The Model Is Not the Problem: OpenAI and Anthropic Spend Billions on New AI Deployment Ventures

In May 2026, OpenAI and Anthropic each put money behind the same conclusion, independently, nine weeks apart. Better models aren't what's holding enterprise AI back. Getting them installed, integrated, and used is. Neither company built another model to prove it. They built deployment companies, and this month Anthropic's went fully public with the kind of detail that makes the bet easy to see clearly.

Two labs, the same bet, weeks apart

OpenAI moved first. On May 11, it launched The Deployment Company, a joint venture majority-owned by OpenAI that raised more than $4 billion from 19 investors anchored by TPG, with Advent International, Bain Capital, Brookfield, and Goldman Sachs among the named partners. The venture is built around embedding Forward Deployed Engineers, OpenAI's term for hands-on implementation staff working inside a customer's environment, a model borrowed from Palantir's playbook. OpenAI paired the launch with an acquisition: Tomoro, a London-based AI consulting firm with roughly 150 Forward Deployed Engineers already on staff and an existing client list including Mattel, Red Bull, Tesco, and Virgin Atlantic.

Nine weeks later, Anthropic followed with Ode, a $1.5 billion joint venture backed by Blackstone, Hellman & Friedman, and Goldman Sachs. Where OpenAI's move first read as a quieter services play, Anthropic's went fully public this month with its structure, its staffing, and its ambitions on the record. Ode launched with 100 engineers, described by one Blackstone executive as "special forces" rather than a large implementation army, and its CEO told TechCrunch plainly: "It's pretty easy to imagine this as a trillion-dollar company someday if we execute well."

What these companies are selling

Plenty of companies already use ChatGPT or Claude. That's not what either venture is built to fix. The role at the center of both, Forward Deployed Engineering, borrowed from Palantir's playbook, is a different thing from API access, and a different thing from traditional consulting too. Job postings from both labs describe the same concrete deliverables: build custom integrations, sub-agents, and connectors into a client's existing systems, not a generic template; deploy so the result runs behind the client's authentication, logging, and incident response, not just in a demo; then codify what was built into a reusable pattern the client's team can extend afterward. OpenAI calls that last step "durable productization." Anthropic calls it "repeatable deployment patterns." One writer summarized the distinction from traditional consulting bluntly: Consultants write reports and recommendations. A Forward Deployed Engineer builds the system and stays until it runs in production.

In practice, that spans functions rather than sitting in one department. One reported Anthropic engagement embedded engineers with a fintech infrastructure company to co-build an anti-money-laundering agent, a compliance and finance workflow. Industry analysis of over a thousand Forward Deployed Engineer job postings found healthcare and insurance, claims automation, underwriting, policy document processing, among the most common verticals, alongside finance. The large majority of postings don't name a specific department at all, because the role is built to be vertical-agnostic: Whatever workflow a client needs built, in marketing, sales, finance, HR, or elsewhere, the engineer embeds and builds it into that team's systems rather than handing over a generic tool and leaving.

MIT's Project NANDA studied 300 public AI projects and found 95% of enterprise pilots produced little or no measurable financial return. The reporting on both new ventures states the diagnosis plainly: The problem was not the models, but how they were put to use. That's not a rationalization built after the fact. Both labs put billions behind that exact read on the market, in the same six-to-nine-week window, without coordinating with each other.

Neither lab is alone in reaching that conclusion. Deloitte has stood up a dedicated forward-deployed engineering practice, and Accenture launched a Microsoft-aligned forward-deployed engineering offering on a similar timeline. The traditional systems integrators, the firms that have run enterprise implementation for decades, are racing to match the labs' move rather than dismissing it as a passing trend.

What to watch next

Ode's backers are expected to route their portfolio companies to it as early customers, giving the venture a built-in pipeline that The Deployment Company doesn't have in the same form. Both ventures are still young enough that neither has published enterprise case studies showing results at scale, which is the test of whether embedded-engineer implementation closes the gap the failure-rate research describes, or simply moves the same challenges into a better-funded services wrapper. Worth tracking whether either company's results, not its launch coverage, back up the thesis a year from now.

Sources: AI Pulse · Big Picture · workplaceai.ai. OpenAI's Deployment Company and Tomoro acquisition: OpenAI's announcement, May 11, 2026, corroborated by Forbes, May 28, 2026, The New Stack, May 28, 2026, and MarkTechPost, May 21, 2026. Ode's formal launch details, staffing, the Fractional AI acquisition, and the Chris Taylor quote: TechCrunch reporting by Rebecca Bellan, July 2026, as compiled by MarketScale, MLQ News, and Technology.org. MIT Project NANDA's 95% figure: as reported by The New Stack, May 28, 2026. Deloitte and Accenture's forward-deployed engineering practices: Deloitte and Accenture newsroom announcements, as compiled by MarketScale. Every figure above is attributed to its original reporting; none is a WorkplaceAI study.