OpenAI and Anthropic tap PE giants in enterprise AI race

After spending billions training giant models, the AI frontier is facing a cold reality check: selling API tokens through a browser is not going to justify a multi-hundred-billion-dollar valuation.

OpenAI has finalized a massive joint venture with top private equity players to aggressively deploy its AI into corporate America. The company raised more than $4 billion from heavyweights including TPG, Brookfield Asset Management, Advent, and Bain Capital. According to a Bloomberg report, the new entity is valued at $10 billion before cash, carries the literal name "The Deployment Company," and will be majority-owned and controlled by OpenAI.

Minutes later, Anthropic fired back with an almost identical playbook. The rival startup announced a joint venture backed by Blackstone, Hellman & Friedman, and Goldman Sachs, aimed at embedding its Claude models directly into midsize enterprise workflows.

The rise of the forward-deployed army

OpenAI's deal also brought in SoftBank, Dragoneer Investment Group, and several consulting outfits. Anthropic lined up its own blue-chip coalition, including Apollo Global Management, General Atlantic, Leonard Green & Partners, GIC, and Sequoia Capital.

Both ventures exist to solve the same headache: getting old-school companies to actually use and pay for AI. To make it work, both labs are aggressively poaching and hiring "forward-deployed engineers"—technical specialists who parachute directly into corporate client offices to rewire workflows around proprietary models.

The timing is hardly a coincidence. Both OpenAI and Anthropic are reportedly eyeing IPOs as soon as this year. To appease public market investors, they need to show massive, recurring enterprise revenue right now, not just promises of AGI.

Why it matters

This looks like the end of the easy software era for foundational AI. Winning enterprise accounts now requires old-school consulting muscle.

For businesses, particularly the vast portfolios owned by private equity firms, this is an express lane to digital transformation. Having model makers send dedicated engineering teams to customize workflows solves the adoption barrier overnight. But it also creates massive vendor lock-in. Once Claude or ChatGPT is baked into your core operational plumbing, switching models will be practically impossible.

Strategically, setting up off-balance-sheet joint ventures is a clever maneuver. It allows OpenAI and Anthropic to capture enterprise market share without shouldering the heavy headcount costs of a massive services division. Yet we think it also carries real risk. IT deployment is notoriously messy and low-margin compared to pure software. Turning model companies into Palantir-style consulting outfits may secure IPO headlines, but Wall Street will eventually look at those blended margins.