Enterprises are running 3.1 orchestration platforms on average, not one. Across 107 companies surveyed by VentureBeat in July 2026, 85% run two or more platforms and 64% run three or more. Microsoft AI Foundry appears in 70% of stacks, OpenAI's Agents SDK in 68%, Anthropic's Claude Platform in 47%. When forced to name a single primary platform, 41% of the 61 respondents who could answer unambiguously said Microsoft, 28% said Anthropic. Nobody is running one orchestration layer and calling it a strategy.
The buying logic is governance, not developer comfort. Flexibility across models is the top purchase driver at 29%, nearly three times the 10% who cite native alignment with a preferred base model. Security and permissions (17%), production reliability (15%), and execution control (15%) follow. By 2026, 53% expect a hybrid control plane combining provider-native and external orchestration. The risk enterprises fear most from provider-resident control is not vendor lock-in at 23%, but security and permissioning limits at 37%. Spend confirms the priority: agent monitoring leads at 31%, security enforcement at 30%, workflow tooling at 19%.
The soft spot is fiscal control, and it is significant. Twenty-one percent of enterprises have no real-time way to stop a runaway agent before the bill arrives, relying only on post-hoc logs. Separately, 47% of respondents say only 26 to 50% of their deployed agents are genuinely orchestrated, with 37% at a quarter or below. Most enterprise agents are chatbots wearing an agent label. The full report is worth reading for how it breaks down the gap between orchestration ambition and orchestration reality, and for what value-for-money satisfaction scores reveal about where this market is heading.
[READ ORIGINAL →]