For the past several years, the defining question in enterprise AI was: whose model wins? That question hasn't gone away, and it won't. With capable models now available across both closed and open weight, choosing well, and increasingly routing across them for the right task at the right cost, still matters enormously.
But it is no longer the question that separates the leaders. A new one has taken its place: who is doing the best job deploying this technology throughout their company, internally in how they operate and externally in how they serve customers?
That comes down to execution. The winners are the ones taking frontier capability and embedding it into the workflows and decisions that actually run the business.
OpenAI is launching the OpenAI Deployment Company (DeployCo) because they see the same shift we do: the bottleneck has moved from capability to deployment.
The deployment gap
Most of the capital in AI over the last decade went into building the technology. But look at how enterprises actually use it and a stubborn pattern emerges. McKinsey found that 88% of organizations now use AI in at least one function, while fewer than 40% have scaled it beyond a pilot. Access is nearly universal. Real deployment is rare.
That gap is an organizational problem, not a technical one. Running a pilot is easy. Redesigning a core workflow around AI, connecting it to live data and business processes, and getting frontline teams to actually rely on it is a different kind of work. It takes people who understand both the technology and the organization, and who can stay inside the problem long enough to solve it.
A company built to stay
DeployCo was built for exactly that. Its Forward Deployed Engineers embed inside organizations, working alongside business leaders and operating teams to find where AI creates real leverage, rebuild workflows around it, and ship production systems. This is neither software delivery nor traditional consulting. Software vendors hand off the implementation. Consultants hand off the recommendation. DeployCo's FDEs stay until the system is running in production and the organization has built the muscle to operate it itself.
To put experienced builders in the field on day one, OpenAI also agreed to acquire Tomoro, an applied AI engineering firm of roughly 150 engineers who have already done this work in demanding environments, including deployments at Tesco, Virgin Atlantic, and Supercell.
Why we're partnering with OpenAI on DeployCo
We have been building conviction around a category we call AI-native services, or AINS. The idea is straightforward: if AI can transform how enterprises operate, it should also transform how services firms operate. A new generation of companies can deliver expertise that once required large, expensive teams, with AI doing much of the work and humans supplying the judgment and the relationship.
We have written about this, mapped the early companies building in the space, and kept refining the thesis as the market has matured. DeployCo is the largest and most ambitious test of the model we have seen. Critically, it is designed to use AI to deploy AI, applying the same frontier technology it helps clients implement to run its own operations. If that works at scale, it resets expectations for what a services business can be.
That is why Emergence invested as a founding partner. It is also why we wanted to be close to it: to learn alongside the team as they push the model further than anyone has taken it. We will update our AINS playbook as the patterns emerge, and we expect this partnership to sharpen our thinking considerably.
The consortium behind DeployCo is worth noting. It launched with 19 founding partners, including TPG, Bain Capital, Brookfield, Goldman Sachs, SoftBank, McKinsey, and Bain & Company, and more than $4 billion in initial investment. The private equity sponsors alone back thousands of portfolio companies that now have a structured path to serious AI deployment. The consulting partners bring the change-management depth that technical deployment alone cannot provide. It is a deliberately constructed coalition, and it reflects how seriously OpenAI is taking the operational side of this transition.
The advantage compounds
Models will keep improving, and that will keep expanding what is possible. But the advantage is starting to compound somewhere else. Early movers get the organizational muscle, the refined workflows, and the institutional knowledge of what actually works. Companies still stuck in pilot mode two years from now will be playing catch-up against organizations that have been running production AI long enough to get good at it.
DeployCo is a bet that the deployment problem can be solved at scale, by a new kind of company built specifically for this moment. We are thrilled to be on the journey.
