Agent interoperability protocols Market Outlook 2026–2033: Growth Drivers, Key Companies, Strategic Developments & Adoption Forecast — A Meticulous Next™ Foresight Brief

This Meticulous Next™ brief examines agent interoperability protocols, the open standards that let AI agents built by different vendors discover each other, prove who they are and hand work to one another. It tests how far open protocols become the common language of multi-agent systems by 2033, and what that means for cloud vendors, software providers, security teams and the enterprises that run agents across company boundaries.

The standards have moved under neutral governance. Google introduced the Agent2Agent protocol on 9 April 2025 and transferred it, with its specification and software development kits, to the Linux Foundation on 23 June 2025, when more than 100 technology companies had announced support. The conclusion is that a protocol started by one cloud vendor is now run as a shared project, which lowers the risk for rivals that build on it.

Support keeps broadening. Microsoft stated in May 2025 that it planned to support the protocol in its AI platforms, AWS, Cisco, Salesforce and SAP joined the project, and the protocol site now lists more than 170 community partners and six official software development kits. The conclusion is that the largest platform vendors have agreed to speak the same language, so the contest has moved from whether a standard exists to which layers each standard covers.

Several efforts run in parallel. AGNTCY, started by Cisco, launched under the Linux Foundation in July 2025 with more than 65 companies and covers discovery, identity, secure messaging and observability, while the Agentic AI Foundation, announced on 9 December 2025, brought the model context protocol, goose and AGENTS.md together and had more than 170 members by April 2026. The conclusion is that tool access, agent to agent calls, identity and observability are being standardised as separate layers that need to fit together.

Gaps remain in identity, trust and governance. A protocol can carry a request between agents, but it does not decide whether the receiving agent may act, who is liable for the result or how the call is audited. The conclusion is that gateways, registries, identity services and audit tools become the paid part of the market, while the protocols themselves stay free.

The brief treats agent interoperability protocols as an accelerating signal with a seven year horizon to 2033. It identifies which vendors, platforms and security providers capture value as agents begin to work across company boundaries. It is a focused 110-page decision brief that presents an indicative trajectory rather than a segmented market model.

A protocol lets one agent ask another to do work. Agent2Agent describes an agent through a card that states what it can do and how to reach it, then lets a client agent send tasks, receive progress updates and collect results without seeing the memory, tools or logic of the other agent. The conclusion is that a company can offer an agent as a service to partners without exposing how it works, which is what makes cross-company use possible.

The standards cover different layers. The model context protocol connects an agent to tools and data, Agent2Agent connects agents to each other, and the AGNTCY project covers discovery, identity, secure messaging and observability. The conclusion is that these standards complement more than compete, and buyers need products that join the layers rather than a single winner.

Security rests on existing web standards. Agent2Agent supports TLS, JSON web tokens and OpenID Connect, and signed agent cards were added in a later version of the specification, with the protocol reaching version 1.0 and gRPC support in the SDKs. The conclusion is that identity can build on tools that enterprises already run, although rules for delegation between agents of different companies are still being written.

Neutral governance reduces the risk of building. The Agentic AI Foundation is a directed fund under the Linux Foundation, with members paying tiered dues and nominating representatives to technical steering committees, and no single member has unilateral say over direction. The conclusion is that enterprises can commit to a protocol for years without depending on one vendor, which supports multi-year agent programmes.

The limits are weak cross-company trust rules, overlapping standards and the cost of testing agents from many vendors against each other. The brief tests each limit against the evidence to 2033.

The agent interoperability protocols market, covering paid gateways, registries, discovery and identity services, observability tools and integration services that let AI agents from different vendors find, authenticate and work with each other across protocols such as Agent2Agent, and excluding the AI models themselves, tool connection servers for single agents, agent development frameworks and internally built connectors that carry no vendor spend, is valued at USD 0.6–0.9 billion in 2026 and is projected to reach USD 5.72–8.58 billion by 2033, at a CAGR of 38%.

The 2026 base rests on gateway, registry, identity, observability and integration spending by cloud providers, software vendors and early enterprise programmes. Growth then follows the number of agents that call agents from other vendors, the number of companies that publish agents to partners and the share of enterprises that buy trust and audit controls rather than build them.

The forecast is sensitive to the pace of enterprise agent adoption, the fit between competing standards and the arrival of liability and audit rules for agent actions.

The base case assumes steady growth in agent programmes with standards that fit together, bringing the inflection to about 2030 and a 2033 value near the middle of the range. An accelerated case assumes fast adoption of agents across companies and early rules on delegation and audit, moving the inflection to 2029 and lifting the 2033 value to the top of the range. A delayed case assumes split standards or a trust failure between agents that slows cross-company use, moving the inflection to 2032 and holding growth at the bottom of the range.

Growth drivers

• Enterprise need for agents from different vendors to work together.
• Support from every large cloud and software platform.
• Neutral governance under the Linux Foundation.
• Falling cost of building and testing agent connections.

Enablers

• Official software development kits in several languages.
• Agent cards, registries and discovery services.
• Identity based on OpenID Connect and signed credentials.
• Gateway, observability and audit products.

Restraints and barriers

• Weak rules for trust and liability between companies.
• Overlapping standards and version changes.
• Cost of testing and governing many agents.
• Security risks from agents that act on delegated authority.

The direction of agent interoperability protocols is set by enterprise need for agents from different vendors to work together, support from every large cloud and software platform, neutral governance, the maturity of identity and audit controls and the readiness of software vendors to publish agents that others can call.

The key participants in agent interoperability protocols and the open standards for agent communication include Alphabet Inc., Microsoft Corporation, Amazon Web Services, Inc., Cisco Systems, Inc., Salesforce, Inc., SAP SE, International Business Machines Corporation, Oracle Corporation, ServiceNow, Inc., Anthropic PBC, OpenAI, Inc., Block, Inc., Dell Technologies Inc., Workday, Inc., Atlassian Corporation, Okta, Inc., Palo Alto Networks, Inc., Kong Inc., LangChain, Inc., The Linux Foundation, Cloudflare, Inc., Bloomberg L.P., Stripe, Inc., Datadog, Inc., Elastic N.V., Zapier, Inc., Workato, Inc., CrowdStrike Holdings, Inc., Zscaler, Inc., Databricks, Inc., Snowflake Inc., Hugging Face, Inc., CrewAI, Inc., Accenture plc, Infosys Limited, Tata Consultancy Services Limited, Wipro Limited, Capgemini SE, Cognizant Technology Solutions Corporation, the Agentic AI Foundation and the AGNTCY project.

The brief profiles representative participants in each archetype and assesses which are positioned to lead as open protocols become the common language of multi-agent systems.

Alphabet Inc., Microsoft Corporation and Amazon Web Services, Inc. lead among cloud platforms that host and register agents, Cisco Systems, Inc., Okta, Inc. and Palo Alto Networks, Inc. lead among providers building identity, security and observability, and Salesforce, Inc., SAP SE and ServiceNow, Inc. lead among software vendors whose agents other agents will call. Competitive intensity is rising as cloud providers, security vendors and start-ups all move into the registry and gateway layer.

Where value migrates.

In 2026 value sits with cloud platforms and enterprise software vendors that host agents and with model builders that create demand, while the protocols are free and most connections are built in house. By 2030 it moves toward registry, identity, gateway and audit providers, because enterprises will not let agents from other companies act on their systems without trust controls. By 2033 it settles with the vendors that own the agents, data and actions that others call, where open protocols are a standard layer and margin comes from trusted access and results. Margins hold for platforms with registries, identity providers with large customer bases and software vendors with widely used agents, and narrow for builders of single point connectors that platform releases replace.

More than 170 community partners for Agent2Agent, more than 170 members of the Agentic AI Foundation, support from every large cloud and a move to neutral governance support growth, while weak cross-company trust rules, overlapping standards and the cost of governing many agents hold back the pace.

A trust failure between agents of different companies, a split between standards backed by large vendors or a slowdown in enterprise agent programmes could slow progress. The brief therefore tracks four indicators: the number of published agents and registered agent cards, the share of agent traffic that crosses company boundaries, the pace of specification changes and the number of reported security incidents involving delegated agents.

Standards settling first, then cross-company agents as a normal part of workflows, then open protocols as a standard layer under agent traffic is the expected order of change.