OpenAI built a ten billion dollar consulting company. Anthropic built a 1.5 billion dollar consulting company. Google is writing a licensing agreement. The difference in approach may determine which AI lab captures the largest new enterprise distribution channel to emerge since the birth of cloud computing itself: the portfolio companies of the world’s biggest private equity firms.

Alphabet is in talks with Blackstone, KKR, and European private equity firm EQT to give their portfolio companies access to Google’s Gemini AI models under omnibus licensing agreements, according to a Bloomberg report published on Monday. The discussions are not exclusive and no deals have been finalised. But the structure Google is proposing is fundamentally different from what its two principal competitors have built, and the difference reveals a strategic bet about how enterprise AI will actually be deployed at scale.

The race

The private equity AI land grab accelerated over the past week with a velocity that suggests the three leading AI labs view buyout firms not as customers but as distribution infrastructure. On Sunday, OpenAI finalised The Deployment Company, a ten billion dollar joint venture anchored by TPG with 19 investors including Brookfield, Advent, and Bain Capital. The structure guarantees investors a 17.5 per cent annual return over five years, with OpenAI committing up to 1.5 billion dollars of its own capital and retaining strategic control through super-voting shares. The business model is Palantir’s forward-deployed-engineer approach applied to AI: teams of OpenAI engineers embedded directly inside client organisations, redesigning workflows across healthcare, logistics, manufacturing, and financial services.

The same day, Anthropic announced its own 1.5 billion dollar enterprise services firm with Blackstone, Hellman and Friedman, and Goldman Sachs, each anchoring at roughly 300 million dollars, with General Atlantic, Leonard Green, Apollo, GIC, and Sequoia also participating. The venture will embed engineers inside portfolio companies to integrate Claude into core business operations, operating as something between a consulting arm and a deployment factory.

Google’s approach is different. Rather than building a joint venture, hiring hundreds of engineers, and sending them into portfolio companies with implementation playbooks, Alphabet is negotiating licensing agreements that give an entire private equity firm’s portfolio access to Gemini models and Google Cloud AI infrastructure under a single commercial arrangement. The distinction is not merely structural. It reflects a fundamentally different theory about what enterprise AI customers actually need.

The logic

OpenAI and Anthropic are betting that the bottleneck in enterprise AI adoption is implementation. Their joint ventures assume that companies need not just access to frontier models but teams of specialist engineers who can redesign processes, build custom integrations, and manage the transition from pilot to production. The model is labour-intensive, high-margin, and slow to scale. It is also, if the implementation succeeds, extraordinarily sticky: once an AI lab’s engineers have rebuilt a company’s core workflows around a specific model family, the switching costs become prohibitive.

Google is betting that the bottleneck is procurement. Google has already committed 750 million dollars to a partner fund financing agentic AI deployments through Accenture, Deloitte, KPMG, PwC, and NTT DATA, the consulting firms that already serve the portfolio companies of Blackstone and KKR. The omnibus licensing model layers on top of this existing channel: rather than building its own consulting operation, Google is offering private equity firms a commercial wrapper that gives their entire portfolio access to Gemini, then relying on the consulting ecosystem it has already financed to handle implementation. The approach trades consulting revenue for distribution speed, prioritising breadth over depth.

The scale of the opportunity explains the urgency. Blackstone and KKR manage combined assets exceeding two trillion dollars across thousands of portfolio companies spanning healthcare, logistics, technology, real estate, and financial services. EQT manages approximately 130 billion euros. If Google secures omnibus deals with all three firms, it would open the largest single new customer channel in Alphabet’s history since the launch of Google Cloud.

The context

Alphabet’s market capitalisation surged past 4.6 trillion dollars after Q1 2026 earnings that beat estimates across every division. Google Cloud crossed 20 billion dollars in quarterly revenue for the first time, growing 63 per cent, with the cloud backlog nearly doubling to more than 460 billion dollars. Revenue from products built on generative AI models grew nearly 800 per cent year on year. The company is not negotiating from weakness. It is negotiating from the position of a platform that already has 750 million Gemini users and an ecosystem of consulting partners that captured up to 7.05 dollars for every dollar spent on Google Cloud.

The competitive dynamics are complicated by the fact that Blackstone appears on both sides of the table. It is simultaneously a founding investor in Anthropic’s 1.5 billion dollar joint venture and a potential customer of Google’s omnibus licensing programme. Blackstone also recently created Blackstone N1, a new West Coast division dedicated exclusively to its AI and high-growth technology investments, which include stakes in both OpenAI and Anthropic. The firm is not choosing a single AI provider. It is positioning itself as a distribution channel for all of them, extracting value from the competition between labs rather than committing to one platform.

The question

Anthropic’s earlier discussions about pushing Claude into enterprise through private equity and the rapid formation of both The Deployment Company and the Anthropic-Blackstone venture suggest that the AI labs believe the next phase of the model wars will be fought inside portfolio companies, not in benchmark leaderboards. The private equity channel is attractive because it bundles thousands of mid-market companies under a single commercial relationship, compresses the enterprise sales cycle from months to weeks, and creates a feedback loop in which implementation data flows back to the AI lab, improving the models that the portfolio companies depend on.

But the three approaches carry different risks. OpenAI’s Deployment Company has raised the most capital and promised the highest returns, which means it must generate substantial recurring revenue quickly or face investor pressure. Anthropic’s venture is smaller but ties it to a specific set of financial sponsors whose portfolio companies may not represent the broadest cross-section of enterprise use cases. Google’s licensing model is the cheapest to operate and the fastest to scale, but it cedes the implementation relationship to third-party consultants, which means Google may never develop the deep understanding of enterprise workflows that OpenAI and Anthropic are building through their embedded-engineer programmes.

The private equity AI race is, at its core, a distribution bet. OpenAI is building a consultancy. Anthropic is building a consultancy. Google is building a licensing business. The first two approaches assume that enterprise AI is a services problem. The third assumes it is a platform problem. The history of enterprise technology suggests that platforms eventually win, but only if the product is good enough to survive without hand-holding. Whether Gemini meets that threshold across thousands of portfolio companies, in industries ranging from healthcare to logistics to financial services, is the question that Google’s omnibus licensing model is designed to answer. The deals have not been signed. The race to sign them has already begun.