Anthropic to Spend $100M Training 10,000 AI Engineers

According to CNBC, Anthropic has unveiled Claude Frontier Academy and is investing $100 million to train 10,000 engineers to help companies move generative AI projects from idea to production. The program is designed to tackle a persistent barrier to enterprise AI adoption: finding people who can build and deploy AI in real business environments.

The initial cohorts include engineers from Morgan Stanley, Accenture and McKinsey, Anthropic said. Participants begin with a multiday, in-person program led by Anthropic engineers and coaches, running a simulated end-to-end deployment process. Those who complete that phase enter a 12-week period in which they lead a real Claude-based project inside their own organization. Anthropic expects the first certified participants to be ready in early 2027.

The Frontier Deployment Engineer role

The company intends to certify 10,000 so-called Frontier Deployment Engineers, or FDEs, by the end of 2027. Anthropic presents the curriculum as practical and hands-on: simulation, then guided execution on live projects. The program’s explicit aim is to give organizations the people who can translate prototypes into working systems.

The FDE role, a term popularized by Palantir, involves working directly with customers to adapt technology to their systems, data and operational processes. These engineers help decide what to build, integrate new models with existing technology stacks and create the conditions for staff to use the new systems effectively.

Consulting firms have long placed teams inside client organizations to solve problems, but Anthropic and others say AI raises the technical bar. Companies therefore need engineers who combine technical skill with a practical understanding of how the client’s business operates. Anthropic says it will train 10,000 Frontier Deployment Engineers by the end of 2027.

Anthropic told CNBC that participants start with the in-person simulation and that the first certified engineers are expected to be ready in early 2027.