The Most Powerful AI Labs in the World Just Became Consulting Companies
Within 72 hours of each other in early May, OpenAI and Anthropic made the same bet: that the path to an IPO runs through an army of consultants. OpenAI launched its Deployment Company on May 11 with $4 billion and 19 investors. Anthropic followed on May 4 with a $1.5 billion venture backed by Blackstone, Goldman Sachs, and Hellman & Friedman. Neither led with a new model. The product being sold, in both cases, was the installation layer.
The inversion was immediate. Within days, The New Stack reported that PwC and OpenAI are co-building CFO-office agents — software that automates the finance function directly, not as a consulting recommendation but as a live system. Anthropic released 10 finance-specific agent templates with live data connectors on the same timeline. The labs are not just building models. They are buying the on-ramp.
The mechanism is straightforward, even if the incentives are tangled. When a consulting firm like Bain or McKinsey wins an enterprise AI implementation deal, it can now embed DeployCo or Anthropic's services into its own engagement — becoming the prime contractor for the installation while routing the underlying AI work through its own labor force. The consulting partner gets a services revenue stream. The lab gets distribution into accounts it could not have reached directly. The combined entity can pitch itself to a CFO as a single implementation outfit, with the AI layer hidden inside the labor cost.
OpenAI accelerated the move by acquiring Tomoro, an applied AI consulting firm, bringing approximately 150 Forward Deployed Engineers from day one of the new venture. DeployCo's founding investor roster includes TPG, Advent, Bain Capital, and Brookfield alongside consulting partners Bain, Capgemini, and McKinsey — firms whose combined client networks sponsor more than 2,000 businesses worldwide. Anthropic's venture is backed by General Atlantic, Leonard Green, Apollo Global Management, GIC, and Sequoia Capital. The money is not coming from AI companies alone. It is coming from the firms that know how to sell into the enterprise.
Traders on prediction market Kalshi give OpenAI an 83 percent chance of reaching the public markets before Anthropic, according to CNBC — a lead that has shifted in recent weeks as both companies intensified the same behind-the-scenes bet: who can build a services business fast enough to show public market investors the revenue is real. Both are preparing for IPOs as early as Q4 2026. What neither says publicly is that the outcome may depend less on which lab's model scores higher on benchmarks and more on who can put the most humans between their AI and a live enterprise workflow.
The conflict of interest is not subtle. Venture capitalist Chamath Palihapitiya warned that consulting firms partnering with these labs are letting the fox into the henhouse, as the labs simultaneously build competing services businesses. McKinsey and Bain will charge enterprises to implement AI. The same labs they partner with are also selling AI directly to those same enterprises. The firms most trusted inside large organizations are now also the distribution arm for products that could make those organizations' internal consulting practices obsolete.
Both ventures are pre-revenue. Neither has disclosed a named enterprise customer deployment. The history of enterprise software suggests the consulting model is a double-edged sword: consulting firms have survived every wave of disintermediation — ERP, cloud, SaaS — by becoming the implementation layer for the very products that were supposed to replace them. Whether the labs can convert that distribution into recurring software revenue rather than one-off services contracts is the unanswered question.
Anthropic and OpenAI declined to make executives available for this article.
What comes next is a test neither company has passed in public. Both are telling investors they can run a services business and a product business without the two colliding. Both are also telling investors they are on a path to an IPO that requires predictable revenue, not a consulting roll-out with long enterprise sales cycles. The model companies are becoming consulting companies right as the market is supposed to reward software companies.
The next signal to watch is whether the consulting partnerships produce named enterprise customers — real deployments inside a CFO office, an IT operation, a supply chain workflow — before the IPO filings start circulating.