OpenAI promises 17.5% returns to win enterprise AI war

OpenAI is getting aggressive on Wall Street to beat Anthropic at the enterprise game.

According to Reuters, the ChatGPT maker is offering private equity firms preferred equity stakes with a guaranteed minimum return of 17.5% to form dedicated AI joint ventures. That is well above standard preferred equity terms, and it comes bundled with early access to OpenAI’s newest frontier models, priority seniority, and downside protection.

Anthropic has been pitching buyout firms on similar joint ventures, but sources say its pitch included zero guaranteed returns.

Buying access to hundreds of corporate boardrooms

OpenAI is in advanced talks with heavyweights including TPG, Bain Capital, Advent International, and Brookfield. The goal is to raise around $4 billion for the joint venture at a $10 billion pre-money valuation.

The mechanics are straightforward. Private equity firms own portfolios of hundreds of established, cash-generating companies. By setting up a dedicated joint venture, OpenAI gains an instant channel to deploy its tools across thousands of corporate desks at scale.

As Matt Kropp from Boston Consulting Group pointed out, enterprise AI is a race for lock-in. Once an enterprise integrates a customized AI model deeply into its workflows, rip-and-replace costs become brutally high.

Anthropic has been courting buyout shops too, including Blackstone, Hellman & Friedman, and Permira. But Anthropic is relying on its enterprise reputation rather than financial sweeteners.

Carving out professional services ahead of an IPO

Customizing foundation models for Fortune 500 companies requires armies of forward-deployed engineers. That is expensive, slow, and terrible for software-like gross margins.

Both OpenAI and Anthropic are positioning themselves for potential IPOs as early as this year. Housing deployment engineers inside an off-balance-sheet joint venture lets them absorb heavy upfront integration costs elsewhere, giving the parent companies cleaner, higher-margin financial reporting for prospective public market investors.

The joint venture plans to monetize by charging integration fees, collecting revenue shares on deployed software, and co-owning intellectual property.

Still, not everyone is biting. Thoma Bravo, one of the biggest software buyout firms in the world, walked away after internal talks led by managing partner Orlando Bravo. The firm questioned the joint ventures' long-term profit profile, noting that many of its portfolio companies already use AI without signing away equity.

Why it matters

We think this move shows just how fiercely competitive the enterprise AI moat has become.

Anthropic has historically held the upper hand in enterprise trust and developer mindshare. By slapping a guaranteed 17.5% hurdle rate on preferred shares, OpenAI is basically using aggressive financial engineering to buy distribution channels that it could not easily win on product alone.

From a corporate structuring standpoint, it looks very sharp. Shunting low-margin consulting work into an affiliated entity protects OpenAI’s core valuation multiple heading into a public listing.

Yet offering guaranteed double-digit returns in private markets always carries real risks. If enterprise deployment cycles stall or portfolio companies resist adopting these customized tools at full price, OpenAI may find itself saddled with a costly financial obligation just to defend market share.