Is MCP Likely to Achieve Industry Adoption? (hint: we think so)

Picture this: you're building an AI agent that needs to talk to your CRM, pull data from your database, and send notifications through Slack. Right now, you'd have to build custom integrations for each tool, dealing with different APIs, authentication methods, and data formats. It's a mess.

That's exactly the problem the Model Context Protocol (MCP) is trying to solve. Think of it as USB-C for AI agents—one standard way for AI models to connect with any tool or data source.

The big players are already on board

Here's what caught my attention: three of the four major AI labs have already lined up behind MCP. We're talking about OpenAI, Google, and Anthropic—plus Microsoft is throwing its weight behind it too.

OpenAI has baked MCP into their Agents SDK, ChatGPT desktop app, and their new Responses API. Google DeepMind integrated it into the Gemini SDK. Microsoft went all-in with native Windows support and their AI Foundry platform.

The only holdout? Meta. They're still focused on open-source frameworks like LlamaIndex and LangChain, but no MCP integration yet.

Why this matters now

Current AI function-calling APIs are a nightmare. Every vendor does it differently. If you want to build an agent that works across platforms, you're basically starting from scratch each time.

MCP solves this the same way REST solved the complexity of SOAP in the early 2000s. It gives developers a standard way for AI agents to discover tools, request resources, and manage context across conversations.

The timing is perfect too. The agent platform ecosystem is still forming, and standards set now will have massive network effects later. Plus, enterprise buyers are demanding vendor-neutral solutions to reduce AI deployment risks.

How fast is MCP growing?

The adoption numbers are honestly impressive. In less than 9 months, there are already thousands of MCP integrations. To put that in perspective, here's how MCP compares to other protocol adoptions:

• SOAP (1998): Took about 20 months to get multi-vendor support from IBM and Microsoft

• REST (2000): About 24 months before eBay and Amazon jumped on board

• GraphQL (2015): Around 14 months to get GitHub and AWS integration

• MCP (2024): Less than 7 months to get Anthropic, OpenAI, Google, AND Microsoft

That's unprecedented speed for a new protocol.

The missing pieces

MCP isn't perfect yet. There are some real challenges:

Security is the big one. AI agents with tool access create new attack vectors—token theft, prompt injection, tool poisoning. The major providers are being cautious, rolling out gated previews with strict guidelines.

Governance is another issue. Right now, it's still vendor-led by Anthropic. For true industry adoption, MCP needs neutral governance—think how GraphQL moved to the GraphQL Foundation in year three.

Competition could fragment things. Proprietary SDKs, gRPC alternatives, or other standards could split the market.

Will MCP actually win?

I'm putting the odds at 65% that MCP becomes the dominant standard within five years. Here's my reasoning:

The vendor alignment is incredibly strong (8/10). Having Anthropic, OpenAI, Google, and Microsoft all backing the same protocol is rare.

Developer experience is solid (7/10). It's built on familiar JSON-RPC patterns, has good documentation, and growing open-source server ecosystem.

But security concerns (5/10) and lack of neutral governance (4/10) are real drags on adoption.

What we're doing about it

At Growth Method, we think MCP has the momentum, money, and product-market fit to win. The security issues are getting addressed, and neutral governance will likely follow as the protocol matures.

So we're standardising our data integrations on MCP, with Zapier and custom integrations as fallbacks. It's a bet on where we think the ecosystem is heading.

The bottom line? MCP feels like being at the REST vs SOAP crossroads in 2002. The simpler, more practical solution usually wins—and that's exactly what MCP represents for the AI agent world.

If you're building anything that connects AI to external tools or data, MCP is worth learning now. The network effects are already starting to kick in, and being early to a winning standard pays dividends for years.

"We are on-track to deliver a 43% increase in inbound leads this year. There is no doubt the adoption of Growth Method is the primary driver behind these results."

Laura Perrott, Colt Technology Services

Growth Method is the GrowthOS built for marketing teams focused on pipeline — not projects. Book a call at https://cal.com/stuartb/30min.