TL;DR
OpenAI and Anthropic made parallel moves in early May 2026 to build enterprise AI deployment businesses around embedded engineers. The shift matters because the labs are moving beyond model sales into the larger services layer where companies struggle to put AI systems into production.
OpenAI and Anthropic moved in early May 2026 to build services-heavy enterprise AI deployment businesses, a shift that could change how frontier model companies earn revenue and how large firms put AI systems into production.
According to the source material, Anthropic announced a $1.5 billion enterprise-services venture with Blackstone, Hellman & Friedman, and Goldman Sachs to place Claude inside mid-market companies. Hours apart, OpenAI announced a $4 billion Deployment Company, known as DeployCo, at a $10 billion pre-money valuation, with 19 investment partners.
OpenAI’s new structure also included the acquisition of consulting firm Tomoro, bringing 150 forward-deployed engineers into the company on day one, according to the source. Those engineers are expected to work directly with client teams, study workflows, build software around frontier models, and remain involved until the systems work in production.
The source describes both moves as modeled on Palantir’s forward-deployed-engineer approach, where technical staff work inside customer operations rather than selling software from a distance. The confirmed development is the parallel move by two major labs into deployment services; the broader claim is that this marks vertical integration into the enterprise services layer.
Why It Matters
The move matters because enterprise AI adoption has been slowed less by model access than by implementation work: integration, security review, evaluation, governance, and process redesign. The source cites OpenAI’s framing that model performance is no longer the main constraint, and says the labs are targeting the work that turns experiments into working systems.
The economic case is tied to the services market. The source states that companies spend about $6 on services for every $1 spent on software, making implementation far larger than model licensing alone. If the labs can convert deployment work into ongoing model usage, the embedded customer could become a recurring, usage-based revenue source rather than a one-time consulting project.
The risk is that the model may require large amounts of human labor. If each customer needs a proportional number of engineers, margins could look more like consulting than software. If the deployment work produces reusable systems and customer-specific software that expands usage over time, the labs may gain stronger retention and higher enterprise revenue.
Autonomous AI-Driven Enterprise Software From Development to Deployment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background
Palantir developed the forward-deployed-engineer model through years of work with defense, intelligence, and large institutional customers. In that model, engineers do not only advise; they build operational systems close to the customer’s real workflows.
The source frames the OpenAI and Anthropic announcements as part of a larger shift in enterprise AI. Many companies have run generative AI pilots, but fewer have moved those systems into production. The source points to MIT research saying 95% of generative AI pilots fail to move past the experimental phase, using that figure as the data behind the labs’ move into deployment.
“the model isn’t the bottleneck, deployment is”
— Thorsten Meyer AI source material
“copied from Palantir almost line for line”
— Thorsten Meyer AI source material
“for every $1 on software, companies spend $6 on services”
— Thorsten Meyer AI source material
“resembles consulting more than pure software licensing”
— Thorsten Meyer AI source material
The Authentic Pocket Engineer – Small Metal Engineering Ruler Protractor Compass Scale, Techie Graduation Gadget Multitool, Mechanical/Civil Engineers Gift, Metric Mini 3 Inch Tool- Genius Lab Gear
Credit Card-Sized Multitool for Engineers: This stainless-steel laser-cut pocket tool includes a straight-edge ruler, protractor, and a compass….
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
It is not yet clear whether these deployment arms will scale like software businesses or remain labor-heavy services operations. The source frames that as the central open question inherited from the Palantir model.
Details are also still limited on customer uptake, pricing, contract structure, margins, and how much of the work done by forward-deployed engineers will become reusable product rather than client-specific implementation.
Your AI Survival Guide: Scraped Knees, Bruised Elbows, and Lessons Learned from Real-World AI Deployments
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
The next test is execution: whether OpenAI and Anthropic can move enterprise customers from pilots into production systems while keeping deployment costs under control. Investors and customers will watch for signs of repeatable workflows, rising model usage inside deployed accounts, and evidence that embedded engineering teams create durable revenue rather than one-off services work.
AI Systems for Churches: How to Use Artificial Intelligence in Teaching, Communication, and Ministry Leadership (The AI Systems Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened in early May 2026?
Anthropic announced a $1.5 billion enterprise-services venture, while OpenAI announced a $4 billion Deployment Company and acquired Tomoro, according to the source material.
What is a forward-deployed engineer?
A forward-deployed engineer works directly with a customer’s operators, studies workflows, builds software around the customer’s real problems, and stays involved until the system works in production.
Why are AI labs moving into services?
The source says the labs see deployment, integration, security review, evaluation, and process redesign as the main barriers to enterprise AI adoption. Services also represent a larger spending pool than software alone.
Is this confirmed or an interpretation?
The announced ventures and acquisition are treated as the reported developments in the source material. The argument that this is vertical integration into the services layer is the source’s interpretation of those moves.
What remains unclear?
The open question is whether the model can scale with software-like margins or whether it will remain constrained by the cost of human deployment teams.
Source: Thorsten Meyer AI
