The Arweave Mainnet was launched on November 18, 2018 and has undergone 13 major upgrades in more than five years. Its mission is to provide permanent decentralized storage services. However, when we observe the changes in network data, we find that the above upgrades have not truly formed a moat for Arweave. By observing Arweave's business data, we can see that:
After entering 2023, the growth of Arweave's storage business has significantly slowed down, with a sharp decline in the increase of stored data. The monthly network storage capacity fluctuates between 2-4 TiB, with the lowest network storage capacity in June, only 1.43 TiB. The total storage capacity for the year 2023 is 32.96 TiB. In comparison, the total network storage capacity of Filecoin in 2023 exceeds 1.8 EiB (1EiB = 1,048,576 TiB). It can be seen that Arweave is unable to shake Filecoin's dominant position in decentralized storage, and its business expansion is very difficult.
Not only is it difficult to break through the horizontal competition in the same type of business, but the track where Arweave is located - decentralized storage, is somewhat distant from ordinary retail investors. Ordinary users do not have strong motivation to use it and cannot quickly perceive changes in the fundamentals. In the new bull market, decentralized storage is somewhat influenced by the AI/DePIN narrative, but it is still nothing new and has not received much attention from the market.
The same dilemma is reflected in the price of the currency. Over a one-year period, before February 2024, the price of AR has been hovering around 6-10 dollars, significantly underperforming BTC, without experiencing an increase with the mainnet upgrade and the start of the bull market, until founder Sam officially announced the launch of AO on February 14th.
As a storage protocol, Arweave can only be regarded as a hard drive. It is impossible to support larger narratives and use cases with just a hard drive. For a long time, Arweave has relied on other protocols to utilize its hard drive, but the results have been minimal. It is too far away from the infrastructure of ordinary users and lacks market attention. Therefore, Arweave has created a CPU - AO that is fully compatible with its own hard drive, which has also caused its token price to quickly pump from $8 to nearly $50. This article will mainly introduce the AO technology principles and related ecosystem.
1.1 AO Technology Principle: How to Achieve Verifiable Infinite Computation?
AO is an Actor Oriented computer running on Arweave, designed to be an environment that can accommodate any number of parallel processes, with processes coordinated through an open message passing layer.
The most core feature of AO can be summarized as two:
(1) Any number of processes can run in parallel, i.e. infinite expansion of computing power;
(2) The verifiability and reproducibility of the calculation results, thus achieving minimal trust.
Before introducing how AO achieves the above functions, let's first understand the basic components of AO. The AO system consists of two basic units: processes and messages, as well as three basic units (which can also be seen as three important roles): scheduling units (SU), computing units (CU), and messaging units (MU).
Process: A unit of computing in a network. The state of a process can be calculated by computing units, and the process can accept messages from users and other processes. In terms of definition, $P_i$ represents the $i ^{th}$ process. Define $P_i$ = ($Log_i,Init_i,Env_i$), where $Log_i$ is $Pi$ $Init_i$ is the initialization data of $P_i$, $Sched_i$ is the scheduler of $P_i$, $Env_i$ is the computing environment of $P_i$, and at a given time step, the state of $P_i$ $S(P_i) = F(Log_i, Env_i)$, where F is a function, defined by $Env_i$, calculates the state from the message log.
Message: Every interaction with the process in AO is represented as a message. The nature of a message is a data item that conforms to the ANS-104 standard. The uniformity of message format is crucial, and the entire AO environment settles and processes on the decentralized data layer of Arweave through unified messages.
Scheduling Unit: The scheduling unit is responsible for assigning atomic increment time slot numbers (similar to Ethereum nonce) to the messages sent to the process, that is, sorting the process messages. Once allocated, the scheduler needs to ensure that the data is uploaded to Arweave for permanent access by others.
Calculation Unit: The calculation unit is a node responsible for calculating the process status in AO. The calculation units form a decentralized computing market similar to the Akash computing protocol. The calculation units compete with each other to complete the computation of process status. After the service is completed, the calculation unit needs to return the computation result and the state proof with a signature. If users do not trust a single calculation unit, they can further request from more calculation units. The calculation unit needs to stake a certain amount, and if it provides incorrect states, its stake will be subject to slashing.
Messenger Unit: The Messenger Unit is responsible for receiving incoming messages from the client and routing these messages to the designated Dispatch Unit, then retrieving the results from the Computing Unit.
AO achieves verifiable infinite computing power, with key technical points mainly including the following aspects:
(1) Based on the storage consensus paradigm (SCP): AO computers achieve consensus by holographic storage of message logs in Arweave. Arweave is treated as a continuous and immutable logbook to store all message logs, ensuring that interactive logs are permanently available, allowing any network participant to calculate the state.
(2) Consensus only needs to be reached on data order and storage, and no Consensus is required on state: Blockchain such as Bitcoin and Ethereum adopt the traditional Consensus Mechanism, that is, let network participants agree on the state of the ledger, and all nodes need to verify and reach consensus on the current state, which causes a waste of computing resources and limits the speed and scalability of the network. AO does not need to reach a consensus on the state, only needs to agree on the order and storage of the interaction log in Arweave, that is, the state of AO is "holographic" implied in the message log hosted by Arweave, although there is no consensus on the state, but everyone can calculate the state through the data stored in Arweave in principle, if the user wants to get the state, you can request the computing unit to complete the calculation and provide proof.
(3) Functional separation of AR and AO: Combined with the above analysis, AR and AO perform their respective duties, and AO does not solve the verification problem, but is only responsible for completing the transmission, sorting and state calculation of the message, mainly dealing with the calculation problem; Arweave handles security and verifiability issues, builds consensus on data order, and guarantees permanently immutable Decentralization storage. AO is stored based on the interaction log on Arweave, but cannot modify the Arweave consensus.
(4) Parallel computing architecture based on message passing: The two basic methods to achieve parallel computing are shared memory and message passing, and AO adopts the architecture of message passing, as opposed to the shared memory used by parallel Blockchain such as Solana and Sei. In the shared memory mode, when one user accesses and modifies some data, other users cannot modify it, that is, they need to wait for each other in the process of "lock contention" until the user "unlocks" the data, and then others can access it, so there is a certain upper limit to the scalability of this method. AO, on the other hand, only needs to send interaction information at the time of interaction, and there is no waiting for "lock contention", which realizes horizontal scaling, which makes its parallel scalability reach any scale.
(5) Modular architecture of AO: The modularity of AO is reflected in the separation of CU, SU, and MU. Users can freely choose suitable sorters, message relays, computing units, and even system-level virtual machines to support the introduction of processes in various smart contract systems. CU, SU, and MU can achieve horizontal scaling, ensuring that computing power increases as needed.
Based on the above analysis, the key processes in the AO network include: the interactive information in the AO network is passed to SU after being verified and signed by MU, uploaded to Arweave after being sorted by SU, and consensus is reached and stored in sequence on Arweave. When a user needs to obtain the status, the message will be passed to SU through MU, SU will select the appropriate CU and deliver the message to CU, and CU will complete the status calculation through the data on Arweave, and return the output to the user through MU.
**1.2 Does AO's technology have a moat?
To explore this issue, we will compare AO with two similar projects: high-performance parallel chains (such as Parallel EVM, Solana, etc.) and decentralized computing protocols (Akash).
Comparison of high-performance public chains and AO in parallelism:
The main difference lies in the difference in the basic architecture of parallel computing. Here we take Ethereum as a benchmark object to better illustrate the differences of AO. The traditional EVM represented by Ethereum processes transactions in order, and only one transaction can modify the state at a time, presenting a single linear forward mode for the entire system.
The parallel public chain generally completes the parallel processing of non-conflicting or non-overlapping transactions (conflicting transactions: generally refer to multiple transactions attempting to simultaneously access and modify the same data or state, which usually leads to data inconsistency), for example, Sealevel allows Solana to simultaneously process thousands of smart contracts, each transaction describes the state it will read or write, the system completes the parallel execution of these non-overlapping and non-conflicting transactions by identifying non-overlapping transactions, similar to parallel EVM, it processes non-conflicting transactions in parallel. Taking Monad as an example, its core process includes three steps: (1) optimistic execution, which optimistically assumes that all transactions are non-conflicting and conducts parallel execution, but may lead to errors. By tracking input-output comparisons, transactions will be re-executed in places where the data is inconsistent; (2) scheduling and dependencies, to reduce unnecessary duplicate execution, Monad will use a static code analyzer to predict the dependency relationships between transactions, that is, to pre-identify possible conflicts between some transactions to optimize execution; (3) state merging, after parallel execution of transactions, the updated states of each transaction need to be merged to ensure the consistency of the entire block's state.
Although the efficiency of the system is improved, there is an obvious bottleneck in the expansion of the parallelism public chain: only non-conflicting transactions can be processed in parallel, and if it involves the access and modification of the same state, it still involves the problem of "lock contention". The differences between AO and parallel public chains are: (1) a parallel computing architecture that uses message passing instead of shared memory; (2) Consensus is only reached on the order of stored data, and Consensus on state is not required. This makes AO have stronger parallel scalability, and the scheduling unit, messenger unit and computing unit can be horizontally and infinitely expanded, ensuring the infinite expansion of computing power.
Comparison between decentralized computing market and AO:
Networks such as Akash provide a Decentralization computing marketplace for container-hosted services, but sacrifice the ability to create trustless services, i.e., their computational results are not verifiable and reproducible, and therefore lose the ability of smart contracts. In contrast, AO's computation is verifiable computation, and thanks to its holographic state storage mechanism, AO can even maintain the properties of traditional smart contracts. AO ensures that interaction logs are written to Arweave with persistent availability, the state can be computed by any participant, and everyone can reproduce this computation process and verify the correctness of other computations. Measures to ensure verifiability and trustless services can be summarized as follows: (1) holographic storage of all interaction logs on Arweave, making the computation process reproducible; (2) The computing unit needs to provide an encryption signature statement on the calculation result; (3) The computing unit needs to be staked, and when the calculation result is wrong, it will face slashing.
In summary, the architecture combining AO and AR ensures both the unlimited scalability of computing power and the verifiability and minimal trust of computation. Compared to similar existing projects, it has certain differences and moats.
1.3 AO ecology is booming
The AO ecosystem is still in its very early stages, but it is thriving. According to the overall data from AOlink, the current number of messages processed by the AO network has exceeded 116M, with a daily user count reaching over 5K during its peak period, but recently falling to around 1.5K. The number of holders of the test token for the AO network ($AOCRED, which is used to reward AO network builders) has also surpassed 4100.
Within three months after the release of the AO testnet, the AO network has initially established infrastructure and financial systems, including cross-chain bridges, oracle machines, wallets, AMM, stablecoin protocols, etc. At the same time, applications such as gaming, social networking, Memecoin, AI, etc. are also under development.
（As of 2024/4/26, AO Ecological Panorama, source: @everPayHQ @ArweaveSCP）
The key protocols include:
AOX: The first cross-chain bridge in the AO ecosystem, which provides asset cross-chain services for the AO network based on MPC technology. It is currently in the Beta stage and only supports cross-chain between $AR on Arweave and the AO network. In AO, it corresponds to Wrapped AR. Currently, incentive tasks are being launched, and users can earn $TAOX test tokens by completing cross-chain operations and other tasks, which may correspond to future official token airdrops.
0rbit: The oracle machine of the AO network, which supports bringing any data to the AO process through a valid URL. Users can send message requests for data to 0rbit, and the 0rbit node will retrieve the data and deliver it to the user's process.
Arconnect, aoWebWallet: Wallet infrastructure for the AO network.
Astro: The stablecoin protocol of the AO network, which went online with the testnet in April. Currently, you can receive test tokens tAR and mint stablecoin USDA through tAR.
Permaswap, ArSwap, and Bark: DEX in the AO network. Permaswap has recently launched on AO, currently supporting the exchange between Wrapped AR and AOCRED; ArSwap and Bark were launched earlier, supporting a wider range of assets, in addition to AOCRED and Wrapped AR, they also support other projects or meme tokens within the ecosystem.
typr: AO ecosystem's Twitter, a social application with open functionalities and a UI that is basically the same as Twitter, including post, long-form story, and chatroom, but supporting rewards with four types of assets: TRUNK/Wrapped AR/AOCRED/typr test token.
Permaverse: AO Network's game and Metaverse issuance platform. The currently issued game is dumdum. Players can earn points by touching their dumdum (a green elephant), which corresponds to potential airdrop rewards. It also provides a very simple Metaverse environment for dumdum.
AO Games: The game and Metaverse issuance platform just launched this week, the product features mentioned in Twitter include: Web2-like gaming experience, the ability to integrate on-chain AI, etc.
outcome_gg: The prediction market of the AO ecosystem, with prediction indicators including AO ecosystem, games, defi, meme, business, technology, etc. After the project is announced, it may introduce AI autonomous agents and conduct prediction contests relying on LLM.
TRUNK, Aetheris: Memecoins in the AO ecosystem.
AOVM: AI tools of AO ecosystem, for end users, can use artificial intelligence as personal assistant, market data analysis and other tools, for developers, can be used as smart contract development assistant, the product has not yet been launched.
Overall, the AO ecosystem is still in the very early stage, just entering the stage of "laying the foundation", long applications are still in the White Paper stage, and even only Twitter and website pages, no specific products and technical documents are online, and the technical capabilities of the AO network have not yet been released and tested in the Application Layer. The development of the AO ecosystem will also feed back Arweave's storage business - if the external business development is not smooth, then the self-built CPU will unleash the potential of this hard drive.
What is the relationship between AO and AI tracks?
After the launch of AO, Arweave was recompared with projects in the AI track and parallel EVM track. In the previous section, we have introduced the differences between AO and parallel EVM. This section will briefly analyze the position of AO in the AI track.
AO itself adopts the Actor model for design, and the Actor model itself has a close relationship with AI research. The core idea is that each component of the system can be an independent and autonomous agent. When interaction is needed, it is achieved by sending messages. The Actor in this model is very similar to an AI Agent, which makes hosting AI models and building AI applications in the AO ecosystem an attractive direction. So specifically, how will AO empower AI?
In a nutshell, AO makes it more feasible to put AI models on the blockchain, enabling verifiability of AI models and promoting the integration of AI models with smart contracts, thus expanding the use of AI in the Crypto world.
First, on-chain modeling refers to storing the ML model in the smart contract of the blockchain and using the model by calling the methods of the smart contract, but this requires: (1) storing the AI model and data on-chain, that is, storing the complete model and the required data among thousands of nodes, the storage cost is extremely high, and on-chain storage of models, especially large language models, is economically infeasible; (2) limited computing resources, while the blockchain has high latency and low throughput, which will limit the high-performance computing of AI models, computing AI models on-chain requires all nodes to complete the computation process simultaneously, this single-threaded architecture is obviously unable to support.
Therefore, the main computation model is currently mainly executed off-chain, and the results are returned on-chain. A compromise method is to use opml/zkml to prove the inference results and put them on-chain to improve the verifiability of off-chain computation.
Compared to traditional blockchains (such as Ethereum), AO's technical advantages are: (1) the ability to natively integrate with Arweave, AR provides a storage layer, making it possible to store large-scale data at low cost; (2) the ability to achieve verifiable and infinitely scalable parallel computing. This allows AO to solve many problems in traditional blockchains, such as storing AI models, making it possible to host large language models, and parallel computing effectively alleviates the demand for computing resources. It is not necessary for all nodes to redundantly complete model calculations, reducing computational redundancy and improving computational efficiency. In addition, all data in Arweave can be used as input for AO computation, greatly increasing the data available for on-chain models, promoting on-chain agents or AI applications to rely on more reliable data for decision-making.
The first focus of AO Ecology in the field of AI will be the combination of AI and finance, proposing AgentFi. AgentFi refers to the use of AI's reasoning ability to create and adjust complex strategies similar to fund managers. Manipulating money using AI models is a sensitive transaction, especially when trustworthiness is crucial. Compared to introducing AgentFi on other chains, AO Ecology is the first to establish the verifiability of computations. The first project launched is Autonomous Finance, which aims to include financial agents such as DCA asset management agents, autonomously balanced index fund agents, autonomously hedged funds agents with customized risk strategies, on-chain prediction agents, high-frequency trading agents, etc. Currently, the DCA investment agent product has been launched, where users can set parameters such as investment asset types, slippage range, liquidity pool, and investment time. However, this has not yet achieved the use of AI wisdom to formulate investment strategies, but rather remains at the level of contract automation that does not require off-chain triggers. We still need to track the capabilities of subsequent products to determine whether they are serious products or just gimmicks.
3 Tokenomics and Chip Analysis
On May 30, 2024, AO announced that the issuance of $AO tokens will be completed soon, and the token will go live on June 13th, Beijing time. In addition, $AR is a token related to the project, and it remains one of the speculative targets before the completion of AO's Token Generation Event (TGE).
The maximum supply of $AR tokens is 66 million, with an initial minting of 55 million. Currently, all of them have entered the market circulation, and the remaining 11 million are used as mining rewards, of which 10,744,796 tokens have been mined. The current mining reward per block is about 0.75 $AR, which will halve every year. Arweave mines approximately 660 blocks per day, resulting in a daily circulation increase of about 500 $AR, with minimal selling pressure from mining. The current circulating supply of $AR is 65,744,796, and the unlocking rate has reached 99.61%, indicating that it can be considered to be in a fully circulated state.
In terms of token use cases, $AR is used as a payment medium for users to store data, as well as an incentive for miners to produce blocks and store data. Arweave has a Storage Endowment mechanism, where the storage fees paid by users are not fully distributed to miners. Currently, only 16.67% is distributed to miners, while the rest automatically goes into the storage endowment. Therefore, whenever data is uploaded, the Arweave network moves the corresponding amount of tokens in circulation to the endowment, which is used to pay for accumulated data storage fees over time. This endowment is only distributed when the storage costs for miners exceed the sum of new mining rewards and transaction fees, ensuring that mining is always profitable. However, since the birth of Arweave, no one has withdrawn a token from the storage endowment, which has led to the storage endowment being seen as a burning mechanism for $AR. When the growth rate of the storage endowment exceeds the addition of new $AR, $AR can be considered to enter a deflationary state.
From the perspective of the price trend, after the release of AO, the price of $AR rose rapidly, quadrupling within a month. During the recent brief bear market, the price of AR rose against the trend and is approaching a new high. Currently, the concentration of chips is still around $10, and the other main concentration areas are around $20-$40. $AR has tested the $47 area twice, and the price is currently approaching this level again.
In terms of valuation analysis, there are currently no projects in the same category as Arweave and AO, with similar projects including parallel EVM, high-performance public chain decentralized storage, and decentralized computing protocols. In terms of market capitalization, Arweave is basically on par with high-performance public chains and Filecoin, higher than Akash and Sei, but much lower in terms of FDV compared to public chains and storage projects. Therefore, considering that the AO ecosystem is still in its very early stages and the mainnet is not yet online, the current market capitalization of $AR is not significantly undervalued. However, the almost fully circulating state reduces the dilution effect of token issuance, which may result in relatively minor resistance to potential future pumps.
Token is closely related to the future issuance of AO tokens. According to official news, $AO will be launched 100% fairly, without pre-mining, pre-sale, or preferential acquisition. The total supply of tokens is 21M, and it will undergo halving every four years. It is important to note the ways to obtain tokens: (1) Asset bridging to AO (2) Holding $AR (3) Participating in the construction of AO ecosystem. The specific token model of AO has not been announced yet. According to the answer of outprog, one of the proposers of AO, in X Space, AO token and AR token will have their respective functions. AR token mainly focuses on the storage function and consensus maintenance of Arweave, while AO token focuses on solving the communication problem between computation and application, that is, AO and AR respectively maintain the computation and storage functions of the network.
After the announcement of the $AO token, the maximum short-term increase of $AR exceeded 18%. On one hand, $AR is currently almost the only target to take on this event, and on the other hand, it may be related to the rules for holding $AR to obtain $AO tokens. The rules of bridging assets to AO (currently $AR is basically the only asset that can be cross-chain to the AO network) and holding $AR are actually both digesting the selling pressure on $AR from this event. However, it is important to note that the $AR token currently carries a dual valuation of the market for computing and storage. When the $AO is about to be released, it also needs to reevaluate the $AR or dilute a portion of the market capitalization of $AR.
4 Summary
Although the market always says that "the new is not the old", because the old narrative often mobilizes the new attention of the market, some old projects are bringing new surprises with technological innovation, and compared with the new coin projects, the Token of these old projects has basically entered the full circulation stage, unlocking and selling smaller, and fully bottoming out in the Bear Market, then these old projects may bring greater upside and pump short, Arweave is an example of this kind of project. The combination of AO and AR with verifiable infinite parallel computing capabilities provides Arweave with a technical moat and injects new vitality and narrative.
From the perspective of ecological development, AO is still in the testnet stage, ecological construction is in the extremely early stage, and the real computing power of AO has not been actually tested and played. We expect that the AO ecosystem should give birth to some projects that can give full play to its unique technical capabilities such as storage, verifiable computing, and massively parallel computing, such as decentralization social applications (which can ensure the permanent storage and recall of social data), AI infrastructure and applications, etc. From the perspective of market attention, although the coin price of AR has been steadily pump some time ago, the degree of discussion among ordinary users is not high, and AO Testnet has not become a "big hair" to attract market attention. Due to its complex technical details, it is difficult for ordinary users to recognize the difference between AO and high-performance public chains and Decentralization protocol, and how AO can cut into the AI track, so they do not fully realize the rise potential of AO. AO is still in the extreme early stage of protocol development, and the follow-up development short time is large, but if the ecosystem development is not as expected, especially if the Mainnet does not meet expectations in terms of performance and user experience after launch, and does not run out of AI projects with sufficient technical capabilities and out of the circle, the valuation of $AO will be greatly discounted.
The content is for reference only, not a solicitation or offer. No investment, tax, or legal advice provided. See Disclaimer for more risks disclosure.
Metrics Ventures Research Report | Coin Issuance Imminent, Comprehensive Interpretation of AO Technology Principles and Ecological Potential
1 Storage-based, AO helps Arweave make a comeback
The Arweave Mainnet was launched on November 18, 2018 and has undergone 13 major upgrades in more than five years. Its mission is to provide permanent decentralized storage services. However, when we observe the changes in network data, we find that the above upgrades have not truly formed a moat for Arweave. By observing Arweave's business data, we can see that:
After entering 2023, the growth of Arweave's storage business has significantly slowed down, with a sharp decline in the increase of stored data. The monthly network storage capacity fluctuates between 2-4 TiB, with the lowest network storage capacity in June, only 1.43 TiB. The total storage capacity for the year 2023 is 32.96 TiB. In comparison, the total network storage capacity of Filecoin in 2023 exceeds 1.8 EiB (1EiB = 1,048,576 TiB). It can be seen that Arweave is unable to shake Filecoin's dominant position in decentralized storage, and its business expansion is very difficult.
Not only is it difficult to break through the horizontal competition in the same type of business, but the track where Arweave is located - decentralized storage, is somewhat distant from ordinary retail investors. Ordinary users do not have strong motivation to use it and cannot quickly perceive changes in the fundamentals. In the new bull market, decentralized storage is somewhat influenced by the AI/DePIN narrative, but it is still nothing new and has not received much attention from the market.
The same dilemma is reflected in the price of the currency. Over a one-year period, before February 2024, the price of AR has been hovering around 6-10 dollars, significantly underperforming BTC, without experiencing an increase with the mainnet upgrade and the start of the bull market, until founder Sam officially announced the launch of AO on February 14th.
As a storage protocol, Arweave can only be regarded as a hard drive. It is impossible to support larger narratives and use cases with just a hard drive. For a long time, Arweave has relied on other protocols to utilize its hard drive, but the results have been minimal. It is too far away from the infrastructure of ordinary users and lacks market attention. Therefore, Arweave has created a CPU - AO that is fully compatible with its own hard drive, which has also caused its token price to quickly pump from $8 to nearly $50. This article will mainly introduce the AO technology principles and related ecosystem.
1.1 AO Technology Principle: How to Achieve Verifiable Infinite Computation?
AO is an Actor Oriented computer running on Arweave, designed to be an environment that can accommodate any number of parallel processes, with processes coordinated through an open message passing layer.
The most core feature of AO can be summarized as two:
(1) Any number of processes can run in parallel, i.e. infinite expansion of computing power;
(2) The verifiability and reproducibility of the calculation results, thus achieving minimal trust.
Before introducing how AO achieves the above functions, let's first understand the basic components of AO. The AO system consists of two basic units: processes and messages, as well as three basic units (which can also be seen as three important roles): scheduling units (SU), computing units (CU), and messaging units (MU).
AO achieves verifiable infinite computing power, with key technical points mainly including the following aspects:
(1) Based on the storage consensus paradigm (SCP): AO computers achieve consensus by holographic storage of message logs in Arweave. Arweave is treated as a continuous and immutable logbook to store all message logs, ensuring that interactive logs are permanently available, allowing any network participant to calculate the state.
(2) Consensus only needs to be reached on data order and storage, and no Consensus is required on state: Blockchain such as Bitcoin and Ethereum adopt the traditional Consensus Mechanism, that is, let network participants agree on the state of the ledger, and all nodes need to verify and reach consensus on the current state, which causes a waste of computing resources and limits the speed and scalability of the network. AO does not need to reach a consensus on the state, only needs to agree on the order and storage of the interaction log in Arweave, that is, the state of AO is "holographic" implied in the message log hosted by Arweave, although there is no consensus on the state, but everyone can calculate the state through the data stored in Arweave in principle, if the user wants to get the state, you can request the computing unit to complete the calculation and provide proof.
(3) Functional separation of AR and AO: Combined with the above analysis, AR and AO perform their respective duties, and AO does not solve the verification problem, but is only responsible for completing the transmission, sorting and state calculation of the message, mainly dealing with the calculation problem; Arweave handles security and verifiability issues, builds consensus on data order, and guarantees permanently immutable Decentralization storage. AO is stored based on the interaction log on Arweave, but cannot modify the Arweave consensus.
(4) Parallel computing architecture based on message passing: The two basic methods to achieve parallel computing are shared memory and message passing, and AO adopts the architecture of message passing, as opposed to the shared memory used by parallel Blockchain such as Solana and Sei. In the shared memory mode, when one user accesses and modifies some data, other users cannot modify it, that is, they need to wait for each other in the process of "lock contention" until the user "unlocks" the data, and then others can access it, so there is a certain upper limit to the scalability of this method. AO, on the other hand, only needs to send interaction information at the time of interaction, and there is no waiting for "lock contention", which realizes horizontal scaling, which makes its parallel scalability reach any scale.
(5) Modular architecture of AO: The modularity of AO is reflected in the separation of CU, SU, and MU. Users can freely choose suitable sorters, message relays, computing units, and even system-level virtual machines to support the introduction of processes in various smart contract systems. CU, SU, and MU can achieve horizontal scaling, ensuring that computing power increases as needed.
Based on the above analysis, the key processes in the AO network include: the interactive information in the AO network is passed to SU after being verified and signed by MU, uploaded to Arweave after being sorted by SU, and consensus is reached and stored in sequence on Arweave. When a user needs to obtain the status, the message will be passed to SU through MU, SU will select the appropriate CU and deliver the message to CU, and CU will complete the status calculation through the data on Arweave, and return the output to the user through MU.
**1.2 Does AO's technology have a moat?
To explore this issue, we will compare AO with two similar projects: high-performance parallel chains (such as Parallel EVM, Solana, etc.) and decentralized computing protocols (Akash).
Comparison of high-performance public chains and AO in parallelism:
The main difference lies in the difference in the basic architecture of parallel computing. Here we take Ethereum as a benchmark object to better illustrate the differences of AO. The traditional EVM represented by Ethereum processes transactions in order, and only one transaction can modify the state at a time, presenting a single linear forward mode for the entire system.
The parallel public chain generally completes the parallel processing of non-conflicting or non-overlapping transactions (conflicting transactions: generally refer to multiple transactions attempting to simultaneously access and modify the same data or state, which usually leads to data inconsistency), for example, Sealevel allows Solana to simultaneously process thousands of smart contracts, each transaction describes the state it will read or write, the system completes the parallel execution of these non-overlapping and non-conflicting transactions by identifying non-overlapping transactions, similar to parallel EVM, it processes non-conflicting transactions in parallel. Taking Monad as an example, its core process includes three steps: (1) optimistic execution, which optimistically assumes that all transactions are non-conflicting and conducts parallel execution, but may lead to errors. By tracking input-output comparisons, transactions will be re-executed in places where the data is inconsistent; (2) scheduling and dependencies, to reduce unnecessary duplicate execution, Monad will use a static code analyzer to predict the dependency relationships between transactions, that is, to pre-identify possible conflicts between some transactions to optimize execution; (3) state merging, after parallel execution of transactions, the updated states of each transaction need to be merged to ensure the consistency of the entire block's state.
Although the efficiency of the system is improved, there is an obvious bottleneck in the expansion of the parallelism public chain: only non-conflicting transactions can be processed in parallel, and if it involves the access and modification of the same state, it still involves the problem of "lock contention". The differences between AO and parallel public chains are: (1) a parallel computing architecture that uses message passing instead of shared memory; (2) Consensus is only reached on the order of stored data, and Consensus on state is not required. This makes AO have stronger parallel scalability, and the scheduling unit, messenger unit and computing unit can be horizontally and infinitely expanded, ensuring the infinite expansion of computing power.
Comparison between decentralized computing market and AO:
Networks such as Akash provide a Decentralization computing marketplace for container-hosted services, but sacrifice the ability to create trustless services, i.e., their computational results are not verifiable and reproducible, and therefore lose the ability of smart contracts. In contrast, AO's computation is verifiable computation, and thanks to its holographic state storage mechanism, AO can even maintain the properties of traditional smart contracts. AO ensures that interaction logs are written to Arweave with persistent availability, the state can be computed by any participant, and everyone can reproduce this computation process and verify the correctness of other computations. Measures to ensure verifiability and trustless services can be summarized as follows: (1) holographic storage of all interaction logs on Arweave, making the computation process reproducible; (2) The computing unit needs to provide an encryption signature statement on the calculation result; (3) The computing unit needs to be staked, and when the calculation result is wrong, it will face slashing.
In summary, the architecture combining AO and AR ensures both the unlimited scalability of computing power and the verifiability and minimal trust of computation. Compared to similar existing projects, it has certain differences and moats.
1.3 AO ecology is booming
The AO ecosystem is still in its very early stages, but it is thriving. According to the overall data from AOlink, the current number of messages processed by the AO network has exceeded 116M, with a daily user count reaching over 5K during its peak period, but recently falling to around 1.5K. The number of holders of the test token for the AO network ($AOCRED, which is used to reward AO network builders) has also surpassed 4100.
Within three months after the release of the AO testnet, the AO network has initially established infrastructure and financial systems, including cross-chain bridges, oracle machines, wallets, AMM, stablecoin protocols, etc. At the same time, applications such as gaming, social networking, Memecoin, AI, etc. are also under development.
（As of 2024/4/26, AO Ecological Panorama, source: @everPayHQ @ArweaveSCP）
The key protocols include:
Overall, the AO ecosystem is still in the very early stage, just entering the stage of "laying the foundation", long applications are still in the White Paper stage, and even only Twitter and website pages, no specific products and technical documents are online, and the technical capabilities of the AO network have not yet been released and tested in the Application Layer. The development of the AO ecosystem will also feed back Arweave's storage business - if the external business development is not smooth, then the self-built CPU will unleash the potential of this hard drive.
What is the relationship between AO and AI tracks?
After the launch of AO, Arweave was recompared with projects in the AI track and parallel EVM track. In the previous section, we have introduced the differences between AO and parallel EVM. This section will briefly analyze the position of AO in the AI track.
AO itself adopts the Actor model for design, and the Actor model itself has a close relationship with AI research. The core idea is that each component of the system can be an independent and autonomous agent. When interaction is needed, it is achieved by sending messages. The Actor in this model is very similar to an AI Agent, which makes hosting AI models and building AI applications in the AO ecosystem an attractive direction. So specifically, how will AO empower AI?
In a nutshell, AO makes it more feasible to put AI models on the blockchain, enabling verifiability of AI models and promoting the integration of AI models with smart contracts, thus expanding the use of AI in the Crypto world.
First, on-chain modeling refers to storing the ML model in the smart contract of the blockchain and using the model by calling the methods of the smart contract, but this requires: (1) storing the AI model and data on-chain, that is, storing the complete model and the required data among thousands of nodes, the storage cost is extremely high, and on-chain storage of models, especially large language models, is economically infeasible; (2) limited computing resources, while the blockchain has high latency and low throughput, which will limit the high-performance computing of AI models, computing AI models on-chain requires all nodes to complete the computation process simultaneously, this single-threaded architecture is obviously unable to support.
Therefore, the main computation model is currently mainly executed off-chain, and the results are returned on-chain. A compromise method is to use opml/zkml to prove the inference results and put them on-chain to improve the verifiability of off-chain computation.
Compared to traditional blockchains (such as Ethereum), AO's technical advantages are: (1) the ability to natively integrate with Arweave, AR provides a storage layer, making it possible to store large-scale data at low cost; (2) the ability to achieve verifiable and infinitely scalable parallel computing. This allows AO to solve many problems in traditional blockchains, such as storing AI models, making it possible to host large language models, and parallel computing effectively alleviates the demand for computing resources. It is not necessary for all nodes to redundantly complete model calculations, reducing computational redundancy and improving computational efficiency. In addition, all data in Arweave can be used as input for AO computation, greatly increasing the data available for on-chain models, promoting on-chain agents or AI applications to rely on more reliable data for decision-making.
The first focus of AO Ecology in the field of AI will be the combination of AI and finance, proposing AgentFi. AgentFi refers to the use of AI's reasoning ability to create and adjust complex strategies similar to fund managers. Manipulating money using AI models is a sensitive transaction, especially when trustworthiness is crucial. Compared to introducing AgentFi on other chains, AO Ecology is the first to establish the verifiability of computations. The first project launched is Autonomous Finance, which aims to include financial agents such as DCA asset management agents, autonomously balanced index fund agents, autonomously hedged funds agents with customized risk strategies, on-chain prediction agents, high-frequency trading agents, etc. Currently, the DCA investment agent product has been launched, where users can set parameters such as investment asset types, slippage range, liquidity pool, and investment time. However, this has not yet achieved the use of AI wisdom to formulate investment strategies, but rather remains at the level of contract automation that does not require off-chain triggers. We still need to track the capabilities of subsequent products to determine whether they are serious products or just gimmicks.
3 Tokenomics and Chip Analysis
On May 30, 2024, AO announced that the issuance of $AO tokens will be completed soon, and the token will go live on June 13th, Beijing time. In addition, $AR is a token related to the project, and it remains one of the speculative targets before the completion of AO's Token Generation Event (TGE).
The maximum supply of $AR tokens is 66 million, with an initial minting of 55 million. Currently, all of them have entered the market circulation, and the remaining 11 million are used as mining rewards, of which 10,744,796 tokens have been mined. The current mining reward per block is about 0.75 $AR, which will halve every year. Arweave mines approximately 660 blocks per day, resulting in a daily circulation increase of about 500 $AR, with minimal selling pressure from mining. The current circulating supply of $AR is 65,744,796, and the unlocking rate has reached 99.61%, indicating that it can be considered to be in a fully circulated state.
In terms of token use cases, $AR is used as a payment medium for users to store data, as well as an incentive for miners to produce blocks and store data. Arweave has a Storage Endowment mechanism, where the storage fees paid by users are not fully distributed to miners. Currently, only 16.67% is distributed to miners, while the rest automatically goes into the storage endowment. Therefore, whenever data is uploaded, the Arweave network moves the corresponding amount of tokens in circulation to the endowment, which is used to pay for accumulated data storage fees over time. This endowment is only distributed when the storage costs for miners exceed the sum of new mining rewards and transaction fees, ensuring that mining is always profitable. However, since the birth of Arweave, no one has withdrawn a token from the storage endowment, which has led to the storage endowment being seen as a burning mechanism for $AR. When the growth rate of the storage endowment exceeds the addition of new $AR, $AR can be considered to enter a deflationary state.
From the perspective of the price trend, after the release of AO, the price of $AR rose rapidly, quadrupling within a month. During the recent brief bear market, the price of AR rose against the trend and is approaching a new high. Currently, the concentration of chips is still around $10, and the other main concentration areas are around $20-$40. $AR has tested the $47 area twice, and the price is currently approaching this level again.
In terms of valuation analysis, there are currently no projects in the same category as Arweave and AO, with similar projects including parallel EVM, high-performance public chain decentralized storage, and decentralized computing protocols. In terms of market capitalization, Arweave is basically on par with high-performance public chains and Filecoin, higher than Akash and Sei, but much lower in terms of FDV compared to public chains and storage projects. Therefore, considering that the AO ecosystem is still in its very early stages and the mainnet is not yet online, the current market capitalization of $AR is not significantly undervalued. However, the almost fully circulating state reduces the dilution effect of token issuance, which may result in relatively minor resistance to potential future pumps.
| Project Name | Project Type | MC | FDV | | --- | --- | --- | --- | | Arweave | Decentralized Storage and Parallel Computing | $2,872,218,566 | $2,898,720,000 | | Sei | Parallel EVM | $1,525,554,400 | $5,215,570,597 | | Sui | High-performance public chain | $2,402,908,724 | $10,272,365,941 | | Aptos | High-performance public chain | $4,116,070,332 | $10,356,559,209 | | FIL | Decentralized Storage | $3,284,109,281 | $11,548,151,694 | | Akash | Decentralized Computing | $1,182,479,902 | $1,205,387,098 |
Token is closely related to the future issuance of AO tokens. According to official news, $AO will be launched 100% fairly, without pre-mining, pre-sale, or preferential acquisition. The total supply of tokens is 21M, and it will undergo halving every four years. It is important to note the ways to obtain tokens: (1) Asset bridging to AO (2) Holding $AR (3) Participating in the construction of AO ecosystem. The specific token model of AO has not been announced yet. According to the answer of outprog, one of the proposers of AO, in X Space, AO token and AR token will have their respective functions. AR token mainly focuses on the storage function and consensus maintenance of Arweave, while AO token focuses on solving the communication problem between computation and application, that is, AO and AR respectively maintain the computation and storage functions of the network.
After the announcement of the $AO token, the maximum short-term increase of $AR exceeded 18%. On one hand, $AR is currently almost the only target to take on this event, and on the other hand, it may be related to the rules for holding $AR to obtain $AO tokens. The rules of bridging assets to AO (currently $AR is basically the only asset that can be cross-chain to the AO network) and holding $AR are actually both digesting the selling pressure on $AR from this event. However, it is important to note that the $AR token currently carries a dual valuation of the market for computing and storage. When the $AO is about to be released, it also needs to reevaluate the $AR or dilute a portion of the market capitalization of $AR.
4 Summary
Although the market always says that "the new is not the old", because the old narrative often mobilizes the new attention of the market, some old projects are bringing new surprises with technological innovation, and compared with the new coin projects, the Token of these old projects has basically entered the full circulation stage, unlocking and selling smaller, and fully bottoming out in the Bear Market, then these old projects may bring greater upside and pump short, Arweave is an example of this kind of project. The combination of AO and AR with verifiable infinite parallel computing capabilities provides Arweave with a technical moat and injects new vitality and narrative.
From the perspective of ecological development, AO is still in the testnet stage, ecological construction is in the extremely early stage, and the real computing power of AO has not been actually tested and played. We expect that the AO ecosystem should give birth to some projects that can give full play to its unique technical capabilities such as storage, verifiable computing, and massively parallel computing, such as decentralization social applications (which can ensure the permanent storage and recall of social data), AI infrastructure and applications, etc. From the perspective of market attention, although the coin price of AR has been steadily pump some time ago, the degree of discussion among ordinary users is not high, and AO Testnet has not become a "big hair" to attract market attention. Due to its complex technical details, it is difficult for ordinary users to recognize the difference between AO and high-performance public chains and Decentralization protocol, and how AO can cut into the AI track, so they do not fully realize the rise potential of AO. AO is still in the extreme early stage of protocol development, and the follow-up development short time is large, but if the ecosystem development is not as expected, especially if the Mainnet does not meet expectations in terms of performance and user experience after launch, and does not run out of AI projects with sufficient technical capabilities and out of the circle, the valuation of $AO will be greatly discounted.