The AI Control Plane covers the gateways, governance, orchestration, and observability infrastructure now emerging to govern autonomous AI agents. A few takeaways stand out:
Enterprises are adopting agents faster than they can control them. Hyperscaler capex across Amazon, Alphabet, Meta, Microsoft, and Oracle has gone from $37B to $182B per quarter since early 2022, and the AI agent market itself is on a 46% CAGR to roughly $53B by 2030. Yet more than half of enterprises already have agents in production or proof-of-concept, and only 12% say they can actually control them. That gap is what's pulling budget toward this category. Recent high profile security breaches of production systems with the LLMs has turned control into a board-level issue.
Different acquirer types are converging on the same layer. Security platforms are buying their way in (Palo Alto/Portkey), data platforms are building and buying against it (Databricks' $188B round earmarked partly for gateway and M&A; Snowflake/Natoma), and hyperscalers are bundling it into infrastructure. Meanwhile OpenAI, Anthropic, and Google each shipped enterprise governance in 2026, but every one of them governs only its own estate – cross-provider control is structurally an independent's market, and the last time this setup appeared, in API management, the neutral layer didn't stay independent for long (Apigee went to Google, MuleSoft to Salesforce, both at strategic premiums).
Capital is moving ahead of the category's own maturity. M&A volume more than tripled from 15 deals in 2023 to 48 in 2025 and is tracking at a similar pace in 2026, while disclosed financing has scaled from $2.5B in 2022 to an $8.1B annualized run rate this year. Minority rounds can be a fast lane to M&A: Oasis Security went from a Series B to a roughly $1B sale in four months.
AGC has completed over 70 Infrastructure Software transactions, and we are in market with several AI-native infrastructure software financings and M&A transactions. We believe the control plane is forming now, ahead of any consensus on what a mature version of it looks like, which is exactly why the buyers defining the category are moving early.
