Anthropic Just Hired Wall Street to Kill McKinsey. $1.5B and 'Forward-Deployed' Engineers.

The headline that should have made every McKinsey partner spit out their coffee on Monday morning was buried in a press release. Anthropic, Blackstone, Hellman & Friedman, Goldman Sachs and General Atlantic just put $1.5 billion into a new firm with one job: replace the consultant. Not augment them. Not give them a copilot. Replace them.

The deal landed on May 4, 2026. Anthropic and Blackstone each committed $300M. Hellman & Friedman matched at $300M. Goldman Sachs and General Atlantic put in $150M each. Apollo, Leonard Green, GIC and Sequoia rounded out the cap table. Hours later, OpenAI announced a parallel joint venture aimed at the same prize. The race to dismantle the $200B consulting industry started before lunch.

The mechanism is what makes this different from every previous AI-for-enterprise pitch. The new entity does not sell a license. It sells forward-deployed engineers — people who fly to your factory, sit at the desk next to your operations VP, and ship Claude-powered systems that automate work the consultants used to recommend. Palantir invented this model selling to the Pentagon. Anthropic just bought the playbook and pointed it at the Fortune 5000.

Why Wall Street wrote the check

Blackstone manages roughly $1.1 trillion. Hellman & Friedman sits on about $115B. Together they own controlling stakes in hundreds of mid-market portfolio companies — the kind of business with $200M to $2B in revenue, a CFO who still uses pivot tables, and a consulting line item bigger than the IT budget. These companies are the dream customer for embedded AI services. They have real money. They have real inefficiency. And their owners have a fiduciary obligation to extract every last point of EBITDA.

Anthropic's annual recurring revenue hit roughly $19B in early 2026. OpenAI is closer to $25B. Both companies have run out of casual API customers and need a way to capture the enterprise dollars that currently fund Accenture's $64B revenue line. The math is brutal: a single Big Three engagement runs $5M to $50M for a six-month rollout. Replace 5% of those with embedded engineers running Claude, and you have built a $10B revenue stream with software margins on top of it.

The Goldman investment matters for a separate reason. Goldman is not just writing a check — the firm is the prototype customer. Anthropic engineers have already been embedded inside Goldman's banking and asset management businesses for over a year. The bank is the proof point. If the model works inside the most paranoid client base on earth, it works anywhere.

Ranking the 7 industries about to get gutted

Most coverage of the announcement focused on consulting itself. That misses the point. Consulting gets repriced. The industries the new firm walks into get gutted. Here is the disruption tower, ranked by AI-replaceability, headcount exposure, and PE ownership concentration — the three variables that decide who gets restructured first.

1. Healthcare services and revenue cycle management

Top of the tower, and it is not close. PE owns vast swaths of physician practice management, revenue cycle outsourcers, and billing operations. The work is rules-based, regulated, and currently performed by hundreds of thousands of human reviewers. Anthropic's Claude 4 family already handles prior authorization, denial appeals, and coding audits at parity with senior billers. A forward-deployed team can erase 40% of the headcount cost inside a single fiscal quarter. PE owners will sign that contract before they finish reading it.

2. Financial services back office

Goldman is not the customer because Goldman is broken. Goldman is the customer because the rest of the industry is. Mid-tier banks, insurance carriers and asset managers spend roughly 40 cents of every operations dollar on KYC, AML, claims adjudication and regulatory reporting. All of it is text-in, text-out work that Claude does at human accuracy and 1/30th the cost. The combined PE exposure across financial back-office firms is in the tens of billions in annual labor spend. The new JV will have a dedicated vertical here within 90 days. Bet on it.

3. Infrastructure and engineering services

Construction management, engineering review, environmental compliance — the unsexy plumbing of every PE-owned infrastructure roll-up. Each project generates thousands of pages of PDFs that get manually reviewed by senior engineers billing $400 an hour. Forward-deployed teams plug Claude into the document pipelines and turn six-week reviews into six-day reviews. The labor savings are smaller in absolute headcount than healthcare, but the margin uplift per project is higher. This is where the JV books its first lighthouse case study.

4. Manufacturing and industrial operations

This one ranks high on opportunity, lower on speed. Industrial operations have AI-ready data, but the cycle time to deploy is longer because real-world equipment integration takes months. Expect the JV to run pilots at Blackstone-owned industrial portfolio companies through 2026 and start booking material revenue in 2027. The endgame is autonomous procurement, predictive maintenance scheduling and supplier risk management run by agent fleets. Net headcount impact: 15-25% of indirect labor over three years.

5. Retail and consumer brands

PE owns hundreds of consumer brands and specialty retailers. The work that gets automated here is merchandising analytics, customer service tier one, returns processing and supplier negotiation. Lower margin per engagement than banking, but higher count of engagements, which is exactly what a forward-deployed model needs to scale repeatable playbooks. Watch for an off-the-shelf Retail Operations Module from the JV by Q3 2026.

6. Real estate operations

Blackstone is the world's largest real estate owner. Its portfolio runs on tenant servicing, lease administration, asset valuation and property management — all heavily document-driven, all highly automatable. The political dynamic inside Blackstone makes this an obvious early-customer pilot. The JV will use Blackstone's own portfolio as the demo, then sell the resulting playbook to every other real estate sponsor on the planet.

7. Logistics and transportation

Last on the tower because the data fragmentation problem is real. Logistics operations span dozens of carriers, ERP systems and EDI feeds that have not played nicely with each other since 1998. Claude can absolutely handle the complexity, but each deployment is bespoke. Expect the JV to land a few flagship logistics customers, generate impressive case studies, and let the consulting market figure out the rest.

The surprise: consulting itself is not number one. Consulting gets repriced. McKinsey, BCG and Bain still get hired — just at lower rates and for different work. The industries above are the ones that will see actual headcount reduction.

What Anthropic is really buying

The strategic prize for Anthropic is not the services revenue. It is the workflow data that comes with embedded engagements. Every forward-deployed team generates a high-fidelity record of how real enterprise work actually flows — what the inputs look like, where the bottlenecks sit, which approvals get rubber-stamped, which ones do not. That data is the training fuel for the next generation of agents that ship without an engineer at all.

This is why the structure is a joint venture and not an acquisition. Anthropic does not want to build a 5,000-person services firm. It wants to use Blackstone's portfolio access and Goldman's enterprise credibility to harvest the workflow corpus, then productize it. In three years, the JV's headcount stops growing and the agent revenue compounds. That is the bet.

The OpenAI parallel announcement on the same day tells you Sam Altman saw the same thing. OpenAI's services play is reportedly larger in headcount but narrower in PE access. Anthropic got the better cap table. OpenAI got the bigger first-year revenue projection. Both will ship the same quarter, both will close the same logos, and the consulting industry will learn what enterprise software went through in 2010.

What this means if you sit on either side

If you are a junior or mid-level consultant: your billable rate is about to compress 30-50%. The work you have been doing — market scans, slide stacks, "synthesis" deliverables — is exactly what Claude does in a 12-second response. Your survival path is owning a function the AI cannot embed into easily: relationship management, executive coaching, and politically delicate change management.

If you are a partner: your firm just got a 24-month window to either build its own forward-deployed practice or watch the JV eat the mid-market. McKinsey's QuantumBlack and Accenture's Center for Advanced AI are the obvious responses. Neither has the founding-model relationship Anthropic just locked up.

If you run a PE-backed portfolio company: your sponsor is about to call you. They will offer to fund a JV pilot that the new firm leads. The right answer is yes — but negotiate hard on data ownership. Every workflow the embedded team observes becomes training data for the next agent. Make sure you are not paying for the privilege of training your replacement's replacement.

If you build with AI: this announcement just validated the forward-deployed model as the dominant enterprise distribution strategy. Expect a wave of smaller specialist firms copying it — Anthropic alums starting verticals, ex-Palantir engineers spinning out industry-specific shops. The next year of enterprise AI looks less like SaaS and more like a return to the consulting roll-up era of the 1990s, with software margins layered on top.

The number that matters most

$1.5B is not a meaningful check by Wall Street standards. Blackstone alone deploys multiples of that on a normal quarter. The number that matters is the portfolio access: by our count, the four PE backers collectively own or have meaningful stakes in over 600 mid-market companies generating north of $1 trillion in combined revenue. That is the largest pre-installed customer base any AI startup has ever had on day one.

The new firm does not need to sell. It needs to deploy. The customers are already inside the building.

Related Resources

• See how enterprises are governing the resulting agent fleet: Microsoft Agent 365 launched the same week.
• Engineering teams are responding with their own AI integrations: Jama Connect MCP Server brings spec-driven development to Claude.
• The forward-deployed model also explains the rise of multi-agent harnesses like jcode trending on GitHub.

Frequently Asked Questions

What is the Anthropic Blackstone Goldman Sachs joint venture?

It is a $1.5B enterprise AI services firm announced May 4, 2026, founded by Anthropic, Blackstone, Hellman & Friedman, Goldman Sachs and General Atlantic. The firm sells forward-deployed engineers who embed inside customer companies and ship Claude-powered systems instead of slide-deck recommendations.

How does the Anthropic JV compare to McKinsey or BCG?

McKinsey and BCG sell senior consultants who write strategy decks and recommendations. The Anthropic JV sells engineers who embed on-site and build the working software the consultants would otherwise tell you to buy. The pricing is closer to a software contract than an hourly billable model, and the deliverable is a running production system, not a PDF.

How much money did Anthropic put into the new firm?

Anthropic committed $300M, matched by Blackstone and Hellman & Friedman at $300M each. Goldman Sachs and General Atlantic each put in $150M, with additional capital from Apollo, Leonard Green, GIC and Sequoia. Total announced commitment: $1.5B.

Is consulting actually going to disappear?

No — consulting gets repriced, not erased. The big losers are industries the JV walks into through PE portfolios: healthcare back office, financial services operations and infrastructure engineering. Strategy consulting at the partner level still exists, but mid-tier billable work compresses 30-50% over the next 24 months.

Why did OpenAI announce its own enterprise services JV the same day?

OpenAI saw the same opportunity Anthropic did and could not afford to cede the enterprise services category. Sam Altman's announcement matches the structural move — embedded engineers, enterprise verticals, PE distribution — but with a different cap table. The result is a duopoly race to capture the $200B consulting market rather than a single winner running unopposed.