- Mplify Alliance is reframing its work around network-as-a-service and AI workloads
- Automation APIs, common data models and MCP servers are central to its strategy
- The group is positioning operators for what it calls an emerging 'AI internet,' not best‑effort networks
MOBILE WORLD CONGRESS 2026, BARCELONA — For years, the organization now known as Mplify Alliance was closely tied to Metro Ethernet standards and carrier connectivity. That heritage still matters, but it no longer defines the mission, according to CTO Pascal Menezez who caught up with Fierce Network at Mobile World Congress 2026, where it was demoing work with GSMA in the latter's exhibit area.
The demo showed off a collaboration with Colt Technology Services, Orange, Google Cloud, and GSMA Open Gateway, showing how agentic AI combined with standardized network APIs, enables closed-loop orchestration across wireless and wired networks. Automotive innovators DENSO and Tata Elxsi provided the real-world use cases for how these aforementioned work can support connected vehicles and autonomous systems.
Formerly known as the MEF Forum, the name change to Mplify in June 2025, was a broader shift for the organization: from point technologies like Ethernet to a more expansive framework built around network-as-a-service (NaaS), said Menezez. The group has steadily layered in IP services, wavelengths, SD‑WAN and SASE, using open APIs to automate how services are ordered, provisioned and managed across providers.
“That’s what the market needed for business workloads — until AI changed the equation,” Menezez said.
Re-architecting for AI workloads
AI requires a fundamentally different network fabric than traditional enterprise traffic, he said. Rather than starting from scratch, Mplify is repurposing its existing assets — automation APIs, service models and certifications — into an AI context.
A major step is wrapping those APIs into model context protocol (MCP), allowing AI systems to interact with networks natively. Equally important are the data models that define relationships, states and behaviors across Ethernet, IP, wireless and security services — context that AI agents need to operate reliably.
Mplify is also developing agent blueprints, not agents themselves. The idea is to give operators and developers a standardized guide that can be dropped into any agent framework — whether from hyperscalers or open platforms — to generate agents that work consistently against provider MCP servers.
Toward an 'AI internet'
The longer-term vision is an AI federation or AI exchange, where providers interconnect to support training and inference workloads with cloud-like speed and automation. That shift also pulls security into the spotlight, with new work underway on AI-focused cybersecurity, policy enforcement and auditing, said Menezez.
Read all of our coverage from Mobile World Congress 2026 in Barcelona here.