Foxconn, SAP Partner to Expand AI Manufacturing in APAC

Hon Hai Technology Group (Foxconn) has announced a partnership with SAP, the enterprise software provider, to accelerate enterprise AI adoption across the Asia-Pacific (APAC) region. The partnership was announced at NVIDIA’s GTC event in San Jose.

The collaboration aims to combine Foxconn’s AI computing capabilities with SAP’s enterprise software to advance AI-led manufacturing and supply chain systems. The companies will also explore co-innovation in areas such as physical AI, smart manufacturing, and supply chain management.

Foxconn said the partnership builds on its AI Factory initiative, which focuses on transforming manufacturing processes and supply chains. The company plans to use SAP’s enterprise applications to support its transition to a fully digitised, AI-driven organisation.

“By bringing together Foxconn’s AI computing capabilities with SAP’s expertise in enterprise applications, we are addressing one of the most complex challenges facing global industries today,” said Young Liu, Chairman of Foxconn.

Foxconn and SAP also aim to support go-to-market initiatives across APAC. The effort will focus on industries that require large-scale manufacturing and logistics operations.

Foxconn said the partnership will help define new models for enterprise AI deployment. The company expects the collaboration to create pathways for real-world industrial use of AI technologies.

SAP said the collaboration will help companies in the region adopt AI faster and scale operations. “Together, we are helping organisations harness technology that helps them move faster, operate more efficiently, and scale with confidence as they transform in the AI era,” said Christian Klein, CEO of SAP.

In October last year, Foxconn had also launched a traditional Chinese LLM code-named FoxBrain. Initially designed for internal group applications, it was set to become an important engine for Foxconn’s three major platforms: smart manufacturing, smart EV, and smart cities.

In an attempt to extend AI manufacturing, FoxBrain was trained on 120 NVIDIA H100 GPUs and scaled with NVIDIA Quantum-2 InfiniBand network. The model is also based on Meta’s Llama 3.1 architecture, with 70B parameters and a context window length of 128k tokens. 

This kind of unique adaptive reasoning reflection technique was used to train the model in autonomous reasoning.