Anthropic is spending $100 million to train 10,000 engineers inside other companies, betting that getting AI actually working inside large organizations is now a bottleneck worth solving at institutional scale.
Announced Friday under the name Claude Frontier Academy, the program targets what Anthropic calls Frontier Deployed Engineers, or FDEs, people who can take Claude from a concept to a production system in someone else’s business. The goal is 10,000 of them by the end of 2027. Cohorts are already running in San Francisco, New York and London, and participation is by nomination rather than open enrollment.
Nominations come from the companies themselves, which pick engineers with strong fundamentals, a track record of building with large language models, and a named Claude project to lead when they go back. Each nomination passes two graded assessments before acceptance, and the curriculum borrows deliberately from medical training: multi-day instruction followed by a twelve-week on-the-job rotation, on the theory that enterprise AI work, like clinical work, is something you learn by doing under supervision before you do it alone.
Who is actually in the room
The first eight organizations putting engineers through the program read like a map of how enterprise AI budgets get signed off: Accenture, Bain, Capgemini, Commonwealth Bank of Australia, Deloitte, McKinsey, Morgan Stanley and Novo Nordisk. Five of the eight are consultancies or IT service firms rather than technology companies, and that mix is the point. Anthropic is not building an internal talent bench, it is seeding the intermediaries whose recommendations then cascade out to thousands of client companies.
Accenture’s participation illustrates the leverage. The firm has said it intends to use the Academy’s resources to train roughly 30,000 of its own professionals on Claude, about three times the Academy’s total headcount target, via its own internal programs. If even a fraction of those employees end up recommending or deploying Claude in client work, Anthropic’s $100 million is buying distribution rather than merely education.
Anthropic’s global head of business development and partnerships, Steve Corfield, framed the investment in those terms. “Claude Frontier Academy trains people the way our own engineers learn, and we want those who graduate to set the standard for how AI gets built inside a business,” he said in the announcement.
Competing for the same engineers
The move slots into a wider land grab for AI implementation talent. Anthropic started its Claude Partner Network in March 2026 with an initial $100 million and had logged more than 175,000 certifications by this week, while OpenAI announced a partner effort in June with a $150 million commitment targeting 300,000 certified professionals by the end of 2026, including a pilot program for forward-deployed experts. Each program trains a slightly different type of worker, but they all converge on the same market: enterprises unsure how to actually use frontier models and willing to pay whoever can credibly guide them.
The gap in these numbers is worth noting. Ten thousand rigorously trained engineers is a rounding error next to 300,000 certifications, and that is deliberate. OpenAI’s program is built for volume and certification breadth, Anthropic’s is built for depth and placed acceptance into named companies. Whether depth or breadth wins is likely to depend less on program quality than on enterprise budget cycles over the next 18 months.
It also lands against a tighter labor backdrop than most AI training programs have faced. The U.S. added just 29,000 jobs in September against forecasts near 90,000, per Friday’s Bureau of Labor Statistics release, and unemployment touched 4.2%. A company investing $100 million in training others’ engineers is, implicitly, a bet that firms will pay to arm the people they already have rather than compete in an open market for a small pool of AI-native hires.
What it costs, and who pays
Anthropic has not disclosed how the $100 million is allocated across instruction, subsidies for nominating firms or program staff, so it is too early to say whether the residency is free to participants or subsidized at a discount to commercial rates. What is clear is the structure: organizations nominate engineers, Anthropic handles the training pipeline, and the cost of sending an employee away for weeks of work falls to the nominating firm, not to Anthropic.
Enterprises looking to take part have one practical channel: they can ask their Anthropic account team or partner account manager whether they are eligible. Anthropic says cohorts will grow from the three cities currently running toward the 10,000-engineer target through the end of 2027, with further locations implied but not confirmed.
For Anthropic, the bet also hedges a financing story that has been growing more aggressive. Its expanded partnership with Google and Broadcom announced in April involves multiple gigawatts of next-generation compute, and reporting on its IPO filing has put Broadcom on the hook for up to $42 billion in lending against a $125.2 billion five-year TPU lease commitment, a figure large enough that Reuters framed it as the AI lab financing infrastructure like a utility rather than a startup.
That context raises the stakes for revenue, because debt service does not wait for enterprise adoption curves to play out. Spending $100 million to seed enterprise demand is cheap by comparison, and it gives Anthropic’s $30 billion run-rate revenue, up from roughly $9 billion at the end of 2025, a deeper pipeline of customers who already have people on staff who know how to use the product. It also creates a second-order effect worth watching: each trained FDE is now an internal advocate for a specific vendor’s stack, which makes switching costs in enterprise AI higher than they were a year ago.
There is a risk on the other side of the ledger. Enterprise transformation programs led by consultants have a mixed record historically, and a twelve-week rotation is short next to the multi-year ERP and cloud rollouts these same firms typically run. If cohort graduates find their recommendations stalling in procurement, the Academy’s 10,000 could produce fewer deployed systems than the headline suggests. Anthropic is betting the moved needles are worth the spend either way, since even partial deployment shifts enterprise mindshare at a moment when the company is competing for the biggest infrastructure commitments anyone in the industry has ever signed.