model has shown up at software firms of all sizes looking to drive faster adoption of their tools, and the race to deploy enterprise AI has made it the dominant go-to-market strategy of the moment.
Berger/Getty Images
How AWS’s approach differs from OpenAI and Anthropic
AWS is not the first AI company to move in this direction. OpenAI and Anthropic both launched FDE offerings earlier in 2026.
OpenAI structured its venture with TPG, Advent International, Bain Capital, and Brookfield, and it was valued at $4 billion. Anthropic built its deployment company in May 2026 with Blackstone, Hellman & Friedman, and Goldman Sachs, valued at $1.5 billion. Both are joint ventures with outside investors and consulting partners attached.
Amazon is writing this one itself. The $1 billion comes from its own balance sheet, with no co-investors and no outside consulting firms.
Vasquez said it is also the first time AWS has pulled its various engineering capabilities into a single unit with a shared deployment framework.
“We’ve had capabilities over the years, but structurally this is like getting everybody together in one business unit with a common rubric of deployment,” she said. “It’s the first time we’re doing it in that way.”
More Amazon:
- Amazon Prime Day gives Wall Street a $22B reason to take notice
- Amazon quietly building a moat to outlast the AI boom
- Bank of America resets Amazon stock forecast on key service launch
AWS is also the first major hyperscaler to announce this kind of initiative, according to TechCrunch. Google has made its own move in enterprise AI deployment, but through a $750 million partner fund aimed at agentic AI rather than an internal engineering corps.
“Customers leave AWS FDE deployments with both new solutions and new engineering capabilities. Along with agentic systems running in their own AWS environment, they gain lasting AI skills, workflows, and patterns they can use to innovate independently,” Vasquez wrote in the AWS announcement.
What this deal means for customers, investors, and cloud competition
The problem AWS is trying to solve is one most large companies know well. They have access to AI models. They have budgets approved for AI projects. What they do not have is the internal engineering depth to take any of that from a proof of concept into something their operations can actually run.
An external team embedded inside their environment for six weeks, working on their actual systems with their actual data, is a faster fix than hiring and training.
What happens when the AWS team leaves is the part that makes the model interesting. The engineers hand over the code and the agents they built, but they also leave behind the internal knowledge, the documented patterns, and the skills transfer that let the client keep building on their own. Clients own everything.
Related: Bank of America spots Prime Day signal for Amazon investors
For AWS’s cloud market position, six weeks inside a client’s environment building production systems alongside their teams does something that selling compute and storage never quite does.
It makes switching providers harder, in a way that the client probably does not fully appreciate until they try. The agent architecture, the integration patterns, the deployment setup — all of it is built to run on AWS. Moving it somewhere else means rebuilding from scratch.
Amazon holds financial stakes in both Anthropic and OpenAI. It now has a unit competing with both of them for the same enterprise deployment contracts.
For investors tracking AWS, the questions worth watching are whether the FDE model accelerates enterprise contract sizes, how quickly AWS can scale thousands of engineers without quality dropping, and whether the self-sufficiency promise holds up when customers try to operate independently after the 45-day window closes.
The company’s reading is that selling access to an AI model is only the first part of the opportunity. Getting that model running inside a customer’s actual business is where durable revenue gets built.
More on Amazon & its stock:
- History of Amazon: From garage startup to tech titan
- Is Amazon a good long-term investment? Its buy-and-hold prospects explained
- Amazon’s dividends and stock splits: What you need to know
- Amazon’s stock buybacks explained
- What is Amazon’s free cash flow in 2026?
- Who owns Amazon? Top executives and institutional investors
- Where are Amazon’s headquarters? Seattle and beyond
- How many employees does Amazon have in 2026? Its workforce explained