The Labs Are Coming Inside Your Business
On May 11, 2026, OpenAI announced the launch of the OpenAI Deployment Company, a $4 billion venture majority-controlled by OpenAI and backed by 19 global investment firms, consultancies, and system integrators, including TPG, Bain Capital, Goldman Sachs, SoftBank, McKinsey, and Capgemini. Alongside the launch, OpenAI announced the acquisition of Tomoro, an applied AI consulting firm that has built production AI systems for companies including Tesco, Virgin Atlantic, and Supercell. The acquisition brings approximately 150 forward deployed engineers into OpenAI's new enterprise unit from day one.
The premise is straightforward. Enterprise AI has moved past demos. Companies now want systems that plug into legacy software, reshape workflows, and hold up under daily operational pressure. OpenAI's answer is a new subsidiary built to embed engineers directly inside client organizations to identify where AI creates the most value, redesign workflows around it, and turn those gains into durable production systems.
One week earlier, Anthropic had made an almost identical move. Anthropic launched a $1.5 billion joint venture backed by Blackstone, Goldman Sachs, Hellman and Friedman, Apollo, and General Atlantic to accelerate AI deployment across private equity portfolio companies, embedding Anthropic engineers inside midsized businesses to implement AI systems including Claude Code.
Two of the three largest AI labs announced enterprise deployment arms within seven days of each other. That is not a coincidence. It is a structural signal about where enterprise AI value is being created and who the labs believe should be capturing it.
Why the Labs Are Making This Move
The decision by OpenAI and Anthropic to build deployment businesses is a direct response to a problem they have been watching accumulate for two years: the gap between what their models can do and what enterprises are actually getting out of them.
Deloitte's 2026 State of AI in the Enterprise report says worker access to AI rose by 50% in 2025, but only 34% of surveyed organizations are deeply transforming by creating new products, services, or business models. 37% are still using AI at a surface level with little or no change to existing processes. Enterprise now makes up more than 40% of OpenAI's revenue, and the company expects that figure to reach parity with consumer by the end of 2026. That trajectory depends on enterprises generating enough value from AI to justify continued and expanding investment. If enterprises keep running surface-level deployments without production-grade outcomes, the renewal and expansion revenue that OpenAI's growth model depends on does not materialize.
The model behind the OpenAI Deployment Company closely resembles the Palantir enterprise model. Forward deployed engineers are meant to embed inside client organizations, connect models to legacy systems, and redesign workflows around actual operational needs. Palantir refined that approach over years of defense and intelligence engagements, where software had to work inside complex institutions rather than sit on top of them. OpenAI is borrowing a playbook that has already been proven to turn advanced software into something businesses can reliably use at scale.
The strategic logic is also defensive. Anthropic and Google Gemini are putting more pressure on OpenAI in enterprise accounts, and the battle is increasingly about delivery, not just model quality. When model performance between frontier labs has converged to the point where differentiation is marginal, the organization that wins enterprise relationships is the one with the best implementation capability, the deepest workflow knowledge, and the most trusted presence inside the client's operations. That is what both OpenAI and Anthropic are now building.
What This Means for the Consulting and Implementation Market
The immediate implication of two $4 billion and $1.5 billion deployment ventures entering the market is that the AI consulting and implementation landscape is being disrupted from above. McKinsey, Capgemini, and Bain are backing partners in the OpenAI Deployment Company rather than competing with it directly, which is itself a signal about how those firms assess their own positioning.
For enterprises, this creates both options and risks that require clear-eyed evaluation.
The options side is straightforward. Organizations that have been struggling to find experienced AI implementation talent now have a higher-capitalized, more structured option. The OpenAI Deployment Company's investment and consulting partners sponsor more than 2,000 businesses around the world. The venture's network gives OpenAI access to enterprise relationships at a scale that would have taken years to build through direct sales alone. For organizations already deeply embedded in the OpenAI ecosystem, the deployment company offers a more direct path from pilot to production.
The risks require equal attention, and enterprise buyers should evaluate them carefully before committing.
The first is model lock-in structured as implementation dependency. The new company will focus exclusively on OpenAI's technology ecosystem. Forward deployed engineers embedded inside your organization will be building workflows, integrations, and operational systems around OpenAI's models. When those engineers leave, they take knowledge that is increasingly model-specific rather than transferable. When OpenAI's pricing changes, and it has been changing, the cost of switching away from the infrastructure your operations now depend on rises accordingly. As we reported last week, with the launch of GPT-5.5, OpenAI once again increased its API pricing, with output costs doubling that of its predecessor.
The second risk is the vendor-client dynamic that comes with having your strategic technology partner's employees embedded in your operations. The forward deployed engineer is there to help you succeed. They are also there to ensure that success is built on OpenAI's stack. Those two goals are aligned most of the time. They are in tension when evaluating whether a different model or a different architecture might serve your specific needs better.
The third risk is concentration. If the primary implementation expertise available to your organization is channeled through a vendor's own deployment arm, your ability to get objective counsel about your AI architecture diminishes over time.
The Deeper Signal: Delivery Is Now the Competitive Moat
The launch of both ventures this week confirms something that the most perceptive observers of enterprise AI have been saying for 18 months: the competitive moat in enterprise AI is no longer primarily in the models. It is in the implementation layer.
The AI industry is increasingly acknowledging that deployment services, integration expertise, and operational implementation may become just as strategically important as model development itself. That shift could reshape enterprise AI buying decisions, agency offerings, consulting markets, and technology partnerships.
OpenAI's enterprise revenue chief described the deployment company's value proposition directly: forward deployed engineers can sit with an organization, sit with their users, understand the workflow, and then help them take that capability from their back-office applications, connecting it to the model, and really building intelligence into each workflow.
That description is worth reading carefully, because it describes work that requires deep organizational knowledge, not just technical skill. Understanding a workflow well enough to redesign it around AI requires understanding the business, the regulatory environment, the data landscape, and the people who will use the system. It requires the kind of contextual judgment that comes from time spent inside an organization, not from model benchmarks or capability announcements.
This is why the Palantir comparison is apt and instructive. Palantir has maintained pricing power and deep enterprise relationships not because its models are the most advanced, but because its forward deployed engineers accumulate irreplaceable institutional knowledge inside client organizations over time. That knowledge becomes the product, and the model becomes the infrastructure underneath it. OpenAI and Anthropic are building toward the same dynamic.
What Enterprise Leaders Should Do With This Information
The entry of AI labs into the deployment market does not eliminate the need for independent AI strategy. It makes it more important.
Organizations that build their AI implementation capability entirely through a single vendor's deployment arm are making a strategic bet that the vendor's interests and their own will remain aligned over a multi-year horizon. Given the pricing dynamics, competitive pressures, and model strategy shifts that both OpenAI and Anthropic are navigating in real time, that is a bet worth examining carefully before making.
The organizations best positioned to take advantage of the deployment ventures without being captured by them are the ones that enter those relationships with clear strategic intent: defined business outcomes, explicit governance requirements, model-agnostic architecture where possible, and internal capability that gives them genuine optionality rather than dependency.
That means building internal knowledge of your own workflows, data landscape, and AI requirements before inviting any vendor's engineers to redesign them on your behalf. It means understanding your total cost of ownership across different model and implementation options before committing to a single vendor's ecosystem. And it means treating the deployment relationship as a capability transfer program rather than a managed service, so that the organizational knowledge built during the engagement stays inside your organization when the engagement ends.
The entry of OpenAI and Anthropic into enterprise deployment is net positive for the market. It will accelerate adoption, raise the quality of implementation, and help more organizations move from pilot to production. The organizations that benefit most will be the ones that approach these relationships as informed buyers with clear requirements, not as organizations handing their AI strategy to a vendor because the vendor showed up with engineers and a compelling pitch.
That is the work KAIDATA is built for: helping organizations build the strategic clarity, data foundation, and governance infrastructure that makes every implementation relationship, including relationships with the labs themselves, deliver on its stated value rather than create the dependency it was supposed to solve.
