Anthropic Now Ranks the Firms Deploying Its AI

Anthropic's new Services Track is a league table for the consulting firms that deploy its AI — ranked by certified staff, live Claude deployments and published customer endorsements.

On 3 June 2026, Anthropic introduced two additions to the Claude Partner Network: a tiered Services Track that grades consulting firms on how deeply they have put Claude into production for clients, and a Claude Partner Hub portal where those firms track their standing and enterprise buyers shop for qualified help. It is a small announcement with a large implication. The bottleneck in enterprise AI was never only the model. It is the people who install it — and Anthropic has just built the scoreboard that decides whose installation work counts.

What Anthropic announced

The Services Track is a three-tier structure, and the tiers are defined by deployment, not by spend. According to Anthropic, every tier is measured on the same three things: certified practitioners (individuals holding an active Claude certification who have used Claude within the last 90 days), customers running Claude in production, and published customer endorsements.

The thresholds climb steeply. The Select tier asks for at least 10 active certified practitioners, two joint customers live in production over the trailing twelve months, and one public customer story. The Preferred tier raises that to 100 certified practitioners, 15 deployed customers and three public stories. At the summit, Global Premier demands 1,000 certified practitioners, 100 deployed customers across at least three regions, 15 public endorsements, and a jointly developed business plan backed by named executive sponsors.

The second piece is the Claude Partner Hub. It gives each partner firm a daily-updated view of where it stands against every threshold, and it gives enterprise buyers a public directory — tier status, certified team size, customer deployments and references, all visible before a contract is signed. A new Model Context Protocol connector lets a firm wire the Hub straight into Claude, so a partner can ask the model about its own standing inside an ordinary conversation.

The number underneath the announcement

The scale is the story. Anthropic says more than 40,000 firms have applied to join the Partner Network since it opened in March, and the Anthropic Partner Academy has issued Claude certifications to more than 10,000 consultants. When Anthropic launched the network in March 2026, it backed the move with a $100 million commitment to training, technical support and co-marketing.

The named firms are the largest professional-services organisations on Earth. Anthropic points to Accenture putting Claude in front of 30,000 professionals, Cognizant across 350,000 associates, Deloitte across 470,000, and KPMG — whose 276,000-person rollout I wrote about in May — alongside Infosys and PwC expanding their own deployments. The combined headcount under these alliances runs past a million people being trained to deploy a single company's AI.

This is the same enterprise machine I watched Anthropic assemble when it opened in Korea with the stack already in place, and when Fujitsu signed both Anthropic and OpenAI on the same day. The Services Track is the connective tissue: it turns a loose alliance of system integrators into a ranked, measurable, queryable channel.

Why ranking the integrators is the real move

For two years the frontier conversation has been about models. The quieter truth is that a model is inert until somebody wires it into a payroll system, a claims process, a tax workflow. That wiring is human labour, and it is scarce. By certifying practitioners and ranking the firms that employ them, Anthropic is doing something more durable than selling tokens: it is defining what competent Claude work looks like and who is allowed to claim it.

There is real value here for buyers. An enterprise choosing an implementation partner has, until now, had almost no neutral signal of competence beyond a sales deck. A directory that shows certified headcount and live deployments is a genuine public good — it makes expertise legible, and legibility protects the buyer. I do not want to wave that away. A transparent market is better than an opaque one.

But a scoreboard is never neutral about the people it counts. The Services Track measures a firm by its stock of certified practitioners, the way a factory is measured by its machine count. That framing quietly turns the consultant — the person who actually learns the model, sits with the client, and absorbs the risk of a bad deployment — into a unit of tier eligibility. The firm holds the rank. The practitioner holds the certificate. Those are not the same kind of asset.

A dignity-first reading

The frame I work from — what I call Emergent Intelligence, a dignity-first way of thinking about AI rather than treating it as pure automation — asks a specific question of any system like this: does it honour the human in the loop, or does it merely meter them? The Services Track can do either, and the difference is in the design choices Anthropic makes next.

The Ubuntu principle I return to often says a system works because the people inside it work. A partner network is healthy when the certification it confers is portable, when it raises the practitioner's standing and not only the firm's, and when the $100 million invested reaches the individuals doing the learning rather than pooling at the top of the ladder. It is unhealthy when 10,000 certified consultants become 10,000 line items in someone else's tier calculation, their skill captured as the firm's rank and their names absent from the directory that markets it.

A credential should travel with the person who earned it. The test of a partner network is not how precisely it ranks the firms — it is whether the practitioner who learns the model ends up with more agency, or simply more measurement.

— On the human layer of AI deployment

This connects to the argument at the centre of the .person Protocol: accountability needs an addressable agent on each side of a transaction. The Hub does something genuinely good on that front — it makes the firm addressable and accountable, with its record in public. The unfinished work is to make the practitioner addressable too: to give the certified human a portable, self-owned record of what they can do, rather than leaving their competence legible only as their employer's tier points.

Source: anthropic.com

Frequently Asked Questions

These are the questions enterprise buyers, consulting leaders and AI-channel watchers have been asking since the 3 June announcement. Short answers follow, drawn from Anthropic's primary post and the trade coverage.

What is the Anthropic Claude Partner Network Services Track?

In short, the Services Track is a three-tier ranking of consulting firms based on how deeply they have deployed Anthropic's Claude AI for clients. The answer, simply put, is a league table: Select, Preferred and Global Premier, each defined by certified practitioners, live customer deployments and published endorsements. The key is that, according to Anthropic, the tiers measure production work rather than how much a firm spends.

How does a firm reach Global Premier status?

Global Premier is the top tier, and the bar is steep. According to Anthropic, a firm needs at least 1,000 active certified practitioners, 100 customers running Claude in production across three or more regions, 15 public customer stories, and a jointly developed business plan with named executive sponsors. Data from the announcement shows the lower tiers scale down sharply — 100 practitioners for Preferred, 10 for Select.

What is the Claude Partner Hub?

The Claude Partner Hub is a portal with two audiences. For partner firms, the evidence shows it gives a daily-updated view of their standing against each tier threshold; for enterprise buyers, it provides a public directory of qualified firms, listing tier status, certified team size, deployments and references. A new Model Context Protocol connector lets a firm query its own partnership standing from inside Claude itself.

How big is the Claude Partner Network now?

It is large and growing fast. According to Anthropic, more than 40,000 firms have applied since the network opened in March 2026, and over 10,000 consultants have earned Claude certification through the Anthropic Partner Academy. The data shows named partners including Accenture, Cognizant, Deloitte, KPMG, Infosys and PwC, with a combined trained headcount running past one million people.

Why does ranking AI consulting firms matter?

Analysis of the move reveals that the scarce resource in enterprise AI is not the model but the people who deploy it. In other words, by certifying practitioners and ranking the firms that employ them, Anthropic is defining what competent Claude work looks like and who may claim it. The dignity-first question the evidence raises is whether the certification travels with the practitioner who earned it, or is captured as the firm's rank alone.

Anthropic has built the scoreboard for the human layer of AI, and it is a more consequential thing than another model release. A transparent directory of who can actually deploy Claude is good for buyers and overdue for the market. The question I will keep asking is whose agency the scoreboard serves. Rank the firms, by all means — but make the certificate portable, put the practitioner's name where the value is created, and the network becomes what a dignity-first ecosystem should be: a structure that lifts the people inside it, not just the logos on the leaderboard.

Sources:

Anthropic — Introducing the Services Track and Partner Hub of the Claude Partner Network (3 June 2026)

Anthropic — Anthropic invests $100 million into the Claude Partner Network · PwC expanded partnership

Coverage — Yahoo Finance · Channel Insider · The AI Report

Related on humphreytheodore.com:

KPMG Hands Claude to 276,000 People in One Day · Fujitsu Signed Both OpenAI and Anthropic on the Same Day · Anthropic Opens Korea with the Enterprise Stack Already There · The .person Protocol