Manufacturers around the world are placing bold bets on artificial intelligence, directing nearly half of their modernisation budgets toward systems they believe will lift profits within the next two years. The shift signals a clear message: AI is no longer a side project. It is becoming the main engine for growth.
According to the Future-Ready Manufacturing Study 2025 from Tata Consultancy Services (TCS) and AWS, 88 percent of manufacturers expect AI to deliver at least a five percent lift in operating margin. One in four believe it will push margins by more than ten percent. The optimism is strong. The infrastructure supporting these ambitions is not.
A race to extract value from AI
The study shows that 75 percent of executives expect AI to rank among their top three contributors to operating margins by 2026. As a result, companies are dedicating 51 percent of their transformation budgets to AI and autonomous systems. This investment eclipses other priorities. Workforce reskilling stands at 19 percent and cloud modernisation at 16 percent, a gap that highlights an uncomfortable tension: sophisticated AI is being deployed on outdated foundations.
Anupam Singhal, President of Manufacturing at TCS, said the industry thrives on precision and reliability. He noted that AI can strengthen these traits by improving predictability, stability, and control. Singhal added that TCS sees an opportunity to help manufacturers build systems that can adapt and succeed in an era shaped by intelligent autonomy.
Physical safeguards still rule
Despite their enthusiasm for predictive technologies, many manufacturers continue to rely on old habits when disruptions hit. Sixty-one percent have increased safety stock. Half have turned to multisourcing. Only 26 percent have used digital twins to guide their responses.
The gap is clear. AI promises smarter inventory optimisation, yet leaders are stockpiling materials as if the digital tools do not exist. As Ozgur Tohumcu of AWS explained, the industry is under intense pressure due to tight margins, unstable supply chains, and workforce shortages. He said AWS is pushing AI systems that learn, adapt, and act with minimal human intervention. True resilience, he noted, comes when technology can predict and respond on its own.
The data problem
The biggest barrier to AI’s promised returns is not the technology itself. It is the data feeding it. Only 21 percent of manufacturers consider themselves fully AI-ready. Most deal with uneven data quality across sites, creating fragmented systems that restrict enterprise-level insights.
More than half of respondents cited integration with legacy equipment as a major challenge. Years of accumulated technical debt make it difficult to layer modern autonomous tools on old operational systems. Security concerns add to the hesitation. With 52 percent listing cyber risks as a key obstacle, trust in autonomous intervention remains limited.
Ryan Daws
The rise of agentic AI
Even with these hurdles, the industry is moving quickly toward AI agents that can make decisions with minimal oversight. Seventy-four percent of manufacturers believe AI agents will manage up to half of routine production decisions by 2028. Two-thirds already allow or plan to allow AI to approve routine work orders within a year.
The workforce impact is evolving. Most companies expect AI-guided robotics to reshape roles, but the emphasis is on support rather than replacement. Productivity gains are strongest in knowledge-heavy functions like quality inspection and IT support. Traditional hands-on roles are seeing slower adoption.
Companies are also avoiding dependence on any single platform. Sixty-three percent prefer hybrid or multi-platform approaches, keeping their options open as AI capabilities mature.
Ryan Daws
Turning investment into real profit
For manufacturers to convert their AI spending into measurable returns, leaders need to confront three foundational issues.
First, data quality must improve. With such a small share of companies fully AI-ready, modernising data infrastructure should outrank flashy algorithm development.
Second, trust in AI must grow. The tendency to fall back on safety stock shows hesitation to rely on digital signals. A phased approach can help, starting with lower-risk administrative tasks before handing over complex operational decisions.
Third, flexibility matters. A multi-platform strategy can prevent companies from locking themselves into systems that may not evolve fast enough.
The manufacturing sector is betting heavily on AI. The path to profit depends not just on advanced models, but on strengthening the fundamentals that allow those models to succeed.
