$1.5B Anthropic + Blackstone Goldman + H&F venture $10B OpenAI Deployment Co. TPG · Brookfield · Bain 6:1 services to software consulting industry ratio THE FORK A·HIRE B·BUILD
Field Notes / AI Strategy

Goldman said the quiet part out loud.

Anthropic's $1.5 billion Blackstone-Goldman venture and OpenAI's $10 billion Deployment Company landed within hours of each other. Both said the same thing on the record: the model isn't the constraint. Methodology is.

By Mike Goetz May 2026 8 min read
Read
On the record
Having the model alone doesn't change your workflows or how you operate. You need people who can combine the technology with what's actually happening in the business and implement those changes.
Marc Nachmann, Global Head of Asset and Wealth Management, Goldman Sachs
That's a $1.5 billion sentence.
$1.5B
Anthropic JV with Blackstone, Goldman, Hellman & Friedman, plus Apollo, GA, GIC, Leonard Green
$10B
OpenAI Deployment Co. valuation, $4B raise from TPG, Brookfield, Advent, Bain
6:1
Services-to-software dollar ratio, the consulting industry both labs are targeting

On Monday, Anthropic announced a joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs, backed by additional capital from Apollo, General Atlantic, GIC, Leonard Green, and Sequoia. The new firm, still unnamed at filing, will embed Anthropic engineers directly inside the operations of mid-size companies. The first customers are portfolio companies of the participating private equity sponsors. The structural model is Palantir's forward-deployed engineering playbook, pointed at the consulting industry.

Hours before the Anthropic announcement, OpenAI announced a parallel venture called The Deployment Company, valued at $10 billion and raising approximately $4 billion from a PE consortium that includes TPG, Brookfield, Advent, and Bain Capital. Same structure. Same target. Different anchor capital.

The two largest AI labs in the world filed press releases that read like they had been written by the same lawyer. Both said the new firms would put engineers inside Fortune 500 companies and PE portfolios to redesign workflows around the models. Both targeted the consulting industry, where companies spend roughly six dollars on services for every dollar they spend on software. That ratio has made consulting a multi-trillion-dollar industry, and the AI labs have decided they want it.

That's the news. Here's what it actually means.

01

The bottleneck was never the model

For the past two years, the public conversation about AI has been mostly about which model is best. Which provider, which benchmark, which agent platform, which prompting technique. The implicit assumption has been that capability is the constraint. Get the right model and the rest follows.

The Goldman partner who just put $150 million into a forward-deployed engineering firm is telling you, on the record, that the model isn't the constraint. The constraint is the systematic human work of mapping AI capability onto how a business actually operates. That's not a new claim. People who've been working with these tools at scale have been saying it for at least eighteen months. What's new is who's saying it now, and how much capital they're putting behind it.

When the Global Head of Asset and Wealth Management at Goldman Sachs says the model alone doesn't change your workflows, that's no longer a contrarian position. That's an institutional position. With $1.5 billion of fresh capital behind it on the Anthropic side, and OpenAI raising approximately $4 billion hours earlier behind its own $10 billion-valued deployment venture.

Jon Gray, President and COO of Blackstone, framed it the same way in the official announcement. The new firm exists to break down what he called "one of the most significant bottlenecks to enterprise AI adoption by expanding the number of highly skilled implementation partners." That bottleneck is the scarcity of engineers who can take a frontier model and turn it into something a real business actually runs on. Not a demo. Not a pilot. The actual day-to-day operating system.

Methodology is the moat. Goldman just said it on stage with $1.5 billion behind it.

02

The fork this announcement creates

Once you accept that methodology is the bottleneck, every company facing AI adoption has the same structural choice. There are two valid responses. They serve different markets, different psychographics, different time horizons. Both work.

The Two Paths

Both options are valid. The wrong choice for your kind of company is expensive in either direction.

A
Hire methodology embedded
The new Anthropic and OpenAI ventures

Forward-deployed engineers sit with your team. They map your workflows. They build the implementations. You pay them ongoing. The implementation depreciates the day you stop paying. Better engineers and better tools than McKinsey or Deloitte have today.

Fit: no internal AI literacy. No learning culture. Immediate revenue at risk. Outcomes over capability transfer. Expensive, fast, works.
B
Build methodology internal
Frameworks the company owns

Capability lives inside the people who already work there. Frameworks compound as institutional IP. The methodology keeps working whether or not anyone keeps paying a vendor. No permanent engineering dependency.

Fit: technical staff already in place. Learning culture. Longer time horizon. Intellectual asset orientation. Cheaper over time, owned outright.

The honest reading of which fork you're on takes about twenty minutes. Most companies haven't done it because the structural choice wasn't visible until last week.

03

What changes from here

Three things are true now that weren't true a week ago.

First, methodology is officially the bottleneck. Not the model. The institutional capital is now behind that claim, which means the conversation in boardrooms is going to shift toward implementation methodology and away from model selection. That shift was overdue, and it's going to happen fast now that Goldman and Blackstone are pulling it forward.

Second, the consulting industry is genuinely under attack. Both AI labs have decided that the services-to-software dollar ratio is a target, not a fact of life. The Big Three consulting firms are going to respond, but they're responding from a position where their billable engineers don't have the same relationship to model development that Anthropic's and OpenAI's do. That's a real disadvantage that capital alone won't close.

Third, the structural fork between Option A and Option B is now a real choice instead of a theoretical one. Until last week, Option A wasn't really available outside of the largest enterprises. Now it is, at scale, with billions of dollars of combined capital backing two competing versions of it. That changes the math for every company evaluating its AI strategy.

The model isn't the constraint, and methodology is.

04

A note on receipts

The work I mentioned earlier covers more than the website. There's an open-source framework-builder repository on GitHub for anyone who wants to build their own methodology infrastructure. There's a Strategic Thinking Academy that teaches the methodology to people who want to internalize it instead of renting it. There are 700-plus frameworks built across that fourteen-month stretch, applied across industries from federal contracting to medical advocacy to creative production. The thesis behind all of it was the same one Nachmann just said on the record: the model isn't the constraint, and methodology is.

That work was a contrarian bet a year ago. It isn't anymore. Goldman just said the quiet part out loud, and the quiet part is exactly what I've been building toward.

If you're on Option A

The new Anthropic venture is probably the right call for you.

Well-funded. The engineers will be excellent. Model access is direct. Take the meeting.

Anthropic.com →
If you're on Option B

This is what the methodology side looks like.

Frameworks the company owns. Open-source tools. Strategic Thinking Academy for internal capability.

HowToFramework.com →

Either way, the structural choice is clearer this week than it was last week. That's worth something on its own.

MG
Mike Goetz

Mike Goetz is the founder of RageDesigner, where he has built systematic thinking methodology since 2003. His framework library now exceeds 700 documented frameworks across federal contracting, AI strategy, content production, sales, medical advocacy, and creative production. He teaches framework generation at whatisaframework.com and howtoframework.com. The open-source framework-builder repository is at github.com/framework-creator/framework-builder.