Mike D’Aurizio and Christophe Defert

Globally, manufacturing and industrial processes account for almost a fifth of global carbon emissions, and consume about half of the world’s energy [1]. So decarbonising manufacturing is an essential part of the transition to a more sustainable economy.

A key element of this is reducing the inefficiencies that persist in so many manufacturing processes around the world.  Issues like overuse of resources, system down-time and excess waste are costly for manufacturers and damaging to the planet. But they are not inevitable. They are just symptoms of systems and processes that have historically operated without adequate data and feedback loops – and if operators can’t measure it, they can’t optimise it.

So decarbonising manufacturing does not have to mean regulating industries out of existence. It means making manufacturing more intelligent: improving processes and resource efficiency in ways that are better for the planet and the bottom line.

Thanks to the rapid advances and cost declines in sensors, optics, robotics, and edge computing, we now have an opportunity to do that in ways that simply were not feasible a decade ago.

Manufacturing in the intelligence era

The global market for AI-driven industrial equipment and systems efficiency could be worth $300bn by 2028, according to BCG [2]). And we’re already starting to see evidence of this on the factory floor, thanks largely to the rise of physical AI – robots and machines that can dynamically interact with and learn from their environments.

AI implementations are delivering energy savings of 10-20% at manufacturing facilities, while companies deploying AI to predict faults are reporting reductions in unplanned downtime of up to 50% [3]. If scaled across the global manufacturing sector, gains of this order potentially represent hundreds of billions of dollars in avoided costs – and a substantial reduction in carbon emissions.

Predictive maintenance is arguably the most mature application in intelligent manufacturing. By analysing sensor data from production equipment, AI platforms can identify failure modes weeks before a breakdown occurs, allowing operators to schedule maintenance during planned downtime – rather than scrambling to recover from unplanned stoppages. For capital-intensive industries (such as chemicals, metals, semiconductors etc.), the economic benefits of this are compelling, even before the environmental benefits are considered.

Servo loop control in industrial robotics is another example of how physical AI can improve manufacturing outcomes.

Legacy industrial robots (e.g. as seen in goods manufacturing) are ‘blind’ to their environments, and follow pre-programmed servo loops (instructions that dictate how the motors and robotic arm move through physical space). These robots are generally poor at responding to changing conditions or positions of objects moving along a factory line, causing breakdowns, delays, and waste.

Computer vision systems, leveraging edge computers, can connect directly with different robotic platforms and steer them to increase precision, reduce costly breakdowns, and improve manufacturing efficiency.

In the longer term, the more transformative opportunity of the intelligence era lies not just in optimising existing industrial process – but in reinventing them completely.

Take data centres for example, which are generating increasing amounts of electronic waste as data servers reach end-of-life faster. There are now robotics companies tackling the difficult problem of how to disassemble these millions of devices globally, to recover their valuable parts and materials for reuse in new data servers.

Where intelligence meets impact

For impact-driven investors, this investment theme is particularly compelling because there’s such a strong alignment between commercial returns and measurable environmental impact. The companies best positioned to win are those that help manufacturers reduce waste, use fewer inputs, and extend the life of materials – which also helps to reduce emissions, cut pollution, and use fewer natural resources.

Two external forces are accelerating this alignment. The first is regulatory pressure: the EU’s Corporate Sustainability Reporting Directive, evolving supply chain due diligence legislation, and growing pressure from institutional investors are forcing manufacturers to measure and report their environmental footprint in ways they never have before. This creates clear incentives for manufacturers to optimise and improve their processes – and again, the intelligence era is giving them the tools they need to do that.

The second tailwind is geopolitical. Supply chain disruption and the accelerating fragmentation of global trade is driving a wave of industrial reshoring and nearshoring across North America and Europe. These new factories are being designed and built on digital infrastructure from the outset, so they are far more receptive to AI-enabled optimisation than the legacy plants they are replacing.

Companies that help these facilities run more intelligently – and can prove it quantitatively, with real operational data – clearly have a strong commercial opportunity. And in a sector responsible for nearly a quarter of global emissions, making manufacturing smarter is not a niche sustainability play. It is one of the most important levers available to accelerate the transition to a genuinely sustainable economy.

This is the fourth in our series on the intelligence era of climate technology. Next: why intelligence is critical to Europe’s energy security and competitiveness.

[1]: https://www.iea.org/data-and-statistics/data-product/greenhouse-gas-emissions-from-energy

[2]: BCG, 2026: https://www.bcg.com/publications/2026/the-capital-opportunity-in-ai-enabled-sustainability

[3]: McKinsey Global Institute, The Next Normal in Manufacturing, 2024; Deloitte, 2024 Manufacturing Industry Outlook