Anthropic will put $100 million into a school for people who deploy its models inside companies. The Claude Frontier Academy is supposed to train 10,000 of what it calls frontier deployed engineers, drawn from the Claude Partner Network, by the end of 2027.
CNBC's report of the Friday release names the first seats: engineers from Accenture, Morgan Stanley, Novo Nordisk, and several consulting firms. The program came out of Anthropic's work installing Claude for customers. Shambhavi Shambhavi, head of strategy and operations for partnerships, told CNBC the request she hears: people who already know the enterprise systems, and who can use the model at the highest level on the problems the business actually has. Corfield's line in the release is that the academy trains people the way Anthropic's own engineers learn, and that graduates should set the standard inside a company.
The residency copies a teaching hospital
Training starts with a simulated enterprise deployment and a graded test. Those who pass enter a residency modeled on medical teaching: instructors, casework, then a credential. Steve Corfield, global head of business development and partnerships, said graduates are meant to set the standard for how AI gets built inside a company. The first engineers are expected to be certified in early 2027.
Anthropic's own IPO filing has already warned that its models could, in the company's words, endanger humanity. The academy is a hiring and training budget sitting next to that warning. Reuters, cited by CNBC from a leaked prospectus, put last year's revenue at almost $4.6 billion and the operating loss at over $8 billion. The $100 million school is a line item beside both.
Draup's figures, supplied to CNBC, are the labor-market side of the same announcement: postings for forward deployed engineers and related AI roles have surged in finance this year. The academy is Anthropic's attempt to mint that worker inside customers it already has, rather than hire all 10,000 onto its own payroll. The end-of-2027 headcount and the early-2027 first class are the two clocks. The $100 million is the budget attached to both.
The prospectus shows a loss larger than the revenue
OpenAI's own forecast still shows a $278 billion hole through 2030. Anthropic's published comparison, via a prospectus Reuters described to CNBC, is smaller and already on the books: almost $4.6 billion of revenue last year, and an operating loss of over $8 billion. The academy's $100 million sits beside that loss, aimed at the people who would install the product.
The $2 trillion figure is the one to keep in its box. CNBC says Anthropic could seek that valuation in a listing later this year. It is not a term sheet, and it is not the $100 million. Reuters' leaked prospectus, as CNBC relays it, is the backward-looking pair: almost $4.6 billion of revenue, an operating loss above $8 billion. The academy is funded at $100 million against that loss, with a headcount target of 10,000 by the end of 2027 and a first graduating group in early 2027. Accenture, Morgan Stanley, and Novo Nordisk are named in the opening class, plus unnamed consulting firms. Draup's data, given to CNBC, says finance job ads for this kind of engineer have surged this year. The residency's structure is the part a buyer of the credential can check: a simulated deployment, a grade, then casework under instructors, borrowed from how hospitals train doctors. Corfield wants the graduates to become the standard inside the companies that already pay for Claude. Shambhavi's version of the same demand is fluency plus knowledge of the business system the model is dropped into.