← Back to feed

OpenAI has built out the enterprise AI services model it spent 2025 drawing on whiteboards. The Wall Street Journal published a deep profile today of how DeployCo and the $150M OpenAI Partner Network are operating in practice, with BBVA as the named early engagement.

The structure is deliberate: DeployCo provides the direct, high-touch arm for the accounts that matter most, while the partner program funds the ecosystem that does the same work at scale for everyone else.

The DeployCo Operating Model

The OpenAI Deployment Company launched in May 2026 with the acquisition of Tomoro, an AI consulting firm that contributed approximately 150 forward deployed engineers. A second acquisition, Northslope — announced July 8 via Axios — pushed the FDE headcount into the hundreds. A subset of engineers from OpenAI’s own internal FDE organization is also on temporary loan as DeployCo scales.

The model is borrowed directly from enterprise software firms that have used embedded technical staff with strategic accounts for decades. FDEs arrive on-site, work alongside the client’s own engineers, and build production AI systems rather than handing over API credentials and documentation.

The firm is still selecting its first formal customers. DeployCo is prioritizing organizations with existing OpenAI relationships, which limits the sample but sets the credibility bar high from the start.

The BBVA engagement gives the first concrete data point. Fran Martínez López, head of risk transformation at BBVA, told the Journal that an OpenAI FDE joined his team after months of internal work on a credit risk analysis application had stalled. The engineer redesigned the system’s architecture using AI agents, and the bank completed a project it had not been able to close under its own effort. BBVA is named as a founding partner of DeployCo, and the relationship is expected to deepen.

The Partner Program Layer

The $150M OpenAI Partner Network, formally announced via the OpenAI blog this summer, creates the channel that carries the FDE methodology into accounts OpenAI and DeployCo cannot directly staff. The program runs three tiers — Select, Advanced, and Elite — with advancement tied to sales performance, technical capability, co-sell engagement, and deployment track record.

Partners can also earn specializations in Codex, cybersecurity, and agents, which signal verified depth in areas enterprise buyers treat as high-priority. OpenAI is targeting 300,000 certified consultants within the program by end of 2026.

A smaller Forward Deployed Experts program layers on top of the partner tier structure: certified partner practitioners gain exposure to OpenAI’s playbooks and engineering patterns, aligning their approach with DeployCo’s methodology for complex deployments.

The Strategic Logic

OpenAI’s leadership framed the problem precisely in the Journal: there has never been a greater gap between what AI models can do and what businesses are actually doing with them. DeployCo is the direct answer for the top of the market. The $150M partner program is the answer for everyone below that.

The competition for the same enterprise accounts is significant. Anthropic, which launched its own consulting arm on the same day as DeployCo in May, is competing for the same large accounts. Microsoft’s Frontier Company ($2.5B, 6,000 engineers) is operating in adjacent territory. A growing field of open-weight models gives cost-conscious enterprise buyers a third option that requires no per-token spend at all.

For enterprises evaluating AI vendors in the second half of 2026, OpenAI’s moves reframe the selection criterion. The question is no longer only which model benchmarks best on SWE-bench Verified. It is also which vendor can put credentialed engineers on-site, instrument the workflow, and iterate in production.

Whether the FDE model scales — and whether BBVA’s credit risk result is reproducible across the first cohort of formal DeployCo customers — is the next metric to watch.