Exclusive Interview with Arweave Founder Sam Williams: What Does the Arrival of AO Mean?

Author: OdailyNews

With the development and application of decentralized technology, Arweave's latest AO platform has attracted widespread attention.

Recently, Odaily Planet Daily had the honor of interviewing Arweave's founder, Sam Williams, to delve into how Arweave is redefining data storage and access through its innovative decentralized storage model and the upcoming AO ecosystem.

Sam also shared insights on what the Arweave community and developers should look forward to in 2024, as well as his grand vision for the future development of Arweave. Let's explore how Sam articulates the uniqueness of Arweave and how it challenges the current digital storage paradigm, bringing innovative changes to the world.

Sam in the video announcing AO, source: Twitter

Odaily: First of all, thank you very much for accepting the interview, Sam. The launch of AO on the Arweave permanent storage network, this decentralized parallel computing layer, what does it mean for Arweave's positioning and the development of its ecosystem?

Sam: Arweave is a permanent information storage layer that can support on-chain storage of any scale. When we started this project, we primarily focused on archival use cases for long-term data preservation. However, we gradually realized that the permanence of information enables many other functionalities. In fact, it is precisely the permanence of information that allows us to build highly trustworthy web applications on top of this system. These web applications have the same properties as smart contracts— for example, once developers release the user interface of a web application, it cannot be changed. This was the first point we noticed. Subsequently, we found that permanent storage of computation logs and the use of deterministic virtual machines can enable decentralized computing and verifiable smart contract-like computation.

By introducing AO, we have pushed this concept to its logical extreme, that is, a super parallel computer that can store computation logs for any number of processes simultaneously, relying on Arweave's ability to support on-chain storage of any scale, and facilitating communication between these processes through messaging. This opens up numerous possibilities and allows Arweave to provide a highly scalable smart contract ecosystem that far exceeds any other system currently on the market. We know this is a bold statement, but we absolutely have the capability to prove it.

Moreover, it actually realizes a long-discussed vision in the computer science community: having a distributed or decentralized operating system that is trustworthy, where you can add additional computing units for more utility. Therefore, what we are really seeing is that, in the long run, we have turned Arweave into a permissionless, globally accessible supercomputer that anyone on Earth can access at the lowest feasible cost, which is truly remarkable. So this means a lot for Arweave's positioning, and I think this point is now widely recognized.

This makes Arweave an L1, if you will, with its own native financial ecosystem, but that is just the beginning. It also makes Arweave a native L1 for AI agents, where these autonomous processes run within the system, utilizing any amount of computing resources, and then interact with various entities.

But first, there is finance. We believe this essentially enables a financial ecosystem where both intelligence and settlement occur on-chain. In traditional DeFi, the blockchain is essentially just a settlement layer, right? They allow you to exchange or lend, borrow, atomic operations, like exchanging X tokens for Y tokens, and that settlement happens on the network. But the intelligence about why you should or shouldn't perform these operations is off-chain. When we observe traditional financial economies, yes, 70% of exchange interactions initially come from bots, not humans.

Thus, one obvious thing that AO enables you to do is to bring all the intelligence that actually drives the market into the network itself and make it trustworthy. A simple way to imagine this is to think of a fund that wants to invest in a specific category, but you can make that strategy itself autonomous and trustworthy, executing intelligent computations and acting based on it without having to trust any fund manager. This is a fundamental breakthrough for DeFi that was previously impossible. Then in the long run, we believe we have actually created such a democratized supercomputer. Therefore, any developer on Earth can build a program and run it on a machine with hundreds of thousands of cores if they wish. And they only need to pay for the portion they actually use. So, in the long run, it even has utility far beyond the current problems that cryptocurrencies are trying to solve.

Odaily: Thank you. AO and Ethereum's sharding technology are both Layer 1 scaling solutions aimed at enhancing the scalability and throughput of their respective networks. What are the differences between these two solutions?

Sam: The core difference is that AO is built from a role-oriented perspective. First, we built Arweave, which is an arbitrarily scalable on-chain storage layer. When we say arbitrarily scalable, we mean it seriously. On the mainnet, using real user data, it has already achieved a processing speed of 450 to 500 transactions per second. There was indeed a moment when the speed reached 1200 transactions per second, but that is not important. We have also tested with synthetic data at 50,000 transactions per second. The key point is that it can simply scale. Therefore, we have taken this as the foundation of the system. We did not attempt to create a globally synchronized data space or memory space, which is what almost every smart contract system is trying to do. This is also the core goal of Solana.

We create local memory for each process and then coordinate through a messaging bus. This is the fundamental way traditional distributed system architectures have always scaled. You never try to create a larger single computer; instead, you try to leverage multiple computers simultaneously. We have actually taken this approach and applied it to the blockchain ecosystem. By starting from this layer, it provides us with a foundation to permanently store any amount of data, essentially giving you a correct underlying structure to build upon for free.

So, we did not take the vertical scaling approach that Solana and similar projects are trying to achieve, but instead chose horizontal scaling. You can add more computers to the system to increase capacity, which is a completely different approach. While there is much more to it, I think this is a good overview.

Odaily: From our perspective, AO seems to emphasize Arweave's computational characteristics as an L1 in addition to storage. Has AO announced any updates in Denver? What does the future roadmap look like?

Sam: I believe AO provides a complete stack for decentralized networks. In the past, you could only use your hard drive space to mine Arweave, but now you can also use your CPU or the bandwidth of the information you pass. It actually covers the entire stack of computation, providing availability in this arbitrarily scalable manner, realizing what we call the permanent network. The mid-term roadmap focuses on enabling payments on top of the system and also creating a proof-of-stake network for it. Currently, the testnet operates in a so-called authority proof mode, so when developers initially set up their processes, they can choose which keys they trust to send messages to those processes. This will be the implementation of a proof-of-stake system in the mainnet, through which you can stake tokens on the messaging within the network. This is the core innovation we are focusing on next. We have grand plans for this, and we just need to implement and put those plans into practice.

Odaily: In December 2023, the fork of Arweave by Irys (formerly Bundlr Network) and the reset of token supply raised some concerns among the community and AR holders. Looking back at this event, what impact did it actually have on Arweave?

Sam: In fact, they did not fork the network. This essentially means that people just turned to a different bundle. I also think this decentralized the entry points of the network in a good way. I also believe the community has learned how to handle various forks. Arweave is a permanent information storage system, which is different from ordinary blockchains. If you reset the state every five years, then it cannot execute permanent storage of data.

Fortunately, there is a system in the network that actually incentivizes people to create forks in what you might call the right way and to do so in a "pro-social" manner. This essentially means they can take the Arweave network, create a fork from a specific block height, change the protocol rules in their own version, and also add new tokens to reward themselves. This is very important. It means that the dataset in that new forked system will survive and continue. At the same time, the tokens of holders in the new system will also continue. If that new version is better, people will eventually turn to it and use it. Therefore, token holders will feel satisfied and fairly rewarded, while innovators will be pleased to innovate on the system and receive new rewards for it. The continuity of datasets is the future.

We have elaborated on this extensively. We call them evolutionary forks because it creates a tree of protocols and networks that evolve around changes in the environment. This is one of the fundamental ways we believe Arweave will circumvent issues over the project’s time frame of hundreds of years in the future.

Ultimately, Irys publicly stated that if they fork Arweave, they would follow this framework, this evolutionary fork guiding system, which is a very good thing. The community thinks it is important to understand this, and we need to socially enforce it among ourselves to ensure that this permanent information storage project can start smoothly and continue in the way we hope. Overall, I think this is an inspiring and significant moment for the community.

Odaily: As a representative of decentralized storage, what is the biggest difference between Arweave and other solutions?

Sam: Arweave focuses on permanent information storage rather than temporary information storage. Therefore, we achieve this by creating a storage endowment. Essentially, the way this mechanism works is that you pay for 200 years of storage in advance, which may sound expensive, but it actually costs about half a cent per megabyte. As storage costs decrease over time, the purchasing power of the initially paid storage increases. Yes, therefore essentially, this allows you to indefinitely cover future storage costs. The system is designed to initiate as long a time cycle as possible.

This is an innovation in business models that would not be possible without blockchain, and we think this is very important. Having a temporary storage system, a temporary decentralized storage may be useful in certain areas. But fundamentally, it is just trying to put an existing business model into a network. Blockchain actually makes building more difficult; you have to leverage its inherent properties to achieve the best results or create a meaningful network.

Frankly, Bitcoin achieves this by saying we can have a monetary policy that is static and protocolized. You can trust the monetary policy because no one can change it arbitrarily, right? No private company can provide such a service. Ethereum did the same thing, but in the form of a contractual legal system. Arweave does the same, but around the business model of permanent information storage. If you try to buy this service from a centralized company, you can never ensure that in 50 or 100 years, that company won’t change its model, and your data will disappear.

Therefore, Arweave leverages the mathematical principles of business models to be protocolized, verifiable, and static, so that if you will, it can be trusted without having to trust anyone. So we really believe that it is very important to deeply explore the fundamental properties and utility of blockchain, rather than just transplanting existing systems to run on top. You must highlight the fundamental things that make blockchain possible, rather than just imagining the world in a Web2 way.

Odaily: What are the current key metrics for Arweave, such as storage capacity, market share, and the categories and proportions of stored data?

Sam: Arweave is used to store a large amount of Web3 data, and it is difficult to accurately calculate the amount of data related to Web3. But we know that, for example, in Solana, 70% to 80% of the data in the top 50 NFT collections is stored on Arweave, which gives you a good quick understanding of the situation.

In the past year, the usage of the network on individual data items has increased dramatically, and the growth is astonishing. I think at this time last year, the average operating rate of the network was 30 transactions per second. Now the operating rate is around 450 to 500 transactions per second, depending on the date. So this is a sign of the network's usage increasing dramatically over time. Excitingly, Arweave can indeed scale on-chain data storage arbitrarily. So no matter how much usage increases, it will continue to scale.

That’s why we are so excited about AO. Because the network we are building, Arweave is a foundational layer that will not hinder the scalability of top-layer computing systems. If we built on anything else, there would be a breakpoint. But Arweave provides us with a solid starting point. In terms of AO, just in a little over a week, the AO testnet has already transmitted over 2.5 million messages in about 8 days, which is fantastic. We expected around 100 developers to try it in the first week. Ultimately, it was actually 1500. We need to get the exact numbers within this range. So the enthusiasm of the ecosystem for this is fantastic.

AO has reached 5M in information volume on the testnet, source: Twitter

Odaily: Can you introduce one or two key projects in the Arweave ecosystem? How are they innovating based on Arweave's features, and what different experiences do they bring to end users?

Sam: Yes, of course. There are many projects, and I don’t want to pick favorites, but there are some you might want to know about, such as Accord and ArDrive, which are simple systems that allow you to upload data to the Arweave network. They have simple user interfaces that make them look and feel like Google Drive, but they actually run on a completely decentralized infrastructure. They even allow you to pay with a credit card, but with your own keys signing the data. So you get the benefits of self-sovereign identity while being able to access the system extremely quickly even if you don’t have crypto tokens on hand.

There are also projects like Ar.io, which is a decentralized gateway network running on Arweave, allowing people to access web applications in a decentralized manner. So you can go to any Ario gateway prefixed with Sam, which is my Arns address. Therefore, if you visit sam.ar.dev or sam.gateway.io or any other gateway, you will get my web application, which is really exciting. It allows us to have fully decentralized web applications with a complete stack.

Additionally, there are completely different applications built on the core infrastructure of Arweave. For example, decentralized social media systems like Lens or CyberConnect store their posts on Arweave because it allows for on-chain storage of any scale. And frankly, I believe almost every project I just mentioned started using the network in the past year. Therefore, the growth has been tremendous.

Odaily: What key collaborations has Arweave recently engaged in?

Sam: Yes, I just mentioned a few, such as Lens and CyberConnect, as well as the fact that Arweave is used to store Web3 data across the entire crypto ecosystem. I think one of the most interesting and appropriate descriptions of Arweave is that it is the favorite infrastructure tool of your favorite Web3 developers. So it is really widely used across the tech stack. I am particularly excited about the use of Arweave for storing on-chain data. For example, Lens has created an L3 solution on top of Polygon in the past six months, utilizing Arweave's business data availability network. Then AO is actually the super scaling of that model. So to answer your question, many people are using Arweave, and they typically don’t tell us when they start using it. They just pick it up and leverage the network.

If you want to see where this usage is coming from, I would recommend projects like Data OS for a deeper dive.

One particularly interesting example I want to highlight is a team called Dria (Dria.co) that is using Arweave as a decentralized permanent storage network for AI agents. They leverage Arweave to store LLM embeddings and make them available and authorized so that agents can pay for the data they use. This way, payments are made to all providers of that data, and there is an open, continuously evolving knowledge base. So we think this is very exciting. In the past few months, due to these AI use cases, the usage of the network has seen a significant increase.

Odaily: Can we continue to talk about strategic planning? What is Arweave's overall strategic plan for 2024?

Sam: I mean, Arweave is a protocol; I have my own views on how things should develop. There are many different institutions in the Arweave ecosystem right now. Therefore, I can only really tell you what I think. But because the protocol itself is solidified, it doesn’t change often at this point. I think what is truly exciting is what we are building, which is the application of platforms like AO. This is a lot of our focus, along with some quite exciting news.

We will soon have relatively significant news about decentralized social networks on top of Arweave. This is where our focus lies. We hope to turn Arweave into an L1 ecosystem with the best financial infrastructure, if you will, with autonomous agents roaming within the financial infrastructure, as well as social content. All of these come together in a way that we believe is truly important. If decentralized social media is to happen, it will need to provide better monetization for creators. By having AO as a super scalable computer as the underlying layer, we can scale so that bots can trade content in real-time within those social networks. This provides liquidity for creators, which is a very powerful effect that we believe will push many creators into Web3 in ways never seen before, but we will have news on this soon.

Odaily: So what should the community and developers look forward to the most in 2024?

Sam: I think it is the emergence of the AO ecosystem on our platform. If you haven’t tried it today, I highly recommend it. It is a unique experience in the crypto space. No other machine can work like this. When you want to interact with AO, as a developer, you don’t need to create your contracts in this rigid environment, compile them, and send them to the network, like Ethereum or anything else. Instead, what you do is start a new process inside AO, just like you have a server in EC2 or other cloud hosts, and input commands.

What we see in 2024 is that the tools around this are maturing. Now people are starting to build a fairly large financial ecosystem and infrastructure on it, and in fact, things really started to take off in the first week. So this is very exciting. What you can build on AO is completely impossible outside of it. For example, decentralized, verifiable smart contracts can read the entire state of a webpage, perhaps reading news and then autonomously acting based on the information within it, and so on. Yes, this really realizes the dream of a world computer. This time we really achieved it. It works and has enough scalability to support long-term future use. So we are very excited about this. We believe it will change everything.

Odaily: Sounds great. Finally, if possible, I would like to briefly talk about tokens. From an investor's perspective, what is the fundamental logic of the AR token in the near term, such as security, value retention characteristics, etc.?

Sam: I usually don’t discuss token economics, that’s for sure. Yes, if you want to understand how Arweave works, you can read our documentation.