OpenAI, Anthropic Race to Build Their Own Palantir

On Monday, OpenAI and Anthropic each announced a new company. Neither press release used the word "consulting." Both described ventures that are, structurally, exactly that.

TL;DR

OpenAI closed "The Deployment Company," a $10B vehicle backed by 19 PE investors including TPG and Brookfield, promising backers a 17.5% annual return
Anthropic formed a $1.5B services firm with Blackstone, Hellman & Friedman, and Goldman Sachs to embed Claude in mid-market companies
Both ventures use the forward-deployed engineer model Palantir built its business on: lab engineers working inside client organizations
The target sectors - healthcare, manufacturing, financial services - have historically been McKinsey and Deloitte territory

The two announcements came within hours of each other on May 4. OpenAI had been quietly assembling its investor syndicate since March. Anthropic moved faster with its investor list: alongside Blackstone, Hellman & Friedman, and Goldman Sachs, it drew in General Atlantic, Leonard Green, Apollo Global Management, Singapore's GIC, and Sequoia Capital.

The Comparison That Matters

	Anthropic Venture	The Deployment Company (OpenAI)
Valuation	$1.5B	$10B
PE capital raised	~$900M	$4B
Lab contribution	$300M	$500M (+ $1B option)
Backing firms	Blackstone, H&F, Goldman + 6 others	TPG, Brookfield, Advent, Bain + 15 others
Total investors	9	19
Guaranteed return	Not disclosed	17.5%/year over 5 years
Control structure	Not disclosed	OpenAI retains super-voting shares
Target clients	Mid-market + PE portfolios	PE portfolios
Priority sectors	Healthcare, manufacturing, retail, real estate	Healthcare, logistics, manufacturing

The scale difference is significant. OpenAI is running a $10 billion vehicle with a guaranteed income structure built in. Anthropic is smaller, with a more dispersed investor base. The operational logic is identical.

The Same Playbook

Both ventures are built on what Palantir spent 20 years proving out: send your engineers into client organizations and don't leave until the system works. No software license handed off to an IT department. No implementation left to a third-party integrator. Engineers embedded inside the client, working on the client's actual data and actual workflows.

Two business professionals celebrate a deal at an office desk Both ventures move away from the standard SaaS model toward long-term embedding: lab engineers inside client operations. Source: unsplash.com

Anthropic's announcement described the engagement cycle: small teams first identify where Claude can have impact, then Anthropic Applied AI staff collaborate with the client to build tools, then those tools get embedded into existing workflows. The healthcare example it cited - clinicians getting AI tools for documentation and medical coding without having to change how they work - maps directly to the kind of deployment Palantir made its name on.

OpenAI's Deployment Company is framed identically. Teams of OpenAI engineers working directly inside client operations, starting with healthcare, logistics, manufacturing, and financial services.

"Enterprise demand for Claude is significantly outpacing any single delivery model," said Krishna Rao, Anthropic's CFO, in the company's announcement.

That quote does a lot of work. Claude is already deployed through Accenture, Deloitte, and PwC - all partners in Anthropic's existing delivery network. The new venture doesn't replace those relationships. It adds a direct channel the lab controls end-to-end, with higher margins and tighter feedback loops.

The PE Distribution Channel

Private equity firms aren't in this deal for the mission. They're in it for two reasons: financial returns, and preferred access to AI for their portfolio companies.

A business professional reads a newspaper showing financial data and market prices PE firms backing both ventures gain guaranteed returns and preferred AI deployment access across their portfolio companies. Source: unsplash.com

OpenAI made the economics explicit. PE backers get a guaranteed 17.5% annual return over five years. OpenAI retains super-voting shares, keeping strategic control while the financial sponsors take the income side. It reads like an infrastructure investment, not a startup bet.

Blackstone controls over 230 portfolio companies. TPG has comparable reach in healthcare and financial services. By partnering with these firms, Anthropic and OpenAI each bought access to a pre-assembled customer list. Selling enterprise AI into cold markets is expensive and slow. Selling into a PE portfolio where the GP can influence adoption is a different problem entirely.

Jon Gray, Blackstone's President and COO, was direct about the ambition: "We intend to build a scaled, world-class company to deploy Anthropic's technology across businesses."

That sentence describes a channel play.

Counter-Argument

The consulting industry has things these ventures don't. McKinsey has 45,000 consultants and decades of relationships across every major sector. Accenture has 750,000 employees and the global infrastructure to run enterprise transformation at scale. Deloitte and PwC have embedded compliance and audit relationships that took a generation to build.

The forward-deployed engineer model works when you're deploying surveillance software for the US government, as Palantir did. It works less cleanly when you're trying to redesign procurement workflows at a regional healthcare system where the real obstacle is change management, not technology.

There's a conflict of interest in the structure too. If Blackstone's portfolio companies are simultaneously investors and customers, that relationship isn't arm's length. A community bank without PE backing doesn't get the same access, regardless of Anthropic's stated mid-market ambitions.

The guaranteed return structure is the sharpest risk. OpenAI committed to paying its PE backers 17.5% annually on $4 billion. That's $700 million in guaranteed distributions per year before The Deployment Company earns a profit. OpenAI hasn't turned a profit at the corporate level. The math requires this venture to be substantially cash-generative, fast.

What the Market Is Missing

Green matrix-style data stream representing AI-native enterprise operations AI-native deployment arms aim to reshape how large organisations operate from the inside - a market the major consulting firms have owned for decades. Source: unsplash.com

The real target isn't other AI companies. It's Accenture, McKinsey, Deloitte, and Capgemini. Professional services firms collectively create well over $300 billion annually helping large organizations change how they operate. Both labs are aiming at the same prize.

The irony is that Anthropic already lists Accenture, Deloitte, and PwC as delivery partners. It's building a competitor to firms it's counting on as distribution. OpenAI's recent move to AWS Bedrock follows the same logic: establish the platform, then build the services arm to deploy it directly.

These ventures are partly a hedge. If established consulting firms slow-walk AI adoption to protect their billable hours, the labs now have their own channel. If the big firms accelerate AI faster than expected, the labs have a front-row seat through their partner networks.

Either way, the professional services industry is now competing with the organizations building the tools it's supposed to be selling.

Sources: