Summary: AWS committed $1 billion on June 30, 2026 to a Forward Deployed Engineering organization led by Francessca Vasquez. Its engineers work inside customer companies with business, engineering, and security teams to move agentic AI systems to production grade. Named customers include the Allen Institute, Cox Automotive, NBA, NFL, Ricoh, and Southwest Airlines.
On June 30, 2026, AWS committed $1 billion to standing up a dedicated Forward Deployed Engineering (FDE) organization, to be run by Francessca Vasquez, VP of Frontier AI Engineering and Services. These engineers will not be sitting at Amazon desks; they will move into customer companies and work shoulder to shoulder with each client’s business, engineering, and security staff, bringing agentic AI systems up to production grade, with deployment efforts that once ran in months meant to be measured in days instead.
The timing is impossible to overlook: OpenAI and Anthropic each stepped into enterprise AI services barely two months ago. With this move, AWS has said its piece out loud — the cloud giant now treats the next currency of enterprise AI as “delivering outcomes” rather than “selling tools.”
Where the FDE Model Comes From
It was Palantir that put this term on the industry map. Under the arrangement, engineers stay posted at the client for as long as the system takes to build, sort out whatever goes wrong on the spot, reuse large pieces of the underlying technology across projects, and then reshape everything to match the way each individual company operates. The customer wins twice over: results arrive quickly, and the know-how ends up rooted in the customer’s own team. The vendor, in exchange, has to carry a large staff of engineers working on site — look closely and this is, at bottom, a business powered by headcount.
AI agents stepping into the enterprise have given this well-worn playbook a fresh round of attention. Models are on the market and GPU capacity is available for rent almost anywhere; the wall most companies actually hit is attaching agents to the processes, data, and governance structures they already operate on. TechCrunch’s read on the trend: more and more enterprises are willing to pay for someone to “install it for me,” and dedicated service organizations have sprung up to answer exactly that call.
How the $1 Billion Gets Spent
Amazon has been blunt about the money: the $1 billion is a commitment of internal resources, structured as neither a joint venture nor an investment in the everyday sense. That one decision alone draws the sharpest possible line between AWS and the OpenAI and Anthropic playbooks.
The working method looks like this: engineering teams enter the customer’s environment together with purpose-built agents and run what is called an “AI-Driven Development Lifecycle” — agents accelerate each phase, and human engineers review the output and hold the steering wheel. A deployment stands up a semantic layer within the customer’s own AWS account, wires it to enterprise data, and publishes a governed, versioned knowledge graph; the intent is for domain expertise to reside in the customer’s systems instead of in people who will eventually move on. There is no hourly rate; engagements are designed around business outcomes the two sides share. Customer engineers advance “from observers to co-builders to autonomous operators,” and when the project closes out they are expected to keep the operation running by themselves. Security spans hardware-based isolation and end-to-end encryption, and the data stays within the customer’s governance framework the entire time.
Named Customers and First Results
The organization opens with history behind it: BMW, Jabil, and Lyft had all previously worked with AWS’s Generative AI Innovation Center. As for the new organization, the customers it has named are the Allen Institute, Cox Automotive, the NBA, the NFL, Ricoh, and Southwest Airlines. Gary Brantley, the NFL’s CIO, said the partnership took two fan-facing products, NFL Fantasy AI and NFL IQ, into production “in just weeks.”
Partners are folded in as well: model and industry expertise will come from partners, while AWS invests in partner training and tools. The customers being courted are organizations that have already left the experimentation stage behind — with regulated industries, financial services, and the public sector at the top of the list.
Three Routes, Three Wagers
Rewind to May: OpenAI, under the banner “The Deployment Company,” planned a $4 billion raise at a $10 billion valuation, and TPG, Brookfield, Advent, and Bain Capital all appeared among its backers. Hours afterward, Anthropic put together a joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs — the Wall Street Journal valued it at $1.5 billion, with $300 million committed by each of Anthropic, Blackstone, and Hellman & Friedman. Beneath both plans sits one shared calculation: draw on private equity’s capital, convert the companies in its portfolios into customers, and a fresh enterprise sales channel opens.
AWS chose a third way: it raises nothing from outside investors, splits nothing with private equity, and instead stakes its own balance sheet and the enterprise customers it already has. The wagers point in opposite directions. Amazon is betting that scale and integration win — an FDE organization attaches naturally to AWS accounts and governance tooling — while the two labs are betting on channel innovation. Which side turns out right is still anyone’s guess, but all three already stand together on one belief: the next wave of enterprise AI revenue will come from services, not from subscriptions alone.