In 2024, three of the biggest names in blockchain-based AI — Fetch.ai, SingularityNET, and Ocean Protocol — merged into a single entity called the Artificial Superintelligence Alliance. The goal: build a path to artificial superintelligence that isn't owned by a handful of US tech companies. Here's what the merger actually produced, and whether the bet makes sense.
The ASI Alliance is not a startup. It's a coalition of three pre-existing projects with distinct technology stacks, communities, and token economies — now operating under a shared governance structure and a merged token.
Autonomous AI agent network with its own blockchain. Agents negotiate with each other, find services, and execute tasks without human direction. The most technically active of the three projects.
Decentralized marketplace for AI services. Developers publish AI agents and models; users pay in $ASI to access them. Also home to OpenCog Hyperon — the alliance's AGI architecture research project.
Decentralized data marketplace. Allows data owners to monetize datasets without giving up control. Positioned as the data supply layer for AI systems trained on decentralized data.
Together, the three are meant to form a stack: Ocean supplies the data, SingularityNET hosts the AI services, and Fetch.ai runs the autonomous agent layer that ties it all together. Whether that stack actually integrates smoothly in practice is a fair question — these are three distinct codebases with three distinct communities that now share governance but not a unified product.
The $ASI token functions as both a governance token and a utility token within the ecosystem. Holders vote on alliance decisions proportional to their stake. The token is also used to pay for services on SingularityNET's marketplace, stake within Fetch.ai's agent network, and access data products on Ocean Protocol. OCEAN holders converted at a similar discounted rate (approximately 0.433:1 OCEAN per $ASI) set by governance vote.
The token economy is designed so that demand for AI services across all three platforms flows into $ASI demand. As the alliance's agent networks are used more, more $ASI is needed to pay for services — in theory creating a link between actual AI utility and token value. In practice, like most crypto-AI tokens, $ASI price tracks broader crypto market sentiment as much as underlying AI activity.
The ASI Alliance operates on a fundamentally different model from the centralized labs racing to build AGI. The differences matter:
| Dimension | ASI Alliance | OpenAI / Anthropic / Google |
|---|---|---|
| Structure | Decentralized token governance, Foundation | Private companies (some with non-profit elements) |
| Funding model | Token sales, ecosystem fees, grants | VC, corporate partnerships ($billions) |
| AI approach | Decentralized agent networks, open-source tools, data markets | Large foundation models trained in-house on proprietary infrastructure |
| Current frontier | Agent coordination, data monetization, OpenCog Hyperon research | GPT-6, Claude 4, Gemini Ultra — state-of-the-art LLMs |
| Path to ASI | Emergent from a network of cooperating AI agents and open AI services | Scaling up foundation models + post-training alignment |
| Compute access | Relies on distributed/rented compute; no proprietary supercomputers | Own or partner-controlled GPU clusters (100,000+ H100 scale) |
| Transparency | Open-source codebases, on-chain governance | Variable — OpenAI has moved less open; Anthropic publishes safety research |
The honest answer: the centralized labs are currently ahead on raw capability. GPT-6 Astra scoring 98.6% on ARC-AGI-3 is not something the ASI Alliance ecosystem has matched or approached. What the alliance offers is a different answer to the question of who controls the path to ASI — and whether it should be owned by a handful of US tech companies at all.
Ben Goertzel's thesis: The alliance's Chief AI Scientist argues that centralized AGI built by profit-driven corporations poses existential alignment risks. His OpenCog Hyperon architecture is designed for a "democratic" AGI where no single entity — including the ASI Alliance itself — can dominate. Whether that design philosophy survives contact with the compute realities of frontier AI remains to be seen.
Beneath the ASI branding, the alliance's most concrete products are:
A natural-language interface for finding and coordinating AI agents. Think of it as a search engine where the results are autonomous agents that can execute tasks — booking travel, running data analysis, managing workflows. It's the most consumer-facing product in the ecosystem and has real users.
A deployment platform for autonomous AI agents. Developers build agents that can be discovered, hired, and paid by other agents or users. The agent-to-agent economy is small but functional — micropayments in $ASI, automated task negotiation, persistent agents running without human oversight.
A catalog of AI services — image recognition, NLP, predictive analytics — that developers publish and users pay to access. It predates the merger and has hundreds of listed services. Quality varies significantly; it's more a directory than a curated platform.
Goertzel's long-running AGI architecture project, now under the alliance umbrella. Hyperon is a cognitive architecture built on a "metagraph" knowledge representation — more symbolic AI than neural, though designed to integrate both. It's research-stage, not production. Whether it can compete with transformer-based approaches is genuinely uncertain.
Allows data owners to publish and monetize datasets via "datatokens" — NFT-like access tokens. Buyers purchase access without the data leaving the owner's control (compute-to-data). It's a real product with real volume, particularly in healthcare and enterprise data.
The ASI Alliance's bet is that the next breakthrough in AI won't come from one company training one model on one supercomputer — it'll emerge from a network of specialized agents, open data, and distributed compute cooperating at scale. That's not a crazy bet. It's closer to how the internet itself developed.
But frontier AI has so far rewarded concentration: more compute, more data, more engineering talent in one place produces better models. The companies that got to the AGI-adjacent benchmarks first — OpenAI, Google, Anthropic — did it through centralized, well-funded research. The ASI Alliance's distributed model has different strengths: open governance, no single point of control, incentive alignment through token economics. But it's competing on a different axis than raw capability.
Whether that matters for reaching ASI is the core unanswered question. If ASI requires the kind of compute concentration that only centralized entities can marshal, the alliance's model may produce excellent infrastructure without reaching the ultimate goal. If ASI emerges from coordination and cooperation rather than brute-force scaling, the alliance could matter a great deal.
The honest answer: nobody knows. Both bets could be right on their own terms — centralized labs get to AGI first, the ASI Alliance builds the open infrastructure layer that runs on top. Or the bets could be mutually exclusive. That's what makes it worth watching.
The ASI Alliance is a decentralized AI ecosystem formed by the 2024 merger of Fetch.ai, SingularityNET, and Ocean Protocol. It aims to build open, distributed AI infrastructure as an alternative to the centralized lab model. It is governed by a combined $ASI token and a Foundation structure, not a traditional company.
$ASI is the unified governance and utility token of the ASI Alliance, created by merging the FET (Fetch.ai), AGIX (SingularityNET), and OCEAN (Ocean Protocol) tokens at fixed ratios. Holders use $ASI to pay for AI services on the SingularityNET marketplace, stake in Fetch.ai's agent network, access Ocean data products, and vote on governance decisions. It trades on major exchanges under the ticker $ASI.
Ben Goertzel is the founder of SingularityNET and the Chief AI Scientist of the ASI Alliance. He is one of the most prominent public advocates for AGI via decentralized systems. He created the OpenCog architecture in the early 2000s — one of the longest-running AGI research projects — and has been working on OpenCog Hyperon, the alliance's next-generation AGI architecture, as a successor. He is a frequent speaker at AI and transhumanist conferences and a vocal critic of centralized AI development.
This is a financial question we can't answer — we're not financial advisors and $ASI is a speculative crypto asset. What we can say: the token's price tracks both broader crypto market sentiment and actual AI service demand in the ecosystem. The alliance's products are real and have real users, but the centralized labs currently have a significant capability lead. Anyone considering $ASI should understand both the genuine vision and the competitive reality.
They're pursuing similar goals through radically different models. OpenAI is a centralized company training large foundation models with billions in funding, aiming for AGI through scaling and post-training alignment. The ASI Alliance is a decentralized token-governed ecosystem building open AI agent infrastructure, believing AGI and ASI will emerge from networked cooperation rather than centralized scaling. On current benchmarks, OpenAI's models are more capable. On governance, openness, and decentralization, the ASI Alliance takes a different approach by design.
DeltaV is Fetch.ai's natural language interface for finding and coordinating AI agents. Users can request tasks in plain English — "find me the cheapest flight from Bangkok to Nicosia" or "summarize this dataset" — and DeltaV routes the request to the appropriate agents in the Agentverse ecosystem, which execute the task and return results. It's the most consumer-facing product of the ASI Alliance and has real adoption among developers building agent-based applications.