7*24

Hot topic

Market

Activities

On April 21, during the roundtable discussion 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions', MuleRun CTO Shu Junliang discussed the topic of 'Agent Moat'. He stated that the traditional AI technology moat is being rapidly weakened, primarily due to the accelerated convergence of model capabilities and exponential improvements in development efficiency. He pointed out that the performance gap between mainstream large models is narrowing quickly, especially over the past year, where the capability gap between domestic and international models has significantly converged. Additionally, with the explosive enhancement of code generation capabilities, software development efficiency has greatly increased—features that previously took weeks or even months to complete can now be achieved in just a few days. This means that both Agent frameworks and specific functional modules can be quickly reused or replicated through open-source solutions, making the 'functional moat' at the product level increasingly fragile. In this context, Shu Junliang believes that the core competitiveness of Agents in the future will mainly be reflected in two aspects: first, the ability for continuous high-frequency iteration, meaning whether the team can maintain a leading product update speed over the long term; and second, data advantages, including exclusive data resources and user accumulated data. On one hand, platforms with unique data acquisition capabilities (such as data from specific industries or regions) will form natural barriers; on the other hand, the behavioral and memory data accumulated from long-term user engagement on the platform will become key assets that are difficult to migrate, further enhancing user stickiness and product competitiveness. He concluded that as the trend of 'equalization' in models and technologies continues, the Agent moat is shifting from 'technical capability' to a comprehensive competition of 'data assets and execution efficiency'.

[MuleRun CTO Shu Junliang: Model Gaps Rapidly Converging, Agent Moat Shifts to 'Speed + Data']On April 21, during the roundtable discussion 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions', MuleRun CTO Shu Junliang discussed the topic of 'Agent Moat'. He stated that the traditional AI technology moat is being rapidly weakened, primarily due to the accelerated convergence of model capabilities and exponential improvements in development efficiency. He pointed out that the performance gap between mainstream large models is narrowing quickly, especially over the past year, where the capability gap between domestic and international models has significantly converged. Additionally, with the explosive enhancement of code generation capabilities, software development efficiency has greatly increased—features that previously took weeks or even months to complete can now be achieved in just a few days. This means that both Agent frameworks and specific functional modules can be quickly reused or replicated through open-source solutions, making the 'functional moat' at the product level increasingly fragile. In this context, Shu Junliang believes that the core competitiveness of Agents in the future will mainly be reflected in two aspects: first, the ability for continuous high-frequency iteration, meaning whether the team can maintain a leading product update speed over the long term; and second, data advantages, including exclusive data resources and user accumulated data. On one hand, platforms with unique data acquisition capabilities (such as data from specific industries or regions) will form natural barriers; on the other hand, the behavioral and memory data accumulated from long-term user engagement on the platform will become key assets that are difficult to migrate, further enhancing user stickiness and product competitiveness. He concluded that as the trend of 'equalization' in models and technologies continues, the Agent moat is shifting from 'technical capability' to a comprehensive competition of 'data assets and execution efficiency'.

MuleRun CTO Shu Junliang: Model Gaps Rapidly Converging, Agent Moat Shifts to 'Speed + Data'

On April 21, during the roundtable discussion "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions," Xu Jialong, Associate Vice President of Hong Kong University of Science and Technology, commented on the "moat of AI agents." He noted that there are differences in the model training paths and technical systems behind different AI agents, leading to significant variations in actual user experience. Recently, some new models and tools have shown superior performance in generation quality and execution efficiency, even demonstrating higher ceilings in development output. However, he pointed out that at the current stage, these differences have not yet formed a decisive generational gap, being closer to "efficiency improvement" rather than a "paradigm breakthrough." In other words, competition among agents is still in a rapid evolution phase, with no stable and insurmountable technical barriers emerging yet. The iteration pace of current AI agents and large models is extremely fast, with new products or capabilities appearing almost weekly, driving the industry forward. However, from the perspective of practical use and business decision-making, whether there is a need to continuously and frequently follow these changes still requires careful evaluation.

[HKUST's Xu Jialong: The Moat of AI Agents Is Not Yet Solidified, Model Differences Reflect Efficiency Rather Than Disruptive Breakthroughs]On April 21, during the roundtable discussion "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions," Xu Jialong, Associate Vice President of Hong Kong University of Science and Technology, commented on the "moat of AI agents." He noted that there are differences in the model training paths and technical systems behind different AI agents, leading to significant variations in actual user experience. Recently, some new models and tools have shown superior performance in generation quality and execution efficiency, even demonstrating higher ceilings in development output. However, he pointed out that at the current stage, these differences have not yet formed a decisive generational gap, being closer to "efficiency improvement" rather than a "paradigm breakthrough." In other words, competition among agents is still in a rapid evolution phase, with no stable and insurmountable technical barriers emerging yet. The iteration pace of current AI agents and large models is extremely fast, with new products or capabilities appearing almost weekly, driving the industry forward. However, from the perspective of practical use and business decision-making, whether there is a need to continuously and frequently follow these changes still requires careful evaluation.

HKUST's Xu Jialong: The Moat of AI Agents Is Not Yet Solidified, Model Differences Reflect Efficiency Rather Than Disruptive Breakthroughs

Crypto analyst Ai disclosed data on platform X, revealing that in the first quarter of 2026, AI companies raised approximately $242 billion, accounting for about 80% of the total global venture capital, setting a historical record. However, the funding is highly concentrated in a few 'super funding rounds,' such as OpenAI's $122 billion, Anthropic's $30 billion, xAI's $20 billion, and Waymo's $16 billion. Meanwhile, enterprise-level AI spending continues to rise. Gartner predicts that global AI spending will reach $2.52 trillion in 2026, a year-on-year increase of 44%, with funds rapidly shifting from application layers to infrastructure and computing power construction.

[Analyst: AI Companies Raise $242 Billion in Q1, Setting Historical Record]Crypto analyst Ai disclosed data on platform X, revealing that in the first quarter of 2026, AI companies raised approximately $242 billion, accounting for about 80% of the total global venture capital, setting a historical record. However, the funding is highly concentrated in a few 'super funding rounds,' such as OpenAI's $122 billion, Anthropic's $30 billion, xAI's $20 billion, and Waymo's $16 billion. Meanwhile, enterprise-level AI spending continues to rise. Gartner predicts that global AI spending will reach $2.52 trillion in 2026, a year-on-year increase of 44%, with funds rapidly shifting from application layers to infrastructure and computing power construction.

Analyst: AI Companies Raise $242 Billion in Q1, Setting Historical Record

On April 21, during the roundtable discussion 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions', Box from the Monad Foundation's Greater China Developer Relations addressed the topic of 'Agent Moats'. He stated that the current technical threshold for AI Agents is relatively low, and their so-called 'moat' does not primarily come from the products themselves, but rather depends more on the founders' or teams' understanding of AI, their control capabilities, and external influence. He pointed out that many Agent products at this stage essentially rely on open-source code and existing tools to be quickly assembled, significantly compressing development cycles, with the phenomenon of 'building a product in just a few hours' not being uncommon. However, this also means that product forms are easily replicable; once cloud vendors or large model vendors get involved, they can often replicate or even surpass these products in a very short time, causing early-stage startups to rapidly lose their competitive edge. Box further emphasized that in this environment, the survival logic of Agent projects is changing: on one hand, they need to maintain a lead through high-frequency iterations, and on the other hand, they must expand market influence and user awareness as much as possible to establish a first-mover advantage before the industry fully matures. He believes that if the product update speed is insufficient and lacks dissemination capability, once large companies take notice and quickly follow up, its lifecycle could be significantly compressed or even lead to immediate exit.

[Monad Foundation's Box: The Competitive Edge of Agents Lies in 'Influence + Iteration Speed', Products Easily Replicable]On April 21, during the roundtable discussion 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions', Box from the Monad Foundation's Greater China Developer Relations addressed the topic of 'Agent Moats'. He stated that the current technical threshold for AI Agents is relatively low, and their so-called 'moat' does not primarily come from the products themselves, but rather depends more on the founders' or teams' understanding of AI, their control capabilities, and external influence. He pointed out that many Agent products at this stage essentially rely on open-source code and existing tools to be quickly assembled, significantly compressing development cycles, with the phenomenon of 'building a product in just a few hours' not being uncommon. However, this also means that product forms are easily replicable; once cloud vendors or large model vendors get involved, they can often replicate or even surpass these products in a very short time, causing early-stage startups to rapidly lose their competitive edge. Box further emphasized that in this environment, the survival logic of Agent projects is changing: on one hand, they need to maintain a lead through high-frequency iterations, and on the other hand, they must expand market influence and user awareness as much as possible to establish a first-mover advantage before the industry fully matures. He believes that if the product update speed is insufficient and lacks dissemination capability, once large companies take notice and quickly follow up, its lifecycle could be significantly compressed or even lead to immediate exit.

Monad Foundation's Box: The Competitive Edge of Agents Lies in 'Influence + Iteration Speed', Products Easily Replicable

On April 21, according to The Information: Musk significantly increased his stake in SpaceX last year, spending $1.4 billion.

[Musk Acquired $1.4 Billion in SpaceX Stock Last Year]On April 21, according to The Information: Musk significantly increased his stake in SpaceX last year, spending $1.4 billion.

Musk Acquired $1.4 Billion in SpaceX Stock Last Year

On April 21, according to Nikkei News: The Bank of Japan will maintain interest rates unchanged in April.

[Bank of Japan to Maintain Interest Rates in April]On April 21, according to Nikkei News: The Bank of Japan will maintain interest rates unchanged in April.

Bank of Japan to Maintain Interest Rates in April

On April 21, during the roundtable discussion "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions," Christian, the founder of Infini, discussed the topic of the "Agent Moat." He stated that from the perspective of product and growth, the core competitiveness of AI still primarily depends on the language processing capabilities of large models, particularly the degree of humanization in content output. He pointed out that compared to the earlier models, which tended to provide mechanical and templated responses, current users are more concerned with whether AI can deliver expressions and content quality that closely resemble real human conversations. This directly impacts its practical value in external communication and team collaboration. Christian mentioned that as model capabilities continue to evolve, basic information processing abilities are becoming increasingly homogeneous. Once large models can accomplish about 90% of standardized tasks, solely relying on the models themselves will struggle to create long-term barriers. The competition among AI Agents is gradually shifting from model capabilities to specific application scenarios and workflow design. In the future, truly differentiated capabilities will no longer be limited to simple information processing or desktop automation, but will involve executing higher-level tasks and providing decision support through a deep understanding of business processes in complex scenarios such as finance and trading.

[Christian, Founder of Infini: Humanized Expression in Large Models Becomes a Core Barrier, Agent Moat Shifts from Models to Application Layer]On April 21, during the roundtable discussion "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions," Christian, the founder of Infini, discussed the topic of the "Agent Moat." He stated that from the perspective of product and growth, the core competitiveness of AI still primarily depends on the language processing capabilities of large models, particularly the degree of humanization in content output. He pointed out that compared to the earlier models, which tended to provide mechanical and templated responses, current users are more concerned with whether AI can deliver expressions and content quality that closely resemble real human conversations. This directly impacts its practical value in external communication and team collaboration. Christian mentioned that as model capabilities continue to evolve, basic information processing abilities are becoming increasingly homogeneous. Once large models can accomplish about 90% of standardized tasks, solely relying on the models themselves will struggle to create long-term barriers. The competition among AI Agents is gradually shifting from model capabilities to specific application scenarios and workflow design. In the future, truly differentiated capabilities will no longer be limited to simple information processing or desktop automation, but will involve executing higher-level tasks and providing decision support through a deep understanding of business processes in complex scenarios such as finance and trading.

Christian, Founder of Infini: Humanized Expression in Large Models Becomes a Core Barrier, Agent Moat Shifts from Models to Application Layer

On April 21, Brad Bao, Head of AI Growth at Cobo, shared at the offline event themed 'Decoding Web 4.0: When AI Agents Take Over On-Chain Authority' that Cobo is restructuring the funding control logic of Agents through a structured authority system. In its products, the relationship between users and Agents is no longer a simple wallet authorization but is defined by an executable 'contract (Policy/Agreement)' that outlines the boundaries of authority. This contract includes at least four core elements: Intent, Path, Rules, and Termination conditions, thereby upgrading the original 'all or nothing' authority control to a negotiable, auditable, and revocable fine-grained governance mechanism. On the execution level, Cobo integrates MPC (Multi-Party Computation) technology to ensure that even if an Agent exhibits abnormal behavior, the system is attacked, or risks such as prompt injection occur, the funds remain under user control, and no single entity can bypass the constraints to independently transfer assets. Brad Bao emphasized that under this framework, Agents can gain sufficient freedom of action to complete complex tasks, but their behavior is always confined within verifiable boundaries. He stated that this system essentially builds a layer of 'trust and insurance infrastructure' within the Agent economy, achieving controllable delegation of Agent behavior through the combination of 'contractual authority + secure computation,' ultimately enhancing automation efficiency while ensuring user asset security and outcome certainty.

[Cobo Proposes 'Contractual Authority + MPC' Framework: Building a Controllable Trust Layer for the Agent Economy]On April 21, Brad Bao, Head of AI Growth at Cobo, shared at the offline event themed 'Decoding Web 4.0: When AI Agents Take Over On-Chain Authority' that Cobo is restructuring the funding control logic of Agents through a structured authority system. In its products, the relationship between users and Agents is no longer a simple wallet authorization but is defined by an executable 'contract (Policy/Agreement)' that outlines the boundaries of authority. This contract includes at least four core elements: Intent, Path, Rules, and Termination conditions, thereby upgrading the original 'all or nothing' authority control to a negotiable, auditable, and revocable fine-grained governance mechanism. On the execution level, Cobo integrates MPC (Multi-Party Computation) technology to ensure that even if an Agent exhibits abnormal behavior, the system is attacked, or risks such as prompt injection occur, the funds remain under user control, and no single entity can bypass the constraints to independently transfer assets. Brad Bao emphasized that under this framework, Agents can gain sufficient freedom of action to complete complex tasks, but their behavior is always confined within verifiable boundaries. He stated that this system essentially builds a layer of 'trust and insurance infrastructure' within the Agent economy, achieving controllable delegation of Agent behavior through the combination of 'contractual authority + secure computation,' ultimately enhancing automation efficiency while ensuring user asset security and outcome certainty.

Cobo Proposes 'Contractual Authority + MPC' Framework: Building a Controllable Trust Layer for the Agent Economy

On April 21, Brad Bao, the Growth Head of Cobo AI, shared at the offline event themed 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions' that the involvement of AI Agents in fund allocation and asset management has transitioned from concept to reality. The question of whether Agents will manage money is no longer relevant. As more Agents directly connect to wallets to execute transactions, the industry has begun to see risks such as funds not arriving as expected. In response, the team continuously records issues and risks faced by Agents in real transactions through an experimental system, promoting the improvement of relevant mechanisms. Brad Bao pointed out that the current core contradiction in the industry is: if Agents lack autonomy, their value will be no different from traditional automated trading tools; however, if they lack verifiable certainty, humans cannot confidently entrust fund management to Agents. This tension between 'autonomy vs certainty' is becoming a key bottleneck for AI Agents in advancing towards on-chain financial infrastructure.

[Cobo AI Growth Head Brad Bao: AI Agents Have Started 'Managing Money', Core Challenges Shift to Trust and Certainty]On April 21, Brad Bao, the Growth Head of Cobo AI, shared at the offline event themed 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions' that the involvement of AI Agents in fund allocation and asset management has transitioned from concept to reality. The question of whether Agents will manage money is no longer relevant. As more Agents directly connect to wallets to execute transactions, the industry has begun to see risks such as funds not arriving as expected. In response, the team continuously records issues and risks faced by Agents in real transactions through an experimental system, promoting the improvement of relevant mechanisms. Brad Bao pointed out that the current core contradiction in the industry is: if Agents lack autonomy, their value will be no different from traditional automated trading tools; however, if they lack verifiable certainty, humans cannot confidently entrust fund management to Agents. This tension between 'autonomy vs certainty' is becoming a key bottleneck for AI Agents in advancing towards on-chain financial infrastructure.

Cobo AI Growth Head Brad Bao: AI Agents Have Started 'Managing Money', Core Challenges Shift to Trust and Certainty

Sasha Ivanov, founder of Waves, announced on Twitter that the Wave Level 2 network white paper has been released.

[Founder of Waves: Wave L2 white paper has been released]Sasha Ivanov, founder of Waves, announced on Twitter that the Wave Level 2 network white paper has been released.

Founder of Waves: Wave L2 white paper has been released

Bitcoin Created By The CIA? Chinese Professor Jiang Xueqin Makes Bold Claim

BIT Research: Five Signals Flash Simultaneously, Has the Bitcoin Bear Market Ended?

Bitcoin Recovery Fails To Lift Market Sentiment From Extreme Fear

Nic Carter Says Bitcoin Has 3 Ways To Handle Satoshi’s Coins

Crypto And Financial Industry Giants Reveal What X Money Launch Means

Analyst Exposes Bitcoin Market Maker Buy Strategy, Shows What Happens When Accumulation Ends

Conversation with Bitwise Advisor: From K-Shaped Economy to AI Taking Jobs, How Can Bitcoin Save the Younger Generation?

Analyst Reveals The Chances Of Bitcoin Price Crashing Again

Bitcoin Miners Selling Nears Exhaustion – What Comes Next

Crypto Regulation: Polish Parliament Fails To Overturn Presidential Veto Again

Article

JPYC Completes Additional Financing of Approximately 2.8 Billion Yen with Participation from Metaplanet

Founder of Waves: Wave L2 white paper has been released

Descargo de responsabilidad：El contenido anterior no representa las posiciones de HTX.，HTX no ofrece ninguna recomendación de trading.。

Todos los comentarios0Lo más recientePopular

Temas populares