The Fusion of AI and DePIN: The Rise of Decentralized GPU Networks

Since 2023, AI and DePIN have become hot trends in the Web3 space, with an AI market value of $30 billion and a DePIN market value of $23 billion. This article focuses on the intersection of the two and explores the development of related protocols.

In the AI technology stack, the DePIN network provides practicality for AI through computing resources. Large tech companies have caused a shortage of GPUs, making it difficult for other AI model developers to obtain sufficient GPUs. Traditional options like centralized cloud providers are inefficient and inflexible. DePIN offers a more flexible and cost-effective alternative, utilizing token incentives for resource contributions. DePIN in the AI field integrates individual GPU resources into a unified supply, providing developers with customized on-demand access while creating additional income for GPU owners.

Overview of AI DePIN Network

Render

Render is the first P2P GPU computing network, initially focused on graphics rendering, and later expanded to AI computing tasks.

Highlights:

• Founded by the Oscar-winning technology company OTOY
• Serving entertainment giants such as Paramount Pictures
• Collaborate with Stability AI and others to integrate AI models with 3D rendering
• Approve multiple computation clients and integrate more DePIN network GPUs.

Akash

Akash is positioned as a "super cloud" that supports storage, GPU, and CPU computing, serving as an alternative to traditional platforms like AWS.

Highlights:

• Supports a wide range of tasks from general computing to web hosting
• AkashML can run over 15,000 models on Hugging Face.
• Custody of important applications like Mistral AI's LLM model
• Supports metaverse, AI deployment, and federated learning platform

io.net

io.net provides distributed GPU cloud clusters, focusing on AI and ML use cases.

Highlights:

• IO-SDK is compatible with frameworks such as PyTorch and can be dynamically extended.
• Supports 3 types of clusters, starts within 2 minutes
• Collaborate with Render and integrate other DePIN network GPUs

Gensyn

Gensyn focuses on GPU computing for machine learning and deep learning.

Highlights:

• The cost of a V100 GPU is about $0.40 per hour, significantly saving costs.
• Support fine-tuning of pre-trained base models
• Provide decentralized globally shared foundational model

Aethir

Aethir focuses on enterprise GPUs in compute-intensive fields such as AI, ML, and cloud gaming.

Highlights:

• Expand to cloud phone services, launch decentralized cloud smart phones in collaboration with APHONE.
• Extensive collaboration with Web2 giants such as NVIDIA
• Multiple partners in the Web3 field such as CARV, Magic Eden, etc.

Phala Network

Phala Network serves as the execution layer for Web3 AI solutions, addressing privacy issues.

Highlights:

• Act as a verifiable compute co-processor protocol
• AI agency contracts can integrate with top large language models
• The future will include multiple proof systems such as zk-proofs.
• Plans to support H100 and other TEE GPUs to enhance computing power

Project Comparison

| Features | Render | Akash | io.net | Gensyn | Aethir | Phala | |-----|--------|-------|--------|--------|--------|-------| | Hardware | GPU&CPU | GPU&CPU | GPU&CPU | GPU | GPU | CPU | | Focus | Graphic Rendering and AI | Cloud Computing, Rendering and AI | AI | AI | AI, Cloud Gaming and Telecommunications | On-Chain AI Execution | | AI Task | Reasoning | Both | Both | Training | Training | Execution | | Pricing | Performance-based | Reverse Auction | Market Pricing | Market Pricing | Bidding System | Equity Calculation | | Blockchain | Solana | Cosmos | Solana | Gensyn | Arbitrum | Polkadot | | Data Privacy | Encryption & Hashing | mTLS Authentication | Data Encryption | Secure Mapping | Encryption | TEE | | Fees | 0.5-5%/task | 20% USDC, 4% AKT | 2% USDC, 0.25% reserve | Affordable | 20%/session | Proportional to staking | | Security | Render Proof | Proof of Stake | Proof of Computation | Proof of Stake | Render Capability Proof | Inherited from Relay Chain | | Completion Proof | - | - | Time-Lock Proof | Learning Proof | Rendering Work Proof | TEE Proof | | Quality Assurance | Dispute | - | - | Verifier and Reporter | Check Node | Remote Proof | | GPU Cluster | No | Yes | Yes | Yes | Yes | No |

Hardware Statistics

| Indicator | Render | Akash | io.net | Gensyn | Aethir | Phala | |-----|--------|-------|--------|--------|--------|-------| | Number of GPUs | 5600 | 384 | 38177 | - | 40000+ | - | | Number of CPUs | 114 | 14672 | 5433 | - | - | 30000+ | | H100/A100 Quantity | - | 157 | 2330 | - | 2000+ | - | | H100 Cost/Hour | - | $1.46 | $1.19 | - | - | - | | A100 Cost/Hour | - | $1.37 | $1.50 | $0.55( Estimated ) | $0.33( Estimated ) | - |

Conclusion

The AI DePIN field is still in its early stages, facing many challenges. However, the task volume and hardware quantity on decentralized GPU networks have significantly increased, indicating a rising demand for alternatives to Web2 cloud providers. These networks effectively address the issues on both the supply and demand sides, proving their product-market fit.

Looking ahead, AI is expected to develop into a trillion-dollar market. These decentralized GPU networks will play a key role in providing developers with cost-effective computing alternatives. By continuously bridging the gap between demand and supply, these networks will make significant contributions to the future landscape of AI and computing infrastructure.