The Illusion of Efficiency of the Model Context Protocol (MCP): Analysis of energy surcharges, carbon footprint, and infrastructural vulnerabilities of agentic LLM

The emergence of agentic architectures based on Anthropic's Model Context Protocol (MCP) is widely promoted as a breakthrough in interoperability between Large Language Models (LLMs) and external information systems. However, this paper demonstrates that the uncritical adoption of MCP introduces major engineering flaws that directly contradict the core principles of Green Tech and the Sustainable Web Manifesto. Through a critical evaluation of context token inflation, redundant GPU inference cycles, server thermal stress, and severe security risks (such as indirect prompt injection), this research exposes the hidden environmental and operational costs of autonomous agents. Finally, we propose a digital sobriety framework favoring deterministic algorithms and low-carbon local execution.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-12
DOI
https://doi.org/10.5281/zenodo.24165943
Primary Topic
Green IT and Sustainability
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Illusion of Efficiency of the Model Context Protocol (MCP): Analysis of energy surcharges, carbon footprint, and infrastructural vulnerabilities of agentic LLM

Rabii Jaafri
Zenodo (CERN European Organization for Nuclear Research)
Green IT and Sustainability
preprint

The Illusion of Efficiency of the Model Context Protocol (MCP): Analysis of energy surcharges, carbon footprint, and infrastructural vulnerabilities of agentic LLM

Rabii Jaafri
preprint en

Abstract

The emergence of agentic architectures based on Anthropic's Model Context Protocol (MCP) is widely promoted as a breakthrough in interoperability between Large Language Models (LLMs) and external information systems. However, this paper demonstrates that the uncritical adoption of MCP introduces major engineering flaws that directly contradict the core principles of Green Tech and the Sustainable Web Manifesto. Through a critical evaluation of context token inflation, redundant GPU inference cycles, server thermal stress, and severe security risks (such as indirect prompt injection), this research exposes the hidden environmental and operational costs of autonomous agents. Finally, we propose a digital sobriety framework favoring deterministic algorithms and low-carbon local execution.

Zenodo (CERN European Organization for Nuclear Research)
Climate action
Green IT and Sustainability
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon