Dassault Systèmes: How AI Can Cut its Own Energy Use with ML

By 2030, AI-related data centre energy use could account for as much as 3% of global electricity demand, with certain regions facing far sharper rises.

The International Organization for Standardization has introduced ISO/IEC 42005 to support companies in evaluating AI’s wider societal impact, including environmental costs.

In Ireland, a key hub for global digital services, AI-related data centre energy consumption could climb to 35% of the nation’s total power use.

Yet Philippine de T’Serclaes, Chief Sustainability Officer (CSO) at Dassault Systèmes, remains confident in AI’s long-term potential to deliver environmental benefits.

The potential of machine learning for data centre sustainability 

Recent research indicates ML algorithms can enhance grid efficiency by 15% and improve battery storage performance by between 10% and 20%.

AI also has the potential to cut timelines for new clean energy projects by around 20%, representing savings of hundreds of billions of dollars by 2050.

According to McKinsey, AI and ML could accelerate 47% of the actions required to meet the global 1.5°C pathway outlined in the Paris Agreement.

The technology is already proving a driver of sustainable innovation.

AI has, for example, enabled the creation of lightweight packaging that lowers transport energy use and developed advanced materials for more efficient batteries.

In addition, studies show that AI-designed paint coatings could help reduce building temperatures by as much as 20°C.

However, current deployment patterns raise concerns about energy sourcing. 

Dassault Systèmes’ partnerships for efficiency gains

MIT’s Technology Review recently cautions that the pace of application growth means “data centres are expected to continue trending towards using dirtier, more carbon intensive forms of energy”.

Addressing this challenge, Dassault Systèmeshas partnered with Taiwanese server manufacturer Quanta Cloud Technology (QCT) to drive efficiency gains, while Chief Sustainability Officer Philippine de T’Serclaes promotes “Frugal AI” approaches that prioritise lightweight models and measurable outcomes.

One such method, model pruning, enables developers to strip out unnecessary neural network connections, reducing computational demands while preserving accuracy.

This significantly lowers energy use without compromising performance.

At the same time, optimising data centres offers a direct path to improved efficiency. Cooling systems alone account for up to 40% of overall energy consumption, and selecting advanced solutions can raise efficiency levels by 30%.

At Dassault Systèmes, virtual twin technology on the 3DEXPERIENCE platform is being used to support sustainable operations through deep collaboration with partners.

Scenario modelling performs trade-off analyses across technology configurations, empowering teams to optimise systems and future-proof infrastructure.

Through its collaboration with QCT, Dassault Systèmes also simulates data centre heat and airflow dynamics to design more effective cooling strategies.

The company’s solutions also supports Bouygues Construction, the French building company, in modular construction processes, while also helping Olivier Naar design modular nuclear reactors.

“I don’t see isolated projects,” Philippine says, “I see interconnected, mutually enhancing nodes within a wider value network. 

“I see how those same techniques can help build modular data centers. I see how we can power them with electricity that is low-carbon and convenient.”

The importance of broader thinking 

Network effects are also visible through initiatives such as the Coalition for Sustainable AI, in which Dassault Systèmes takes part.

However, Philippine stresses that technical solutions on their own cannot resolve sustainability challenges.

The true value of processing power, she notes, lies in how it is applied, regardless of gains achieved through data centre efficiency.

“AI will be what we make of it,” she says.

“Circular thinking needs to be embedded not just in our processes and products, but also in our methods and in the way we think about the world.

“We have what we need to make the infrastructure and ecosystems that power AI start working more effectively today.

“What we make of it? Well, that’s up to us.”