When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
The data center industry is struggling to keep up with the massive amounts of computing power that artificial intelligence (AI) requires to train and run large language models (LLMs).
Advanced AI models require thousands of graphics processing units (GPUs) in the pre-training phase and require extremely advanced data center architecture.
Large cloud service providers currently run hyperscale data centers of around 50-200MW costing $1-4 billion but will need to scale up to 1GW+ capacities costing around $10-25 billion by the end of the decade, according to estimates from Bain & Company consultancy.
There are few companies that will have the capacity to build such large facilities, limiting how many can be built, and the space is currently dominated by tech giants like Alphabet, Microsoft, Amazon Web Services (AWS), and Alibaba — making it difficult for smaller competitors to build capacity at a similar scale.
Is there a solution to level the playing field? This is where decentralized physical infrastructure networks (DePINs) come in.
Key Takeaways
- DePIN aggregates small-scale computing resources to solve AI’s growing demand.
- This decentralized computing democratizes AI infrastructure, reducing the dependency on tech giants.
- Crypto tokenization incentivizes participation, enhancing innovation and resource access.
- Decentralized networks can help to lower costs, improve security, and reduce latency.
- We expect DePIN to change how industries scale up as AI adoption increases across 2025.
What Role Can DePIN Play in AI’s Development?
“The scramble to acquire AI resources is already creating extreme competition for resources at the high end of the market, and growing data center requirements will further strain capabilities,” according to Bain & Company.
With demand outpacing cloud capacity, edge AI is emerging as a fix to maximize resources. AI algorithms and models are deployed directly on devices, including sensors or Internet of Things (IoT) devices — at the edge of the network — to enable real-time data processing while reducing the load on cloud networks.
“These networks, which bridge the physical and digital worlds, are no longer just a theoretical framework — they are becoming a cornerstone in the evolution of artificial intelligence development and deployment,” Rock Zhang, chief executive officer (CEO) of Network3.ai, told Techopedia.
But a missing part of the puzzle is DePIN, a Web3 concept that extends beyond digital assets and uses blockchain technology, tokenization, smart contracts and decentralized governance mechanisms.
DePIN incentivizes participants to contribute or build physical resources, including data storage, energy generation and grids, sensors, wireless communication devices, and computing power.
As blockchain networks advance, it becomes easier for decentralized networks to integrate physical assets.
So, DePIN can aggregate small-scale infrastructure but decentralize control, enabling assets to be managed, operated, and governed by distributed networks of individual participants.
This approach to physical infrastructure management improves transparency and security, which reflects the Web3 ideal of decentralized control and community ownership.
As Zhang of Network3 told Techopedia: “By enabling decentralized networks of IoT devices, compute nodes, and sensors, DePIN reduces dependency on centralized cloud infrastructure, creating new avenues for real-time, cost-efficient, and localized AI training and inference.”
Tokenizing assets in DePIN then supports new business models and revenue streams for infrastructure development that are more efficient, cost-effective, and secure than some traditional models. And efficiency is crucial for running AI models with low latency.
Physical Location Can Reduce Latency
DePIN can then reduce the shortage in computing resources for AI by creating networks of individuals with idle GPUs and edge devices and pooling their computing power to rent on-demand out for AI training.
Users can then access the computer power they need geographically located closest to them, reducing latency and without the overhead costs of the big providers — ideally, higher capacity at much lower fees.
This provides AI startups with an affordable alternative to the big cloud computing providers and creates room for more new players to innovate, as the tech giants have less control over computing resources.
In this way, DePIN can democratize access to GPUs beyond big tech, research organizations, and governments and potentially contribute to the acceleration of AI deployments.
At a Glance: How DePINs Work for AI
Resource provision: DePIN participants worldwide connect spare computing capacity to the network.
Smart contracts: Smart contracts monitor resource demand and availability and automate transactions based on predetermined rules.
Resource management: Market-driven allocation of computing power to optimize the use of resources.
Tokenization: Participants receive tokens as rewards for contributing computing power and can contribute to governance; AI companies can quickly and easily pay with tokens for access to computing power.
Outcomes: Cost savings, efficient use of resources, lower latency, transparency.
According to DePIN Scan, there are currently 73 DePIN projects focused on the AI sector.
DePIN projects are deploying on high-speed blockchains with low processing fees, such as Solana and Cosmos, as well as running on Layer 2 networks on the Ethereum (ETH) blockchain.
Networks like Aethir and Render are aggregating GPU power and are increasingly focused on delivering compute capacity to the AI industry.
Generative AI platform TensorOpera demonstrated the effectiveness of DePIN for AI, as it was able to train a 750-million parameter model for 30 consecutive days.
Major DePIN Trends in the AI Space
Network3.ai’s Zhang anticipates three major trends in DePIN and their role in AI development:
1. Massive Scaling of Decentralized Compute
“We are already seeing the rise of decentralized compute networks, where edge devices, GPUs, and idle hardware contribute to AI workloads.
“In 2025, I predict this will lead to a shift in how AI models are trained — transitioning from centralized supercomputing clusters to distributed networks powered by DePINs.”
2. Tokenized Incentives for AI Contributions
“Token-based systems within DePINs will incentivize individuals and businesses to contribute data, compute power, and infrastructure.
“This tokenization will democratize access to AI development and allow smaller players to monetize their resources while participating in global innovation.”
3. Enhanced AI Security and Privacy
“The decentralized nature of DePINs inherently enhances security and privacy. By distributing data and computations across networks, DePINs minimize vulnerabilities and ensure compliance with stricter global data privacy regulations. This is particularly critical as AI applications become more pervasive and sensitive.”
The DePIN Outlook for 2025
“As we approach 2025, the possibilities of DePINs are coming into sharp focus,” according to Zhang.
“DePINs will act as the backbone for a decentralized AI economy, enabling more equitable growth and reducing barriers to entry for underrepresented regions and industries.
“As the technology matures, we can expect DePINs to enable groundbreaking applications in healthcare, agriculture, and energy management, while also providing the infrastructure necessary for emerging innovations like generative AI at scale.”
Zhang sees 2025 as a pivotal year when DePIN and AI move from the experimentation stage to wide-scale adoption, “fundamentally transforming industries and reshaping how we build and interact with technology”.
The Challenge Ahead
The concept of DePIN shows promise, but widespread adoption by a large number of participants prepared to not only contribute but maintain physical infrastructure will be key to these networks reaching their full potential.
Understanding the benefits of participating and high initial setup costs — particularly for GPUs — are hurdles. Education will be crucial to demonstrating the benefits and potential impact of decentralizing physical networks.
Early adopters who can work through these challenges are set to gain a competitive advantage as the market develops.
The Bottom Line
By aggregating computing resources from individuals and organizations around the world, DePIN can help to solve one of the dilemmas of large-scale AI adoption — the need for vast amounts of processing capacity.
DePIN offers the potential to democratize access to AI infrastructure, ensuring that it does not remain solely under the control of the tech giants but allows startups and researchers to participate in development and innovation.
Heading into 2025, DePIN can facilitate the scaling of distributed networks for AI model training, token-based incentives for contributing to AI development, and enhanced security and privacy.
