Respondents believe the integration of Internet of Things (IoT) data into generative AI models will significantly enhance the accuracy and relevance of AI-generated outputs. They also expect the convergence of digital twins with generative AI to improve physical asset performance and supply chain resilience.
“AI is streamlining processes and redefining what’s possible across the entire manufacturing value chain, from supply chain predictions to quality control. Generative AI can help organisations achieve flexibility in fast-changing business environments, especially in the face of uncertain tariff policies worldwide,” says Prasoon Saxena, co-lead for products Industries at NTT Data.
Despite the widespread adoption and benefits of generative AI, Apac manufacturers face significant challenges in fully realising its potential, NTT Data's report reveals.
Ageing infrastructure is a primary concern, with 91% of respondents citing legacy technologies as a hindrance to crucial AI initiatives. Alarmingly, less than half have conducted a thorough assessment of their infrastructure readiness.
See also: Apple, Google warned by Germany that DeepSeek is illegal
Integrating complementary technologies also presents a hurdle. Manufacturers express uncertainty about their capacity to effectively incorporate IoT data into generative AI models to enhance output accuracy.
Furthermore, the establishment of responsible AI frameworks remains a work in progress. While ethical considerations are on the agenda, only 48% of APAC manufacturing leaders strongly believe their organisations adhere to a robust framework that appropriately balances risk and value creation.
Workforce readiness is another key area of concern, with 53% of manufacturers acknowledging a skills gap among their employees in utilising generative AI effectively. This deficiency creates functional and operational disadvantages and introduces potential risks.
Finally, data management capabilities are proving to be a constraint. Less than half (46%) of Apac manufacturers strongly agree they possess sufficient data storage and processing power to support their generative AI workload demands, posing a limitation to successful implementation.
