The Infrastructure Layer Becomes the Battleground

Anthhropic’s Model Context Protocol (MCP) has quietly achieved a significant milestone: 97 million installs by March 2026. What started as a technical specification for connecting AI models to external tools has evolved into foundational infrastructure that every major AI provider now ships natively. This marks a subtle but profound shift in how the AI industry is structuring itself—moving beyond raw model capability competition toward standardized interoperability layers.

Why This Matters More Than It Appears

For the past 18 months, the AI industry narrative has centered on model releases: who has the biggest parameter count, lowest latency, or most creative capabilities. MCP’s ascent signals something different. The protocol enables AI agents—systems that autonomously perform tasks across multiple tools and APIs—to function reliably. With the Linux Foundation now assuming governance responsibilities, MCP is transitioning from proprietary advantage to open infrastructure.

This is particularly significant for European and Irish technology organizations. EU regulatory frameworks like the AI Act increasingly demand transparency and interoperability. A standardized protocol for how AI systems interact with external services could actually facilitate compliance, allowing organizations to audit and control what data flows between systems more effectively.

What This Means for Builders

If you’re developing AI applications or evaluating agent platforms, MCP compatibility is no longer a nice-to-have—it’s becoming table stakes. The 97 million install base represents a network effect. Every integration built on MCP becomes more valuable as the ecosystem grows.

For Irish tech companies specifically, this creates opportunity. Rather than competing on model training (capital-intensive, dominated by US incumbents), Irish and European firms can focus on domain-specific agent applications built atop standardized infrastructure. Companies providing industry-specific integrations—healthcare, fintech, manufacturing—suddenly have a clear technical pathway to interoperability.

The Linux Foundation’s involvement also signals maturity. Open governance structures reduce the risk that a single vendor changes terms or prioritizes their own needs over the community. This matters for enterprise adoption in regulated industries.

Questions Still Unresolved

A few critical gaps remain unclear. First, how will MCP handle real-time security updates as the attack surface grows? With 97 million connections, vulnerabilities in the protocol or popular integrations could cascade quickly. Second, will governance truly be neutral, or will founding members (Anthropic, OpenAI, Google, Meta) hold disproportionate influence? Third, how do compliance and audit trails work for organizations subject to GDPR or AI Act requirements when data flows through distributed agent networks?

As the AI industry enters its infrastructure phase, MCP’s trajectory suggests the next wave of competitive advantage won’t come from who builds the best black-box model, but who creates the most useful, trustworthy systems operating on top of shared standards.

Source: Industry Analysis