Nearly nine in ten organizations report regular AI use. But when McKinsey asked about scaling agents, the picture inverted: 23 percent are scaling them anywhere, and no single business function has crossed 10 percent. A lot of experimentation. Very little that looks like deployment.
Into that gap, OpenAI just announced a $150 million partner network with a target of 300,000 certified consultants by the end of 2026. The framing matters: OpenAI itself argues the enterprise bottleneck is "no longer model capabilities" but integration, workflow redesign, governance, and change management. That's a vendor's positioning, obviously. But it's worth paying attention when the model provider says the constraint has moved downstream.
The pattern fits what Menlo Ventures found in late 2025: 76 percent of enterprise AI use cases are now purchased rather than built internally, up from 53 percent a year earlier. Enterprises aren't short on ambition or budget. They're short on the organizational capacity to make AI operational, and OpenAI is betting that manufacturing that capacity at scale is where the leverage sits now.
The adoption gap: 88% of orgs use AI regularly; only 23% are scaling agents. No function above 10% scaled.
The investment: OpenAI committing $150M to partner ecosystem, targeting 300,000 certified consultants by end of 2026.
Buy over build: 76% of enterprise AI use cases purchased vs. 53% a year prior (Menlo Ventures, U.S. enterprise sample).
What high performers do differently: McKinsey's top cohort, orgs attributing 5%+ of EBIT to AI, are nearly 3x more likely to have fundamentally redesigned workflows.
The EBIT reality: Only 39% of respondents attribute any EBIT impact to AI at all. Most of those say less than 5%.
A caveat on framing: OpenAI's "model capability is no longer the bottleneck" is their positioning, not an independent finding. The partner program demonstrates supply-side investment, not verified enterprise outcomes.