ASI Alliance Fission: From Tripartite Merger to Ocean Exit, Full-Stack AI Architecture Rebuilding and Practical Validation

In 2024, Fetch.ai, SingularityNET, and Ocean Protocol announced a merger to form the Artificial Superintelligence Alliance (ASI Alliance), which the market viewed as the most ambitious structural reorganization in the decentralized AI space. The three parties unified under the FET token, attempting to build a full-stack AI infrastructure spanning compute power, agents, and data. However, this narrative took a major turn in 2025—Ocean Protocol formally exited the alliance and resumed independent operations. As a result, the original full-stack closed-loop concept was broken.

As of May 19, 2026, the FET (ASI) token was quoted at $0.1930 on the Gate platform, with a market cap of approximately $435 million, and a drop of about 73.99% over the past year. The OCEAN token was quoted at $0.1214, with a market cap of approximately $76.39 million. The AGIX token has essentially completed its migration to FET, and market liquidity is extremely low. The core question thus emerges: has the full-stack closed loop formed by the merger of the three protocols already taken shape? Or is the market re-pricing the merger itself?



## Three Turning Points Reshape the Alliance Architecture

At the factual level, the formation and splitting of the ASI Alliance follows a clear timeline.

The first turning point: the merger of the three parties. In March 2024, Fetch.ai, SingularityNET, and Ocean Protocol jointly announced a merger plan. The three parties’ tokens—FET, AGIX, and OCEAN—would be unified and integrated into the FET token ecosystem, with the brand upgraded to ASI. The conversion ratios were set at 0.433350 FET per AGIX token and 0.433226 FET per OCEAN token, and the alliance issued approximately 600 million additional FET tokens for this purpose.

The second turning point: CUDOS joins. In September 2024, the decentralized computing project CUDOS officially joined the alliance. Its network and token supply were integrated into the ASI ecosystem, taking on the role of providing compute infrastructure. The alliance architecture was thus expanded into a four-pillar structure: an agent layer, a service layer, a compute power layer, and a data layer.

The third turning point: Ocean exits. On October 9, 2025, the Ocean Protocol Foundation announced it would immediately exit the ASI Alliance and withdraw its designated directors from the alliance company registered in Singapore. After the exit, the alliance consisted of Fetch.ai, SingularityNET, and CUDOS, leaving a structural gap in the data layer. At the time of exit, approximately 81% of the OCEAN supply had been converted, and the remaining roughly 270 million OCEAN was distributed across 37,334 addresses.

Ocean attributed its exit to seeking independent funding and governance space, while also alleging governance misconduct within the alliance, including financial decisions made without adequate consultation and token management disputes. It is necessary to make a clear distinction: all the above allegations are unilateral statements by the respective parties and have not yet been verified by an independent third party.

## Compute Power and Agents: Can the Alliance’s Triangular Closed Loop Be Self-Consistent?

After Ocean’s exit, the ASI Alliance’s technology stack was redefined as a three-pronged structure, forming a triangular closed loop of “agents—compute power—services.”

In the agent layer, Fetch.ai’s Agentverse platform and the ASI Create tool together form an interface for creating and deploying autonomous agents. The trend toward large-scale agent economies is accelerating: as of April 20, 2026, the number of on-chain AI agent deployments on BNB Chain has exceeded 150,000, while at the beginning of that year there were only 337 across the entire network—an increase of more than 43,750%.

In the compute power layer, CUDOS provides decentralized GPU infrastructure through the ASI Cloud platform, supporting AI inference and training workloads. The alliance also plans to develop the ASI Chain testnet as a dedicated blockchain layer for agent coordination and cross-chain operations.

In the model and service layer, the Web3-native large language model ASI-1 Mini has been released, and SingularityNET continues to operate its AI service marketplace.

This triangular closed loop can support an economic cycle in which agents discover demand, call on compute resources to execute tasks, and exchange model capabilities through the service marketplace. However, the core gap in the loop lies in the acquisition of training data. Agents need data to learn and optimize, and service marketplaces need data to train models—yet after Ocean’s exit, the alliance lacks native decentralized data market support, exposing structural gaps at the architectural level.

## Data Independence: Ocean Protocol’s Privacy Computing Path

After Ocean Protocol exited the alliance, it accelerated the productization of its technology, forming an independent business closed loop of “data—privacy computing—decentralized compute.”

Its core technical architecture revolves around three tools: Data NFT, Compute-to-Data, and Ocean Nodes. Data NFT enables on-chain rights confirmation and version management for data assets. Compute-to-Data allows AI models to run calculations directly in the environment where the data resides, returning only results without exposing the original data—this has key application value in industries with strict privacy and compliance requirements, such as healthcare and finance. The integration of secure multi-party computation further strengthens the capability to train AI models on sensitive data without compromising data privacy.

On the computing resource layer, Ocean Network has handled more than 3,000 compute tasks, providing developers with instant access to NVIDIA H200 GPUs at a pay-as-used rate of about $2.16 per hour. Since Ocean Nodes went live on August 15, 2024, more than 1.7 million nodes have been installed in over 70 countries and regions. Ocean Enterprise Collective, in partnership with 12 organizations covering 8 countries and 9 industries, promotes compliance-oriented enterprise use-case applications.

At the industry macro level, according to a Slator report, the global Data-for-AI market is expected to reach a scale of about $9.3 billion in 2026, and is projected to grow to $21.5 billion by 2031, with an estimated CAGR of approximately 18%. Market demand for data verifiability and compliance is rising. Ocean’s Data NFT and Compute-to-Data technologies provide exactly the kind of on-chain auditable solutions. After exiting the alliance, Ocean announced that it would use part of the profits from technology-derived projects to buy back and burn OCEAN tokens, establishing a persistent deflationary mechanism.

## Controversy Focus: The Governance Game Behind the Exit

The market has formed three levels of public discussion around Ocean’s exit from the ASI Alliance.

At the community level, there is clear polarization. Ocean supporters welcome the project regaining independence, believing that independent governance helps product deployment and token value capture. After the exit news was released, OCEAN saw an approximately 30% short-term increase. Opponents, however, worry that weakened alliance data capabilities will affect their long-term competitiveness.

At the governance level, both sides insist on their own positions. In its exit statement, Ocean alleged that SingularityNET carried out reckless financial operations after the merger and consumed significant market liquidity, including issuing $100 million in tokens and maintaining monthly spending of about $6 million. Fetch.ai’s founder was accused of failing to implement decentralized governance principles and attempting to force Ocean to convert all its independent treasury assets into FET. Ocean said it had requested to exit as early as April 2024 but was prevented by legal threats, and ultimately was forced to exit through legal means after its partner attempted to unilaterally shut down the token exchange bridge in August 2025. Ocean also pointed out that the price of the FET token fell 93% from its peak, mainly due to large-scale sell-offs by its partner and failed high-risk trades, rather than its own exit. The alliance denied the relevant allegations. It must be emphasized again: these are all unilateral statements by the respective parties and have not been verified by an independent third party.

At the industry reflection level, this incident has triggered deeper discussions about the feasibility of governance for decentralized alliances: when participants retain independent governance rights and fiscal sovereignty, does a unified token framework have sufficient incentives to sustain cooperation? And when technical routes and business models differ inherently, can token mergers truly create synergistic value?

## A Market Mirror: Token Price Movements and Value Reassessment

As of May 19, 2026, Gate platform market data shows:

FET (ASI) is quoted at $0.1930, with a 24-hour increase of 1.79%, a market cap of approximately $435 million, a circulating supply of approximately 2.25 billion tokens, and a market rank of 132. Over the past 7 days, it is down 14.38%; over 30 days, down 8.58%; over 90 days, up 16.75%; over one year, down 73.99%.

OCEAN is quoted at $0.1214, with a 24-hour increase of 2.26%, a market cap of approximately $76.39 million, a circulating supply of approximately 629 million tokens, and a total supply of 1.41 billion. AGIX is quoted at $0.0007673, with a market cap of only about $71,000, and has largely exited mainstream trading.

In addition, there is a token called Artificial Superintelligence (ASI) (ranked 6,559), quoted at $0.0009000, with a 24-hour decrease of 9.09% and a market cap of about $253,000. This token is different from the official ASI Alliance token FET/ASI, so investors need to pay attention to the distinction.

From the perspective of market structure, while FET is still one of the assets with a relatively higher market cap in the AI crypto sector, it has retreated significantly compared with its high point at the beginning of the merger. Although OCEAN’s circulating market cap is far lower than FET’s, the technology product supporting it—data markets and Compute-to-Data—has a commercial value logic that stands independently of the alliance narrative.

## Closed-Loop Review: The Real-World Break in the Full-Stack Narrative

Based on the analysis above, here is a judgment on the core question.

The original three-layer closed loop concept has collapsed due to Ocean’s exit. The current ASI Alliance technical architecture is a triangular closed loop of “agents—compute power—services,” and there is a gap in the data layer. Ocean Protocol, meanwhile, has independently formed a business closed loop of “data—privacy computing—decentralized compute.” Although the two are not technically incapable of interoperability—Fetch.ai agents could theoretically call Ocean’s data services—their token economies, governance mechanisms, and business collaboration have already diverged.

This split has dual meanings. On the positive side, it indicates that different functional layers in decentralized AI require differentiated governance and business models, and forcing unification could lead to incentive distortions. On the negative side, the split weakens the network effects of decentralized AI when competing against centralized giants; separating agents and data may reduce overall synergy.

In the future, evolution may follow two paths: first, the ASI Alliance fills the data-layer gap by building or integrating other data protocols, reconstructing a complete closed loop; second, Ocean Protocol, by leveraging privacy computing and data tokenization, attracts more enterprise users and forms an ecosystem that develops in parallel with the alliance. The two paths are not mutually exclusive, but at the level of token economies, it is difficult to integrate again.

## Evolutionary Forecast: Three Possible Futures Under the Three-Part Pattern

Based on the current structure, the decentralized AI space is developing into a landscape characterized by narrative divergence and tiered specialization.

First, the full-stack closed-loop narrative is being replaced by layered specialization. The ASI Alliance focuses on agent economies and compute coordination, while Ocean focuses on data markets and privacy computing. Each targets different value segments along the AI industry chain. This divergence is structurally similar to the evolution of internet infrastructure from vertical integration toward horizontal layering.

Second, the independent value of data tokenization is becoming prominent. The global AI training data market is expanding to the scale of $9.3 billion, and centralized data procurement alternative solutions are being considered by more enterprises. Ocean’s Data NFT and Compute-to-Data provide chain-based, auditable data provenance and privacy-preserving training solutions, aligning with regulatory directions such as the EU AI Act.

Third, multi-scenario evolution is possible. Under an optimistic scenario, scaled deployment of the alliance’s agent economy and enterprise adoption of Ocean’s data market advance in parallel. Through API-level interoperability, they form loosely coupled collaboration at the application layer, benefiting the decentralized AI ecosystem as a whole. Under a conservative scenario, both sides develop independently but face user growth bottlenecks: the agent economy remains at the experimental stage, and data markets struggle to break through the high barriers of enterprise procurement. Under a risk scenario, centralized AI giants continue to squeeze market space, posing fundamental challenges to the commercial viability of decentralized alternatives and putting further pressure on token values.

## Conclusion

The ASI Alliance formed by the merger of three protocols was once the grandest narrative in decentralized AI—a full-stack closed loop spanning compute, agents, and data. However, Ocean Protocol’s exit has pushed this narrative into a real-world test.

The current structure is clear and distinguishable: the ASI Alliance, centered on Fetch.ai, SingularityNET, and CUDOS, has built a triangular closed loop of “agents—services—compute.” Ocean Protocol has independently developed a commercial path focused on “data—privacy computing—decentralized compute.” While the two are technically interoperable, each is exploring its own independent business model and seeking value capture through different token mechanisms.

Has the full-stack closed loop already formed? The original unified closed loop has already split, but two more specialized and deeper independent closed loops are being built in parallel. The future of decentralized AI may not depend on a single grand narrative of a unified alliance, but on whether different functional layers validate product-market fit in their respective tracks. This is both an inevitable consequence of long-standing division and a structural signal that the industry is maturing.