This paper presents a decentralized procurement decision-making framework designed to optimize procurement strategies within distributed manufacturing networks. Traditional centralized procurement models often suffer from inefficiencies, opaque decision-making, and challenges related to multi-level risks, particularly in the context of globalized supply chains. To address these issues, this project proposes a decentralized architecture utilizing predictive modeling and machine learning techniques, supported by decentralized ledger technologies (DLT) such as blockchain. The framework integrates large language models (LLMs) to forecast supply chain risks, demand fluctuations, and pricing trends. Furthermore, smart contracts are employed to automate the procurement process, ensuring transparency, security, and compliance. The system incorporates real-time feedback mechanisms to enhance decision-making accuracy, reduce lead times, and mitigate procurement risks. Testing in collaboration with multiple manufacturing firms revealed improvements in procurement efficiency, supply chain resilience, and risk management. The paper concludes by highlighting the potential of decentralized procurement frameworks to revolutionize supply chain management across various industrial contexts.