Skip to main content

The Rise and Fall of the Artificial Superintelligence Alliance: A $120 Million Crypto Scandal

· 9 min read
Dora Noda
Dora Noda
Software Engineer

What happens when three of crypto's most ambitious AI projects merge to challenge OpenAI and Google—and then publicly implode over $120 million in missing tokens?

The Artificial Superintelligence Alliance was supposed to be Web3's answer to Big Tech's AI monopoly. A $7.5 billion merger between Fetch.ai, SingularityNET, and Ocean Protocol promised to build decentralized artificial general intelligence on blockchain infrastructure. Eighteen months later, Ocean Protocol has withdrawn, lawsuits are threatened, and the dream of democratized superintelligence faces its first existential test.

Yet beneath the drama lies a technical vision that could reshape how AI is built, owned, and governed. Here's the full story.

The Vision: Decentralized AGI as a Global Commons​

The premise behind the ASI Alliance was audacious: artificial superintelligence is too important to be controlled by a handful of corporations or governments.

Ben Goertzel—the researcher who coined the term "artificial general intelligence" in the early 2000s—has spent decades arguing that AGI development shouldn't happen inside proprietary labs. His thesis: if the next wave of AI innovation occurs on decentralized networks rather than behind closed doors, it fundamentally shifts the global power balance. Instead of an "AGI race" between nations and tech giants, superintelligence becomes a global commons—like the internet or Linux.

SingularityNET, founded by Goertzel, built a marketplace where AI developers could publish, monetize, and combine AI services. Fetch.ai created an autonomous agent network where AI programs could discover each other and transact. Ocean Protocol built decentralized data marketplaces so AI could access training data without centralized intermediaries.

The merger aimed to unify these capabilities into a single ecosystem:

  • SingularityNET contributes AGI research, the OpenCog framework, and the Hyperon cognitive architecture
  • Fetch.ai brings autonomous agent infrastructure and the ASI-1 Mini large language model
  • Ocean Protocol provides decentralized data infrastructure and privacy-preserving compute
  • CUDOS (which joined later) supplies distributed computing power

Together, they planned to build ASI:Chain—a dedicated blockchain where AI agents could operate autonomously, access data permissionlessly, and coordinate without human intervention.

The $7.5 Billion Merger​

In March 2024, the three projects announced their merger. By April, token holders had approved the consolidation with overwhelming support.

The mechanics were straightforward:

  • Fetch.ai's FET token served as the base
  • SingularityNET's AGIX converted at 0.433350 ASI per token
  • Ocean Protocol's OCEAN converted at 0.433226 ASI per token
  • Total ASI supply: 2.63 billion tokens

A governing council formed with Humayun Sheikh (Fetch.ai) as Chairman, Ben Goertzel (SingularityNET) as CEO, and Trent McConaghy and Bruce Pon from Ocean Protocol rounding out leadership.

The combined entity became the largest independent player in AI research and development outside of Big Tech. Market capitalization touched $7.5 billion at announcement.

ASI-1 Mini: The First Web3-Native LLM​

The Alliance's flagship product, ASI-1 Mini, launched as the first large language model designed specifically for autonomous AI agents operating on blockchain.

Unlike general-purpose LLMs, ASI-1 Mini is optimized for "agentic" workflows—autonomous decision-making, multi-step planning, and interaction with on-chain infrastructure. Key technical innovations include:

Mixture of Models (MoM): Rather than a monolithic architecture, ASI-1 Mini dynamically selects from multiple specialized models. A gating mechanism activates only the most relevant models for each task, improving efficiency and reducing compute costs.

Mixture of Agents (MoA): Autonomous agents with independent reasoning collaborate to solve complex tasks. A coordination layer handles task distribution, making the system resilient and adaptive.

Hardware Efficiency: ASI-1 Mini delivers enterprise-grade performance on just two GPUs—an eightfold reduction in hardware costs compared to equivalent models. Benchmarks show it matches or surpasses leading LLMs in specialized domains like medicine, history, and business reasoning.

Web3 Integration: The model connects directly to FET token infrastructure, enabling autonomous agents to pay for services, access data, and coordinate without human intervention.

Fetch.ai has already secured partnerships, including a deal with luxury automotive company Mansory to integrate ASI-1 Mini into vehicle systems. The vision extends to DeFi automation, supply chain optimization, and personalized AI assistants—all running on decentralized infrastructure.

The Hyperon Revolution​

While ASI-1 Mini addresses current AI agent needs, SingularityNET's Hyperon project tackles the harder problem: building systems that can actually reason, learn, and generalize like humans.

Hyperon Alpha 1, released in early 2025, introduced:

  • Scalable Atomspace: A distributed "brain space" that stores and processes knowledge across clusters
  • MeTTa Engine: A programming language designed for cognitive architectures, with plans to become the smart contract language for ASI:Chain
  • MORK: Fast, concurrent processing for the Atomspace framework
  • DAS: Distributed Atomspace scaling across multiple machines

These aren't just incremental improvements. Goertzel's team is building infrastructure for AGI systems that can share intermediate cognitive states—essentially allowing AI modules to "think together" rather than just exchange outputs.

The April 2025 presentation "The Ten Reckonings of AGI" addressed fundamental questions the technology raises: Who should control these systems? Will they exacerbate inequality? Can AI truly create, or merely imitate? What happens to human purpose when machines outperform us?

A global study accompanying the presentation found that 48% of Americans believe advanced AI is being developed "primarily by and for the benefit of a select few Big Tech corporations and governments, with insufficient public oversight." The majority (54%) expressed feeling a lack of control over how AI will affect their lives.

The $120 Million Scandal​

Then everything fell apart.

On October 9, 2025, Ocean Protocol Foundation announced its immediate withdrawal from the ASI Alliance. The exit triggered one of crypto's most bitter public disputes.

At the center: 286 million FET tokens worth approximately $120 million.

Fetch.ai CEO Humayun Sheikh alleged that on July 1, 2025, a multisig wallet linked to Ocean Protocol converted 661 million OCEAN tokens into 286 million FET—tokens meant for "community incentives" and "data mining."

Blockchain analytics told the story:

  • Between July 3-14, over 76 million FET moved to specific wallets
  • 21 million FET sent directly to Binance
  • 55 million FET transferred to a GSR Markets-linked address
  • 90 million FET moved to OTC provider GSR Markets
  • Remaining 196 million FET split across 30 newly created wallets
  • By mid-October, nearly 270 million FET had flowed to Binance or GSR

Sheikh's accusation was unequivocal: "If Ocean as a stand-alone project did this, it would be classed as a rug pull." He pledged to personally fund class-action lawsuits across multiple jurisdictions and offered a $250,000 bounty to identify OceanDAO multisig wallet signers.

Ocean Protocol pushed back. The Foundation maintained that the tokens were under stewardship of Ocean Expeditions, a Cayman Islands legal trust established independently in June 2025. They claimed Fetch.ai itself owed 110.9 million FET to the OCEAN:FET bridge contract—an obligation allegedly unfulfilled.

Ocean proposed waiving confidentiality over arbitration findings. They claimed Fetch.ai rejected this.

The market impact was severe. FET declined 92% from March highs. OCEAN fell 87% from its peak.

Resolution and Path Forward​

By late October 2025, both sides moved toward settlement. GeoStaking confirmed Ocean Protocol's willingness to return the disputed tokens pending a formal written proposal.

Roughly 270 million OCEAN tokens across 37,000 wallets remain unconverted. Ocean announced a buyback and burn initiative funded by project profits to restore market confidence.

The ASI Alliance continues, albeit diminished. SingularityNET and Fetch.ai remain committed to their shared infrastructure vision. CUDOS still provides compute resources.

But the episode revealed fundamental tensions in decentralized governance. When three independent foundations merge tokens but retain operational autonomy, who actually controls the treasury? What happens when strategic interests diverge?

What the Alliance Got Right​

Despite the drama, the ASI Alliance addressed real problems:

Fragmented AI infrastructure: Before the merger, each project operated independently. Developers needed to integrate multiple tokens, APIs, and governance systems. The unified ASI token simplified this.

Compute access: By adding CUDOS, the Alliance created a path to decentralized GPU resources—critical as AI training costs explode.

Data availability: Ocean Protocol's data marketplaces provided a mechanism for AI to access training data without relying on scraped internet content or proprietary datasets.

Economic alignment: A single token creates unified incentives. Token holders benefit from the entire ecosystem's success, not just one project's.

Alternative to Big Tech: With OpenAI, Google, and Anthropic controlling the AI frontier, a credible decentralized alternative has geopolitical significance.

The Road to AGI​

Ben Goertzel has consistently predicted AGI emergence within years, not decades. Whether that timeline is realistic, the infrastructure being built will matter.

ASI:Chain, the Alliance's planned dedicated blockchain, would allow:

  • AI agents to hold assets and execute transactions autonomously
  • Smart contracts written in MeTTa, optimized for cognitive architectures
  • Decentralized storage and compute integrated at the protocol level
  • Privacy-preserving data access for training and inference

If the remaining Alliance members execute this vision, they create a credible alternative to centralized AI development. The question is whether decentralized coordination can move fast enough to matter.

The Stakes​

The AI industry moves at unprecedented speed. OpenAI's valuation exceeds $150 billion. Google invests billions annually in AI research. Anthropic raised over $7 billion.

Against this, the ASI Alliance offers a different model: distributed ownership, open-source development, and governance by token holders rather than VCs or corporate boards.

The Ocean Protocol fallout demonstrates the fragility of such arrangements. But it also shows resilience—the core vision survived a major defection and public scandal.

For investors, developers, and AI researchers watching the space, the Alliance represents a high-risk bet on a fundamentally different future. One where superintelligence, if it arrives, belongs to no one and everyone.

Whether that future materializes depends on whether the remaining Alliance members can execute technically while navigating the governance challenges that nearly destroyed them.

Building AI agents that interact with blockchain infrastructure? BlockEden.xyz provides enterprise-grade RPC services and API infrastructure for the chains where AI-crypto convergence is happening. Explore our services to build autonomous systems on reliable foundations.

Share on Twitter