Author: IOSG
TL;DR
TL;DR
As the combination of Web3 and AI has become a focus topic in the cryptocurrency world, the construction of AI infrastructure in the crypto world is thriving, but the actual use of AI or applications built for AI is not much, and the homogenization problem of AI infrastructure is gradually emerging. The recent participation in the first round of financing of RedPill has led to some deeper understanding.
The main toolkits for building AI Dapps include decentralized OpenAI access, GPU networks, inference networks, and proxy networks.
The reason why the GPU network is even more popular than the "Bitcoin mining era" is because: the AI market is larger and growing rapidly and steadily; AI supports millions of applications every day; AI requires diverse GPU models and server locations; the technology is more mature than before; and the customer base is also wider.
Inference networks and proxy networks have similar infrastructure, but with different focuses. Inference networks are mainly for experienced developers to deploy their own models, and running non-LLM models does not necessarily require GPUs. Proxy networks are more focused on LLM, where developers don't need to bring their own models, but are more focused on prompt engineering and how to connect different proxies. Proxy networks always require high-performance GPUs.
AI infrastructure projects promise great potential and are constantly launching new features.
Most native crypto projects are still in the testnet stage, with poor stability, complex configurations, limited functionality, and still need time to prove their security and privacy.
Assuming that AI Dapps become a major trend, there are still many undeveloped areas, such as monitoring, infrastructure related to RAG, Web3 native models, built-in crypto-native APIs and decentralized proxies for data, evaluation networks, etc.
Vertical integration is a significant trend. Infrastructure projects are trying to provide one-stop services to simplify the work of AI Dapp developers.
The future will be hybrid. Some inference will be done on the front-end, while some will be computed on-chain, considering factors such as cost and verifiability.
Source: IOSG
Introduction
The combination of Web3 and AI is one of the most closely watched topics in the crypto field. Talented developers are building AI infrastructure for the crypto world, aiming to bring intelligence into smart contracts. Building AI Dapps is an extremely complex task, with developers needing to handle areas such as data, models, computing power, operations, deployment, and integration with blockchains.
In response to these needs, Web3 founders have developed many preliminary solutions, such as GPU networks, community data labeling, community-trained models, verifiable AI inference and training, and proxy stores. However, in the thriving infrastructure background, the actual use of AI or applications built for AI is not much.
When developers look for tutorials on AI Dapp development, they find that there are not many tutorials related to native crypto AI infrastructure, and most tutorials only involve calling the OpenAI API on the front-end.
Source: IOSG Ventures
The current applications have not fully utilized the decentralization and verifiability of blockchains, but this situation will soon change. Now, most AI infrastructures focused on the crypto field have launched testnet networks and plan to go live in the next 6 months. This research will introduce the main tools available in the crypto AI infrastructure in detail. Let's get ready for the GPT-3.5 moment in the crypto world!
1. RedPill: Providing Decentralized Authorization for OpenAI
The RedPill we mentioned earlier is a good entry point. OpenAI has several world-class powerful models, such as GPT-4-vision, GPT-4-turbo, and GPT-4o, which are the preferred choices for building advanced AI Dapps. Developers can integrate them into Dapps by calling the OpenAI API through oracles or front-end interfaces.
RedPill aggregates the OpenAI APIs of different developers under one interface, providing fast, economical, and verifiable AI services to global users, thereby democratizing access to top-notch AI model resources. RedPill's routing algorithm will direct developer requests to a single contributor. API requests will be executed through its distribution network, bypassing any potential restrictions from OpenAI, solving some common problems faced by crypto developers, such as:
• TPM (Tokens Per Minute) limits: New accounts have limited token usage, unable to meet the needs of popular and AI-dependent Dapps.
• Access restrictions: Some models have restrictions on access for new accounts or certain countries.
By using the same request code but changing the host name, developers can access OpenAI models in a low-cost, highly scalable, and unlimited manner.
2. GPU Networks
In addition to using OpenAI's API, many developers also choose to self-host models at home. They can rely on decentralized GPU networks, such as io.net, Aethir, Akash, and other popular networks, to build GPU clusters and deploy and run various powerful in-house or open-source models.
These decentralized GPU networks can leverage the computing power of individuals or small data centers to provide flexible configurations, more server location options, and lower costs, allowing developers to easily experiment with AI-related tasks on a limited budget. However, due to their decentralized nature, such GPU networks still have certain limitations in functionality, availability, and data privacy.
In the past few months, the demand for GPUs has been booming, exceeding the previous Bitcoin mining craze. The reasons for this phenomenon include:
- Expanded customer base, as GPU networks now serve AI developers, a large and more loyal group that is less affected by cryptocurrency price fluctuations.
- Compared to mining-specific hardware, decentralized GPUs offer more model and specification options to better meet diverse requirements. Especially for large model processing, higher VRAM is needed, while smaller tasks have more suitable GPU options. Decentralized GPUs can also serve end-users with lower latency.
- The technology is becoming more mature, with GPU networks relying on high-speed blockchains like Solana for settlement, Docker virtualization, and Ray compute clusters.
- In terms of investment returns, the AI market is expanding, with more opportunities for new applications and model development. The expected return rate for the H100 model is 60-70%, while Bitcoin mining is more complex, with a winner-take-all dynamic and limited output.
- Bitcoin mining companies like Iris Energy, Core Scientific, and Bitdeer are also starting to support GPU networks, providing AI services and actively purchasing GPUs designed for AI, such as the H100.
Recommendation: For Web2 developers who don't prioritize SLA, io.net provides a simple and user-friendly experience, with a high cost-performance ratio.
This is the core of the crypto-native AI infrastructure. It will support billions of AI inference operations in the future. Many AI layer1 or layer2 projects provide developers with the ability to natively call AI inference on-chain. The market leaders include Ritual, Valence, and Fetch.ai.
These networks differ in the following aspects: performance (latency, computation time), supported models, verifiability, and price (on-chain consumption cost, inference cost) and development experience.
