OpenAI and Anthropic target consulting market with $5.5B in new joint ventures

OpenAI and Anthropic have signed joint ventures with four of the biggest names in private equity, committing a combined $5.5B to a strategy that sounds less like an AI lab and more like McKinsey with GPUs. The deals, struck with TPG, Blackstone, Goldman Sachs, and Bain Capital, will embed AI engineers directly into thousands of portfolio companies across the mid-market economy.

The Palantir playbook, scaled up

The model here isn’t new. Palantir pioneered the concept of “forward-deployed engineers,” essentially sending its own people into client organizations to build custom software on top of Palantir’s platform. OpenAI and Anthropic are now attempting to run that same playbook, but at a scale Palantir never reached. Instead of winning individual enterprise contracts one at a time, they’re partnering with private equity firms that collectively control thousands of portfolio companies.

Goldman Sachs’ involvement is particularly interesting for Anthropic. The partnership reportedly gives Anthropic a pipeline into fintech verticals, specifically wealth management, lending, and insurance.

Two deals, very different risk profiles

One of the deals, likely OpenAI’s, includes a 17.5% guaranteed annual return for investors. That’s a bold commitment from a company reportedly projecting $14B in losses. Anthropic’s deal, by contrast, offers no such guarantee.

Why consulting firms should be nervous

OpenAI and Anthropic are cutting out the middleman. By embedding their own engineers directly, they capture both the technology revenue and the services revenue. That’s a direct threat to the integration practices at firms like Accenture, Deloitte, and yes, even Bain’s consulting arm, which is notably also a partner in these ventures.

OpenAI’s enterprise LLM API market share dropped from 50% to roughly 25% between late 2023 and mid-2025, according to Menlo Ventures data. Services, on the other hand, create stickier relationships. Once your engineers are embedded in a company’s workflows, switching costs go through the roof.