Anthropic just put a number on the biggest bottleneck in corporate AI: people who actually know how to make it work.
The company is spending $100 million to train 10,000 engineers by the end of 2027 through a new program called Claude Frontier Academy. The first group is coming from Accenture, Bain, Capgemini, Deloitte, McKinsey, Morgan Stanley, Commonwealth Bank of Australia, and the drugmaker Novo Nordisk. These are not junior hires. Companies have to nominate their strongest software engineers, and each one arrives with a specific project they plan to lead once they finish.
The training itself borrows from how hospitals train doctors. Engineers start with several days of in-person training, work through a simulated company deployment, and get graded on a practical test. If they pass, they move into a twelve week stretch where they lead a real project at their own company, with Anthropic engineers checking in along the way. Pass that, and they get a badge that says they are qualified. The first badges go out in early 2027.
The job title behind all this, Frontier Deployed Engineer, did not start with Anthropic. Palantir invented the model over a decade ago: instead of selling software and hoping the client figures out how to use it, Palantir put its own engineers inside client offices to build working tools on the spot. The role became so valuable that Palantir's version of it now pays a median total package well above two hundred thousand dollars a year in the United States, and the company itself is worth over 300 billion dollars.
Anthropic is not the only one copying this playbook. Five months before this announcement, OpenAI launched its own version, spending 150 million dollars and buying two firms, Tomoro and Northslope, to build a team of embedded engineers from scratch. The two biggest AI labs are now fighting over the same narrow pool of people: engineers who can translate a chatbot into something that actually changes how a warehouse, a bank, or a hospital runs day to day.
Here is the part worth paying attention to if you work with any of these consulting firms. At the same time Accenture, Deloitte, and McKinsey are putting elite engineers trained directly by Anthropic in front of clients, those same firms have been cutting junior staff, pointing to AI tools that now do the research and drafting work that used to take a full team. The pattern is consistent: a small number of highly trained people become far more valuable, while a much larger group doing routine work becomes easier to replace.
For any business hiring a consulting firm for an AI project, this changes the right question to ask. It is no longer how many people the firm is putting on your account. It is whether any of them hold this specific credential, because that is where the firm itself says the real capability now sits. The hundred million dollar figure is modest next to Anthropic's own revenue, which has been reported near 47 billion dollars annualized earlier this year. The real prize is not the money. It is locking in which consulting firms, and which of their people, become fluent in Claude specifically rather than a rival's tool.