MCPs 2.0: Why The Protocol War Will Define The Next Decade of AI

In partnership with

Have you heard about MCPs yet? If not, you should be paying attention.

While everyone's distracted by the latest AI demos and performance benchmarks, a much more important battle is happening behind the scenes – one that will determine whether AI actually becomes useful or remains a glorified chat toy.

The Protocol Problem

I've been thinking about this a lot since my last letter. Here's the reality: LLMs by themselves are fundamentally limited.

Think about the early web before HTTP standardized how browsers and servers communicate. Every website needed custom programming to talk to every browser. It was a nightmare that prevented scale.

We're at that exact moment with AI right now.

Why This Matters Now

The companies that seem to be "winning" the AI race (OpenAI, Anthropic, Cohere) are actually all hitting the same ceiling: they can chat impressively, but they struggle to do anything useful on their own.

That's because each AI action requires custom integration work:

• Want to check your email? Custom connector needed.

• Need to update your CRM? Another custom connector.

• Update your calendar? You guessed it.

This is the dirty secret of all those cool AI demos. Behind the scenes, there's an army of engineers custom-building every single integration.

The MCP Solution

Machine Communication Protocols (MCPs) are attempting to solve this by creating a standard language for AI systems to communicate with external tools.

Instead of every AI needing to learn how to talk to each service individually, MCPs create one universal translator sitting between your AI and the digital world.

The difference is profound:

• Without MCPs: Each AI-to-service connection needs custom coding

• With MCPs: Build once, connect to everything

The Protocol Wars Have Begun

Here's where it gets interesting. Multiple companies are now competing to establish their protocol as the standard:

1. Anthropic's Claude Protocol: The early frontrunner with major backing

2. OpenAI's Assistants API: Leveraging their market position

3. Open Standards Groups: Working on vendor-neutral alternatives

This isn't just technical minutiae. The company that establishes the dominant protocol will have enormous power in the AI ecosystem – similar to how controlling HTTP would have given you influence over the entire web.

What This Means For You

If you're building or investing in AI, this battle has major implications:

1. Tool Makers: Build for multiple protocols until a winner emerges, but watch market share carefully

2. AI Developers: Don't build deep custom integrations yet; wait for protocol standardization

3. Investors: Look beyond model performance to protocol adoption metrics

The businesses that position themselves correctly for the post-protocol world will have tremendous advantages when AI can finally start doing useful things.

The Real Timeline

Despite all the hype, truly useful AI that can actually accomplish tasks (not just talk about them) is still years away. The protocol standardization is a necessary step, and we're just at the beginning.

But once it happens, the acceleration will be dramatic. Just as the standardization of HTTP unleashed the web revolution, MCPs will unlock the true potential of AI to act on our behalf.

The Big Picture

The most exciting part isn't that AI can chat with you in a more natural way. It's that soon, AI will actually accomplish things for you rather than just discussing them.

The bridge to that future isn't more powerful models with higher token counts. It's the boring, technical, unsexy protocol standardization happening right now.

And the winners of that standardization battle may end up being more important than the model providers themselves.

Until next time...