Anthropic’s $100 Million Academy Trains the Firms That Bill Both Sides

Anthropic is putting $100 million into Claude Frontier Academy to train 10,000 Frontier Deployed Engineers by the end of 2027. The first class is already sitting in San Francisco, New York and London.

The seats do not belong to a public course. They belong to people nominated by Accenture, Bain, Capgemini, Deloitte, McKinsey, Morgan Stanley, Novo Nordisk and Commonwealth Bank of Australia. Three of those consultancies also put money into OpenAI’s deployment company in May.

Anthropic Commits $100 Million to Train Client Engineers

The company published the plan on October 2, 2026. In its $100 million commitment to train 10,000 engineers, Anthropic calls talent the hard part of getting Claude from an idea into a live system. The 10,000 figure is a target, not a headcount of people already trained.

Steve Corfield, global head of business development and partnerships, said the academy copies how Anthropic’s own engineers learn. He wants graduates to set the bar for how AI is built inside a business.

No AI company has invested in developing that talent inside its customers and partners at this depth. 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.

Steve Corfield, Global Head of Business Development and Partnerships, Anthropic

Participation is by nomination. Each engineer arrives with a named Claude project to lead after the classroom work. Prior work building AI agents is not required. The residency is in early access with a select group of customers and partners, and Anthropic says it will grow the program to meet demand.

Enterprise contracts are already the part of the Claude business that large software vendors point to when they explain their own numbers, including a recent profit jump tied to Anthropic deals. Training the people who wire those contracts into production is how that revenue is supposed to stick.

Accenture, McKinsey and the First Nominated Class

Five of the eight named organizations sell advice and delivery. Three run their own operations: a bank, a drugmaker and an investment bank. The mix is the tell. Anthropic is not only teaching in-house staff at Morgan Stanley or Novo Nordisk. It is stocking the firms those companies hire.

THE FIRST NOMINATED CLASS

Accenture described the residency as a test of judgment against a simulated enterprise rollout, not of technical skills in isolation. Capgemini said clients care less about training volume than about people who have solved live problems. McKinsey said the residency extends its technologists’ toolkit for client work, “hands-on-keys.”

Commonwealth Bank of Australia said its teams have produced up to 3x more code changes in the past year and that the residency takes that work deeper. Novo Nordisk said it already uses Claude Science in research and wants its engineers to get the same hands-on training behind Anthropic’s own deployments.

Those operating companies will keep some of the skill on payroll. The consultancies will sell it. That is the part of the class list the launch post treats as a customer roster and that a buyer should read as a channel.

Three Consultancies Already Fund OpenAI’s Deployment Arm

On May 11, 2026, OpenAI launched the OpenAI Deployment Company with more than $4 billion of initial investment from 19 firms. The unit is majority-owned by OpenAI. Its job is to embed Forward Deployed Engineers inside customers, not to badge the customer’s own staff.

OpenAI listed Bain, Capgemini and McKinsey as investors in that company, alongside TPG as lead and Advent, Bain Capital and Brookfield as co-leads. Those three consultancies are also in Claude Frontier Academy’s first cohort. Accenture and Deloitte are in the academy and were not named on OpenAI’s investor list.

WHO SITS ON BOTH SIDES

OpenAI also agreed to acquire Tomoro, an applied AI firm, to bring about 150 Forward Deployed Engineers and deployment specialists into the new unit. That deal was still subject to closing conditions when OpenAI announced it. The Deployment Company’s investment and consulting partners sponsor more than 2,000 businesses, OpenAI said.

The two labs are buying the same scarce job with different balance sheets. OpenAI is putting hired engineers on its own paper. Anthropic is paying to train people who already sit on a consultancy’s or a client’s paper. McKinsey, Bain and Capgemini can collect from both designs. A client that wants a “Claude-ready” team and a client that wants an OpenAI embed can be billed by the same firm.

Residents Drill on a Fake Company, Then Ship a Real One

The academy’s first program is the Frontier Deployed Engineer Residency. Anthropic models it on medical training: learn from people who already do the job, practice on real cases, and sit an assessment before working unsupervised. The bottleneck this is built for is not model choice. It is the small group of people who can pick a use case, get it through security review, and leave someone owning the system after handover.

HOW THE RESIDENCY IS BUILT

1. Nomination: The employer names an engineer and a Claude project that person will lead.
2. Four-day intensive: Three days in person building a Claude system for a simulated company, from the first request through security review to handover; the fourth day is a graded practical on a new scenario.
3. Resident badge: Those who pass earn the Claude Resident Engineer badge and enter the live job.
4. Twelve-week residency: The engineer leads a real Claude use case at their own organization, with cohort mentorship and support from Anthropic engineers.
5. Final practical: A second assessment can earn the Claude Frontier Deployed Engineer badge. Anthropic expects to award the first in early 2027.

The program page calls that opener a four-day in-person intensive. Additional Academy programs are slated to be announced in 2027. The $100 million is described as an initial investment that starts with this residency, so it is not a tuition figure for the 10,000 target alone.

Partner Badges Already Outrun the New Residency

Claude Frontier Academy sits on top of a partner machine Anthropic has been building all year. In March 2026 it launched the Claude Partner Network with a separate $100 million investment. On October 2 it said professionals across 175,000 Claude certifications and 46,000 firms, and that nearly 4,000 people had finished Basecamp, an immersive course that uses the same curriculum Anthropic’s Applied AI teams get when they join.

The July 23 expansion of role-based exams is where the large firms put numbers on their benches. Those figures are pledges, not completions, and they dwarf the academy’s 10,000-engineer aim.

PUBLISHED CERTIFICATION PLEDGES

McKinsey and Bain appear in the October residency class and did not appear with a headcount on that July pledge list. The academy is the elite layer: nomination, a live project, two practicals, a badge Anthropic says matches the standard of its own engineers. The Partner Network is the wide layer: proctored exams, partner tiers, and delivery teams counted in the tens of thousands.

A Global Premier partner on that network needs 1,000 certified practitioners, 100 customers across three regions, and 15 public customer stories. The residency does not replace that ladder. It gives the people at the top of it a harder credential to sell.

What a Claude Credential Changes for the Client

For Morgan Stanley, Novo Nordisk and Commonwealth Bank of Australia, a staff engineer who finishes the 12 weeks is supposed to stay. The knowledge is meant to live on the company’s own systems, which is the pitch against flying in a lab’s people and watching them leave.

For Accenture, Deloitte, McKinsey, Bain and Capgemini, the same badge travels with the firm. The engineer who cleared a simulated security review in New York can be staffed onto the next Claude job in another city. The client pays for that person twice: once as a vendor invoice, and again as switching cost, because the people who know how the last rollout was wired may not be on the payroll.

OpenAI’s answer is to own more of those people directly, starting with the Tomoro team it agreed to buy. Anthropic’s answer is to mint them inside the channel and inside a few large customers, then let the badge do the distribution work. The consultancies that took both offers are not confused. They are the market for deployment skill, and both labs are paying them to hold it.

Anthropic says it will announce more Academy programs in 2027. The first Frontier Deployed Engineer badges are due in early 2027, after the 12-week jobs now under way in San Francisco, New York and London.