Bitcoin is trading sideways and altcoins are basically out of business. Will mining rely on AI computing power next?

Foreword

In today's technology world, the most imaginative thing besides encryption is AI. AI and encryption are now like two dragons and phoenixes. If you can do well in any of them, you can reach the peak of your life.

Of course, there are also people who can do well in both industries, such as the sam altman we mentioned earlier. Through ChatGPT under OpenAI, it can be said that he has penetrated the AI ​​industry. In addition, he has issued an air coin WLD, which is now also the top token in the encryption track. Now it is also far behind Lafite.

1. What is the AI+Depin track

Is there a good project like sam altman that has two major tracks? Actually, there is, and it seems to be more logical than Ultraman. That is to use the web3 method to build AI infrastructure. Since the popularity of large models, the demand for AI infrastructure is actually very strong. AI infrastructure requires strong computing power support. McKinsey has calculated that as early as January 2023, OpenAI has consumed the equivalent of 175,000 Danish households' annual electricity consumption in just one month. Compared with Bitcoin mining, the power consumption of this AI infrastructure is huge. In addition to power consumption, the computing power chip Gpu is also very expensive. H100 and A100 are hundreds of thousands or millions of RMB. These are toys for enterprise players, and various centralized computing power rooms are built duangduang. With such a high threshold, if we encounter our web3, then retail investors must be able to invest! Bring down the threshold of AI technology facilities for me! The computing power of AI must be a decentralized computing power base, and the data storage of AI must be my decentralized storage! The bandwidth used by AI must be my distributed bandwidth.

So this year, many AI+Depin projects emerged.

We are very familiar with this AI, which is Artificial Intelligent in English, and translated into Chinese is artificial intelligence.

The English of Depin is Decentralized Physical Infrastructure Networks, which is translated into Chinese as distributed physical infrastructure. If this narrative is realized, the computers/mobile phones and other devices around you can become part of the distributed physical infrastructure and become working dogs to help you make money.

Because of the strong demand for AI, many Depin projects this year are for AI services.

DePIN core elements include three points:

• Decentralized: First of all, it is decentralized, which is the biggest difference from traditional cloud vendors. In the past, many AI computing power and storage resources were provided by cloud computing power services of large companies such as Amazon/Microsoft. Our encryption revolution is mainly to revolutionize these large companies. The most ideal result is of course to unite the devices of ordinary people like us and realize that a single spark can start a prairie fire. But it is also possible that the final situation is like Bitcoin mining, where the computing power and storage requirements of AI are released in an open and unified way, and anyone with computing power and storage resources can access it.

• Physical infrastructure: These services cannot be provided by manpower or air, but need to be provided by hardware of different types and sizes, such as various IoT, 5G information towers, various servers, or mobile phones.

• Token economy: The economy motivates the community to contribute and build a network together.

If the above three points are achieved, it is a Depin project. In fact, the Bitcoin mining we are familiar with is the most successful Depin

Look at it: the computing power is decentralized, there is also physical infrastructure such as mining machines, and Bitcoin is used as a token economy incentive.

2. Is this concept reliable?

Some viewers may ask, does the current demand for computing power for AI computing also require professional equipment, and does the civilian equipment around ordinary people, such as mobile phones and computers, not use it?

Other AI demands for computing power can be roughly divided into three categories:

One is for training large models, which has the highest requirements, such as the performance-crazy products provided by NVIDIA; the requirements are very high, and only NVIDIA can provide them now;

The second is after the large model is trained and deployed, it is combined with user questions to interact and run the model results, commonly known as inference, the computing power level required is not as high as training, and now NVIDIA, Huawei and 2-3 domestic startups can provide it;

The third is small vertical models deployed on the edge side, which need a lower level of computing power to run the results, about tens of T.

The first type requires professional equipment investment. Even if the depin project is successful, it is likely to be the same as the current Bitcoin mine, which is high-investment and has nothing to do with ordinary people.

For the second/third types of needs, the equipment around ordinary people can still help a little.

3. Inventory of AI+Depin Head Projects

Having said so much, I believe everyone has a certain understanding of the demand for AI and Depin, so let's take a look at what are the popular AI+Depin projects in this round of the industry?

• IO.NET

The first one to be introduced is io.net

io.net is a decentralized AI computing platform based on Solana developed by IO Research. In March this year, it announced that it had completed a $30 million Series A financing led by Hack VC, with a valuation of $1 billion!

In terms of valuation, it is currently the highest among the projects we introduced today.

So what exactly does this Io.Net do?

According to io.net's documentation, io.net can aggregate underutilized GPUs in independent data centers, cryptocurrency miners, and other hardware networks such as Filecoin and Render.

Who will use this computing power?

io.net's current target users are AI startups and developers, because these users are relatively poor and cannot afford the high cost of renting GPU/CPU computing power; and generally these companies will not go to the track of large model training, and the equipment requirements are not so high. Most of them only need to perform inference or edge vertical model calculations.

Who will provide this computing power? Gamers and crypto GPU miners.

Gamers have high-end hardware and fast Internet connections, but usually only have one GPU card; while crypto GPU miners have a large number of GPU resources, but may face limitations in Internet connection speed and storage space.

So if you want to participate in providing computing power for io.net, you still have to weigh the performance of your device. If it is not good enough, don't participate.

Does io.net seem to make sense? There must be some reason for its high valuation, right? (There have been a lot of melons recently)

• Aethir

Another heavyweight player is Aethir, which completed its Pre-A round of financing in mid-2023, with a total fundraising amount of more than US$9 million and a valuation of US$150 million. This round of financing was led by the well-known Sanctor Capital and Hashkey.

This project is a cloud computing platform that promises customers to completely change the current technological landscape by achieving scalability and inclusive changes in the sharing, ownership and use of GPUs. In the fields of artificial intelligence, cloud gaming and machine learning, Aethir allows enterprises and individuals to put idle GPUs into a shared network, which not only optimizes resource utilization overall, but also provides a wide range of user groups with access to top computing power. What does that mean? It means that it purchases GPU resources from users, and then sells the integrated computing power to game giants. Why? Because the core of the CPU is the ability to process graphics. Now high-end games require real-time rendering, and the demand for GPU computing power is still very large. According to public reports, Aethir has signed a contract with the world's largest game studio, with a player base of 150 million. Although it has not been revealed which studio this is, the only one that has reached this scale is probably Riot, which launched "League of Legends"?

In addition, it has official cooperation with WellLink, which is the largest cloud gaming company with more than 64 million monthly active users.

The upstream purchasers have signed well, how are the downstream suppliers doing?

We have seen official information that by March this year, more than 40,000 devices have been connected to the network. It seems that the project party is still working.

It is worth mentioning that on March 20 this year, Aethir launched the "Check Node" public sale, and sold 100 million US dollars in one day. It is said that there will be small GPU mining machines for sale soon, and the operation is more and more like the rhythm of Bitcoin mining machines.

But can Aerthir find so many users to participate in the network co-construction? This is still an unknown, but purely from the potential energy, it is still coming in full force. You can guess in the comment area, who will be listed on Binance first, Aethir or io.net?

• Grass

We mentioned before that whether it is io.net or aethir, the demand on the supply side is mainly computing power.

If some audience friends say, I don’t have computing power, but my network is good, can I just provide the network for mining?

There really is, that is the project grass we are going to introduce now.

Grass is the flagship product of Wynd Network, which allows users to profit from Internet connections by selling unused network resources.

According to the official statement, Grass currently allows users to provide unused Internet bandwidth through a web browser extension to earn points, which will be "retroactively converted into network ownership" at the end of the product beta phase.

In plain words, you can mine by surfing the Internet, and you can mine with your mobile phone.

This project completed a seed round of financing of 3.5 million US dollars at the end of 2023. Although the amount of financing is not high, the leading investment institution is the well-known polychain. We mentioned before that most of the infrastructure projects listed on Binance are invested by polychain.

Grass also had a pre-seed round in July last year, led by No Limit Holdings, which can be said to be an invisible boss in the crypto industry. Its founder is Gin Chao, the third person in Binance, the founder of Binance Labs, who was also on our beautiful talk a few days ago. Its portfolio has a very high chance of being listed on Binance. The stroom we introduced before is also an invested enterprise of this institution. A few days ago, Morgan's COO Jeremy Huff switched to the crypto industry and joined No Limit Holdings.

So Grass was born with a silver spoon in his mouth.

• EMC

If we still have viewers asking, io.net and aethir both require professional equipment, and grass mainly relies on network mining, then I live in the mountains, the network is not that good, and I only have an ordinary laptop at hand, can I mine?

Of course I can

There is also a very technical project called EMC, the full name is EdgeMatrix Computing, also called edge matrix computing. Doesn't the name tell you that the target users are the third type of computing needs we mentioned earlier: small vertical models deployed on the edge side, running results, require a lower level of computing power, about tens of TB.

This project was incubated by Wanxiang Blockchain Lab and HashKey Capital, and completed two rounds of financing, with an estimated financing scale of about 10 million US dollars. The token EMC has been listed on exchanges such as Matcha/Gate, and its current market value is about 1 billion US dollars.

The EMC protocol is an edge computing protocol based on blockchain and peer-to-peer networks, which can integrate computing power and applications from different sources, enabling them to exchange and collaborate on a unified platform.

Individuals or professional developers can run EMC nodes, use personal computers or large GPU clusters like miners to provide AI services and get rewards.

In other words, you can hold your ordinary laptop and go to Starbucks, a place that does not charge electricity bills, and the laptop can be connected to the EMC network to mine and make money. Doesn’t it feel that life is hopeful?

4. AI and encryption, hold together for warmth!

The AI ​​and encryption industries are still in the investment period. To put it bluntly, they are very optimistic in the long run, but it is still difficult to say about the short-term landing scenarios and realization.

But maybe AI and encryption, a pair of crouching dragons and phoenixes, can really be combined through depin.

I hope that one day your mobile phones and computers can support all the viewers through their own efforts, and it will not be in vain that you spend more time looking at your mobile phones and computers than looking at your partners every day.

Bye!