PYMNTS Intelligence Banner June 2024

Photo App Photoroom Taps Green Energy for AI Training

Photoroom, AI, sustainability

Photo editing startup Photoroom plans to power its artificial intelligence (AI) training using 100% renewable energy as tech companies grapple with AI’s growing environmental impact.

The Paris-based company, which claims to have amassed over 150 million downloads since its 2019 launch, is betting that eco-friendly computing will be crucial to staying competitive in the rapidly evolving world of AI-powered mobile apps.

“AI is contributing to a huge increase in demand for energy in the U.S.,” Eliot Andres, Photoroom’s co-founder and CTO, told PYMNTS. “The quantity of GPUs needed to train AI models uses a lot of power, driving up energy demand. In order to service that demand, energy providers are turning to carbon-based solutions.”

AI’s Growing Energy Appetite

Photoroom’s decision to partner with Genesis Cloud comes as the tech industry tries to meet the energy demands of AI. A study by researchers at the University of Massachusetts Amherst found that training a single large AI model can emit as much carbon as five cars in their lifetimes. The computational power required to train and run sophisticated AI models has skyrocketed in recent years, leading to concerns about the technology’s carbon footprint.

This trend is part of a larger pattern in the tech industry. The popularity of massive AI models like ChatGPT highlights a growing trend toward large-scale AI, fueling predictions that data centers could consume as much as 21% of the global electricity supply by 2030.

The partnership gives Photoroom access to a high-performance compute cluster featuring Nvidia HGX H100 GPUs and high-speed networking infrastructure, all powered by renewable energy sources.

“Genesis Cloud’s data centers run on 100% renewable energy. The data center that powers Photoroom’s cluster, for example, is directly connected to a hydroelectric power plant,” Andres said.

AI photo editing automates tasks like background removal and image enhancement, making professional-looking results easy for anyone. Experts say such software helps businesses create better product photos and improve online stores, boosting sales and customer interest.

Balancing Innovation and Sustainability

While environmental concerns were a critical factor in Photoroom’s decision, the move also addresses practical business challenges, Andres said. In the competitive world of mobile apps, computational power can make or break a company.

“On Nvidia, we’re using a technology called Infiniband, which allows GPUs to communicate at very fast speeds — like having an extremely efficient postal service,” he added. “This speed means we can train our models faster, lets our machine learning team launch more experiments and allows us to have the very best tools to train.”

The partnership also helps Photoroom navigate the issue of resource scarcity. The global chip shortage, which began in 2020 and has continued to impact various industries, has made it challenging for many companies to secure the hardware needed for AI development.

“Like many AI companies, the scarcity of GPU resources has been an issue for Photoroom,” Andres said. “Committing to Genesis’ cluster 24/7 means we no longer have these GPU availability issues.”

Looking ahead, both companies are committed to advancing sustainable computing in AI. This aligns with broader industry trends: a report found that 63% of executives believe AI will help solve the climate change crisis, highlighting the complex relationship between AI and environmental sustainability.

As AI becomes more ubiquitous in consumer applications, from photo editing to voice assistants, the energy required to power these systems will likely come under increased scrutiny. According to a report by the International Energy Agency, data centers, which are crucial for AI operations, account for about 1% of global electricity use. This figure is expected to grow significantly as AI adoption increases.

Photoroom’s partnership with Genesis Cloud may serve as a test case for whether green computing can keep pace with the demands of cutting-edge AI development.

“We believe that innovation doesn’t have to come at the cost of sustainability,” Andres said. “As an industry leader, it’s our duty to ensure the decisions we’re making lead to the right choice in the environment and model the behavior we want to see in our sector.”