In May 2026, two things happened in the same week. OpenAI launched a $4 billion deployment company backed by a 19-firm syndicate led by TPG. Anthropic launched Ode, a $1.5 billion joint venture with Blackstone, Goldman Sachs, and Hellman & Friedman. Both ventures do the same thing. They embed engineers inside companies to make AI actually work.
That is $5.5 billion saying the same thing out loud. The model is not the hard part anymore. Deploying it is.
The Gap Everyone Talks About
The numbers are almost absurd. According to McKinsey’s State of AI survey, 88% of organizations use AI in at least one business function. But only about 7% have scaled it enterprise-wide. ModelOp’s 2026 benchmark found that 94% of enterprises have fewer than 25 AI systems in production; despite 67% having between 101 and 250 proposed use cases.
The pipeline is full. The output is nearly empty. The interesting question is why.
It is not the technology. Deloitte found that 70% of AI problems stem from people and process issues. Only 10% come from algorithms. Enterprise IT leaders say AI agent adoption stalls on organizational design, fragmented data, and implementation talent; not model capability. The bottleneck was never the brain. It was the body.
If your team is navigating this gap right now, BTO’s free Enterprise AI Operating Model Blueprint lays out the structural foundations for moving from pilot to production. It is a practical starting point for teams that know the model works but have not figured out the operating system around it yet.
Why Both Labs Made the Same Bet
Here is the background that makes this move make sense. The gap between frontier models has narrowed. Enterprise DNA noted that Gemini 3.5 Pro, Claude Fable 5, and GPT-5-tier models are all competitive in ways that make the “which model” question less important than it was a year ago. When the tool selection decision gets easier, the deployment decision gets all the attention.
OpenAI’s response was massive. The Deployment Company launched with $4 billion and immediately acquired Tomoro, an AI consulting firm with roughly 150 forward-deployed engineers who had already been building production systems for Mattel, Tesco, Virgin Atlantic, and Supercell. OpenAI retained majority ownership. Industry estimates suggest the unit could grow to 2,000 to 4,000 deployment engineers within three years.
Anthropic matched the thesis with a different structure. Ode launched at $1.5 billion, acquiring Fractional AI and its co-founders Chris Taylor and Eddie Siegel. Ode operates on a “Claude-first” principle but is not contractually locked to Anthropic’s models. The investor coalition; Blackstone, Goldman Sachs, General Atlantic, Apollo, Sequoia; has zero overlap with OpenAI’s syndicate. The financial world literally split into two camps, each backing the same thesis with different capital.
What makes this significant is that both companies could have kept selling API access and let system integrators handle deployment. They chose not to. They chose to own the implementation layer. That is a signal every enterprise leader should pay attention to.
What This Means for the Consulting Industry
Fortune reported that the Anthropic venture puts it in direct competition with McKinsey and the traditional consulting industry. Blackstone President Jon Gray framed the whole initiative as solving the “scarcity of engineers who can implement frontier AI systems at speed.”
This is worth sitting with. The companies that build the AI now believe they can deploy it better than the companies that have spent decades deploying technology for enterprises. Whether or not that turns out to be true, the competitive signal matters. It tells us that implementation is no longer a supporting role. It is the main event.
If you are thinking about how to structure this for your own organization, the free AI Readiness Assessment Matrix helps teams evaluate where they stand across six dimensions; from data readiness to governance to team capability. It is the kind of honest inventory that clarifies whether you need to build internal capability, bring in external help, or both.
The New Competitive Question
For years, the competitive question in enterprise AI was “which model are you using?” That question is fading. The new question is “who is deploying it, and how fast?”
Consider what the data says. 76% of enterprise AI use cases are now purchased rather than built internally, a sharp reversal from the year before. Enterprise AI spending passed $300 billion in 2026. The money is flowing. But the value is stuck in the pipeline.
The organizations pulling ahead are not the ones with the best model access. They are the ones that figured out how to rewire their teams, data, and workflows around AI. That is an operating model problem. It is an organizational design problem. It is a change management problem. And it is exactly the problem that $5.5 billion just got committed to solving.
For teams ready to go deeper on the structural work, explore the AI Operating Model Implementation Kit. It is built for teams past the “should we?” phase and into “how do we roll this out across the organization?” Think of it as the blueprint plus the construction manual.
What Teams Should Do Now
Three things become clearer in light of this shift.
First, evaluate your implementation capability with the same rigor you evaluate your models. Who on your team knows how to wire AI into real workflows? Who manages the data pipelines, the governance, the change management? If the answer is “nobody specifically,” that is your actual bottleneck.
Second, watch the “build vs. partner” decision carefully. These new ventures are designed to make “partner with the lab” an option it was not before. That does not mean it is the right option for everyone. But it is a new variable in the equation.
Third, stop treating deployment as a downstream activity. The highest-value AI applications are not the ones that automate an existing process. They are the ones that redesign how work happens. That requires implementation thinking from day one, not after the model is chosen.
For a comprehensive approach to this; from readiness assessment through full operational rollout; dig into the AI Readiness Assessment Complete Toolkit. It connects the strategic layer to the execution layer so that nothing gets lost in translation.
The Shift That Matters
Here is the simplest way to read this moment. For three years, AI labs competed on intelligence. Better benchmarks, bigger context windows, faster inference. That race is not over. But it is no longer the only race.
The new race is about making AI useful at scale, inside real organizations, with real data, real teams, and real constraints. That race just got $5.5 billion in funding from the two companies that know the most about what their own models can and cannot do in production.
The model was always the beginning. The operating system around it was always the point.