The rise of artificial intelligence (AI) and generative AI technologies has been meteoric in the past two years. For some tech-savvy people, every morning begins with the help of AI, from the smart alarm that tracks their sleep cycle to the news app that curates articles based on their interests.

But behind these seamless conveniences lies a hidden reality – these technologies are part of a growing energy crisis. As AI technologies like generative AI advance, they are not just transforming our lives; they’re demanding a huge share of the world’s electricity.

Impact of AI on Energy Infrastructures

The challenge is stark. As one of the most energy-intensive modern IT endeavors, AI systems require considerable carbon emissions and electricity. Indeed, the world might not be ready for their demands.

In 2023, the world became acquainted with the implications of generative AI, and by 2024, its utilization in various sectors magnified. Hence, data centers that power these AI models are becoming massive consumers of electricity.

Indeed, Forbes noted that GPT-4 required over 50 gigawatt-hours to train—equivalent to 0.02% of California’s annual electricity production. Moreover, it requires 50 times more energy than its predecessor, GPT-3.

The statistics are staggering. Globally, data centers and their transmission networks now contribute to 3% of global energy consumption, emitting as much carbon dioxide as Brazil.

Moreover, the escalating energy requirements show no signs of abating. According to an International Energy Agency (IEA) projection, global electricity demand will surge from 460 terawatt-hours (TWh) in 2022 to 1000 TWh by 2026.

Read more: How To Build Your Personal AI Chatbot Using the ChatGPT API

In the United States alone, the power demand from data centers is expected to increase from 200 TWh in 2022 to 260 TWh by 2026, marking a 6% share of the country’s total power usage. Projections suggest this demand will double by 2030.

Amid this backdrop, Ayush Ranjan, CEO of Huddle01, highlighted in an interview with BeInCrypto the urgent need for solutions like DePIN (Decentralized Physical Infrastructure Network).

“AI data centers require a substantial amount of electricity for computation and cooling. If AI applications continue to grow at the current rate, we will see a significant strain on both local and global energy grids that will prove unsustainable. This burden will continue to increase as AI systems get more and more complex with time. This will again lead to higher emissions and grid instability,” Ranjan explained.

The geographic clustering of data centers compounds the challenges. For instance, Northern Virginia hosts the largest hub of data centers globally, consuming electricity equivalent to that of 800,000 homes. This concentration creates dangerous fluctuations in power demand, posing severe risks to energy infrastructures.

How DePIN Solves the Challenges

In response, DePIN offers a promising solution by leveraging underutilized hardware resources to distribute computational tasks more efficiently. By decentralizing energy consumption and incentivizing the use of edge computing, DePIN networks could significantly alleviate the energy burden imposed by AI, offering a pathway to more sustainable and democratized access to AI resources.

Ranjan further elucidated that DePINs distribute energy consumption and workload, easing the burden on any single point. Instead of relying on huge centralized data centers, DePIN deploys multiple nodes—often utilizing underused infrastructure to offload computations closer to end-users.

“This reduces the workload on servers and spreads energy consumption more evenly across regions, easing the burden on energy grids,” Ranjan told BeInCrypto.

Currently, 84% of the data centers are concentrated around the United States, Europe, and China, making data transfers less energy efficient. However, edge computing, integral to DePIN, minimizes long-distance, energy-intensive data transfers typical of centralized data centers.

“Splitting the energy consumption across multiple devices and regions, reducing the load on data centers and energy grids by leveraging existing devices or resources to build the network will prove critical in solving this issue,” Ranjan affirmed.

Read more: What Is DePIN (Decentralized Physical Infrastructure Networks)?

DePin Projects Addressing AI’s Demands

According to Ranjan, several DePIN projects, like Filecoin Green, Akash Network, Render, and Grass, focus on addressing AI’s energy demands.

Notably, the Daylight Energy project, backed by prominent venture capitalist firm Andreessen Horowitz (a16z), aims to transform energy grid operations through distributed energy resources (DERs). This initiative enhances grid responsiveness and facilitates sustainable energy practices by leveraging real-time data from DERs such as solar panels and smart batteries.

Moreover, on September 10, Daylight Energy announced a partnership with DIMO Network to enable electric vehicles (EVs) to support power grids. This collaboration utilizes DIMO’s EV application programming interfaces (APIs) to integrate EVs into the energy management ecosystem, thereby facilitating clean energy usage and real-time energy management for all EV owners. 

DePIN networks also solve other challenges of centralized infrastructure, such as frequent outages. For instance, a recent IT outage involving Microsoft and CrowdStrike disrupted major services worldwide. However, DePIN networks are less susceptible to such outages because they do not have a single point of failure.

Currently, the total market capitalization of DePIN projects stands above $20.5 billion. Additionally, the total number of DePIN devices has crossed 18 million. However, DePIN still faces scalability challenges as the mainstream adoption of these networks requires high computational power.

“Many DePINs rely on a mix of devices, from low-powered edge devices to small-scale data centers. Scaling the network and coordinating the deployed resources to match the computational power of a centralized data center remains a formidable industry challenge,” Ranjan noted.

Read more: Top 10 Web3 Projects That Are Revolutionizing the Industry

However, while the idea of DePIN rescuing the world from a global energy crisis remains nascent, further innovation and adoption are essential. Ranjan believes that token incentives can help bring more adoption.

“Because of hardware limitations of edge devices to handle AI workload, wide adoption is crucial for any DePIN to scale and see a mainstream use case. Token incentives help drive intent to use and participate,” Ranjan concluded.