Exploring ASI Alliance: The Web3 Movement to Decentralise AI

Artificial intelligence is quickly becoming one of the most transformative technologies of our time. Yet, as AI systems grow more powerful, concerns around centralisation, transparency, and control continue to mount. Today, most advanced AI development is concentrated in the hands of a few well-capitalised corporations—raising important questions about who benefits, who decides, and who ultimately owns the intelligence being built.

In response to this, a group of Web3 projects has come together to chart a different course.

The Artificial Superintelligence Alliance, or ASI, is a collaborative initiative between Fetch.ai, SingularityNET, Ocean Protocol, and, more recently, CUDOS. The goal of the alliance is to build a decentralised infrastructure stack for AI—one that prioritises openness, composability, and community governance over closed development and proprietary models.

By combining their respective strengths in agent-based systems, neural network marketplaces, decentralised data sharing, and distributed computing, the ASI Alliance is aiming to lay the groundwork for a modular, interoperable AI ecosystem on the blockchain.

In this article, we’ll examine what the ASI Alliance is, why these projects have chosen to merge, how their architecture is designed to work in practice, and what this could mean for the future of decentralised AI.

Why Was the ASI Alliance Formed?

Artificial Intelligence and Web3 are two of the most transformative technological trends of our time. But until recently, their paths have largely remained separate.

On the AI side, rapid advancements in large language models and machine learning have been led by a handful of well-funded corporations. These firms control the training data, model weights, and infrastructure, leaving little room for transparency, decentralisation, or public oversight. As a result, there’s growing concern that AI development is becoming increasingly centralised, proprietary, and opaque.

Conversely, Web3 has prioritised decentralisation from the outset, focusing on open infrastructure, tokenised incentives, and community-led governance. But while the crypto industry has made notable strides in finance, data, and ownership, it has often lacked the tooling, talent, and resources required to compete with centralised AI research labs.

This is where the Artificial Superintelligence Alliance (ASI) enters the picture.

Rather than continuing to build independently, the teams behind Fetch.ai, SingularityNET, and Ocean Protocol recognised that their individual strengths could form a more compelling proposition when aligned under a shared mission. Each project addresses a different layer of the decentralised AI stack:

• Fetch.ai specialises in autonomous economic agents—software entities capable of negotiating, transacting, and coordinating without human intervention.
   
• SingularityNET brings a decentralised AI service marketplace, allowing developers to publish, combine, and monetise models via smart contracts.
   
• Ocean Protocol provides infrastructure for decentralised data sharing, with privacy-preserving access controls and incentive models for data providers.
   
• And with CUDOS joining the alliance, the network gains access to distributed cloud computing—crucial for running inference workloads and training large models.

By merging tokens and governance under a single umbrella—via the ASI token and a phased integration roadmap—the alliance seeks to avoid duplication, increase network effects, and accelerate the path toward a decentralised AI infrastructure that can rival today’s centralised alternatives.

Crucially, the merger also sends a broader signal: that AI’s future need not be monopolised by a handful of corporate entities. With the right architecture, incentives, and community support, decentralised AI remains a viable—and perhaps necessary—counterbalance.

What Each Project Brings to the Table

The Artificial Superintelligence Alliance may sound ambitious, but its strength lies in the strategic alignment of its founding members. Rather than overlapping efforts, each team brings deep expertise in a distinct area of decentralised AI—whether that’s autonomous agents, data sharing, AI services, or compute infrastructure. 

Fetch.ai

Founded by Humayun Sheikh, an early investor in DeepMind, Fetch.ai was born out of a vision to create a decentralised version of what AI could have become without corporate control. The platform focuses on building autonomous software agents that can act on behalf of users or organisations, performing tasks like negotiating contracts, sourcing data, or interacting with other services—all without ongoing human input.

Fetch.ai initially focused on Autonomous Economic Agents (AEAs), but in 2024, it introduced Microagents—lighter, more accessible AI agents designed to operate in mobile and Web3 environments. These agents can now be integrated directly into dApps or consumer applications, making them far easier to deploy and maintain.

Fetch.ai is also leading the development of ASI-1 Mini, a compact open-source language model designed to be governed by token holders. It represents a first step toward decentralised ownership of AI infrastructure—an important departure from today’s centrally trained foundation models.

SingularityNET 

SingularityNET takes a different approach. Founded by Dr. Ben Goertzel, an AGI researcher with a background in cognitive science and bioinformatics, the project aims to create a global, decentralised marketplace for AI. In simple terms, it allows developers to publish AI models and services on-chain, where anyone can access and use them in exchange for the native $ASI token.

SingularityNET’s architecture is designed for interoperability. AI services listed on the platform can interact, compose, and build on one another—forming what the team calls a “network of AI agents.” These agents aren’t locked into a single application—they can be reused, recombined, and monetised freely across use cases.

The project has also launched several ecosystem spinouts:

• Rejuve.ai – A platform for decentralised health research.
   
• NuNet – Focused on decentralised edge compute.
   
• TrueAGI – Building safer AGI frameworks.
   
• SophiaVerse – A metaverse project using cognitive AI agents.

Ocean Protocol 

Data is essential for training and operating intelligent systems—but sharing it across organisational boundaries remains one of the biggest hurdles in AI development. Ocean Protocol addresses this with tools that allow individuals and enterprises to monetise, share, and use data securely, without losing control or violating privacy regulations.

The project pioneered the concept of Compute-to-Data, a system that sends algorithms to where data is stored (instead of moving the data itself). This allows institutions like hospitals or banks to contribute datasets for AI training without ever exposing the raw data.

Ocean’s tech stack includes:

• Data NFTs – Which represent on-chain access rights to datasets.
   
• Service Execution Agreements (SEAs) – Smart contracts that automate how, when, and by whom data can be used.
   
• Fine-grained access control systems – Letting data owners manage permissions and pricing.
   
• In the context of ASI, Ocean provides the data coordination layer. It ensures that AI agents, compute networks, and applications can access relevant datasets under transparent, secure, and privacy-preserving conditions.

CUDOS

While powerful AI requires good data and intelligent agents, it also needs serious compute power to train and run models. This is where CUDOS steps in. Originally built as a decentralised cloud infrastructure provider, CUDOS offers both virtual and physical compute resources—from GPUs and CPUs to bare-metal servers and edge devices.

The platform supports high-performance training, inference, and storage through a tokenised compute network, where users can pay for resources using crypto, and hardware providers can earn rewards for contributing capacity.

Key features include:

• S3-compatible storage, Infiniband connectivity, and flexible payment rails.
   
• Support for cloud, edge, and hybrid compute environments.
   
• Integration with Web3 wallets and smart contracts for automated provisioning.

In an interview, co-founder Pete Hill explained that CUDOS originated from a traditional Web2 infrastructure company that shifted focus after realising how much underutilised capacity existed in the cloud sector. This “compute sharing economy” vision now sits at the heart of its mission.

Within ASI, CUDOS functions as the compute backbone—powering everything from model training and multi-agent simulations to large-scale deployments across industry applications.

The $ASI Token

At the time of writing, the native token for the Artificial Superintelligence Alliance still trades under the ticker FET, the original asset of Fetch.ai. However, this is due to change to ASI, reflecting the token merger that brings together the four projects into one unified ecosystem. For clarity, we’ll refer to the token as $ASI going forward. 

From Four to One: The Token Merger

The ASI token unifies four tokens—$FET (Fetch.ai), $AGIX (SingularityNET), $OCEAN (Ocean Protocol), and $CUDOS—into a single asset, designed to support shared infrastructure and collective growth.

The merger follows fixed conversion rates:

• $FET holders need not convert; $FET becomes $ASI at a 1:1 ratio.
   
• $AGIX tokens convert at a rate of 1 $AGIX = 0.433350 $ASI.
   
• $OCEAN tokens convert at 1 $OCEAN = 0.433226 $ASI.
   
• $CUDOS holders undergo a two-step process:
  
  
  • The base conversion rate is 112.427 $CUDOS = 1 $FET.
     
  • After applying a 5% merger fee, the effective rate becomes 118.344 $CUDOS = 1 $FET, translating to approximately 1 $ASI per 118 $CUDOS.

Supply Impact and Emissions

Post-merger, the total supply of $ASI will reflect the circulating and vested allocations of all four tokens. Notably, the inclusion of CUDOS introduces an additional ~88.9 million FET (now ASI) into circulation based on the adjusted conversion rate—this figure represents the effective token migration from CUDOS's 10 billion total supply.

To ensure orderly issuance and limit inflationary pressure, vesting is staggered:

• Public CUDOS holders: 3-month linear vesting.
   
• CUDOS treasury allocations: 10-month vesting.
   
• Other migrated assets ($AGIX and $OCEAN): follow their original vesting terms, now applied to $ASI.

No additional supply inflation is introduced beyond the calculated migration. All emissions and future incentives will be governed by the ASI community through on-chain governance.

Utility Across the Ecosystem

With ASI, token holders gain access to services spanning the entire decentralised AI stack:

• Staking: Support network security, governance, and infrastructure services.
   
• Governance: Participate in protocol-wide votes, including AI alignment strategies, ecosystem grants, and resource allocations.
   
• Access: Use $ASI to pay for compute on CUDOS, train agents on Fetch.ai, buy and sell models via SingularityNET, or access token-gated datasets via Ocean Protocol.

Importantly, each network retains functional independence, but $ASI provides the connective tissue—creating a shared economic foundation for cross-platform AI innovation.

Real-World Use Cases

While bold claims are nothing new in crypto, the ASI Alliance has opted for a more grounded approach—prioritising infrastructure over hype. With components ranging from decentralised agents and data marketplaces to compute layers and training frameworks, ASI is laying the technical foundations for real-world use.

Let’s take a closer look at how some of these systems are already being put into practice.

Intelligent Agents and Automated Marketplaces

One of the key building blocks of ASI is Fetch.ai’s autonomous agent technology. These agents are pieces of software that can act independently on behalf of users, companies, or even other agents. They can search for data, negotiate prices, buy services, or complete tasks—all without needing constant human input.

This kind of automation is already being trialled in several areas, from logistics to trading. Imagine a future where your energy provider automatically switches you to the cheapest supplier, or your car negotiates its own parking fees. That’s the kind of agent economy ASI is aiming for—and it’s not as far off as it sounds.

The alliance's launchpad, called Create, makes deploying these agents far easier.

Developers can use templates and pre-built tools to spin up new agents and they can crowdfund or monetise their ideas directly from the platform. The goal is to lower the barrier to entry and kick-start a new ecosystem of AI services that work for users—not just big corporations.

Training AI Without Exposing Private Data

Training powerful AI models typically requires massive datasets—but handing over that data can create huge privacy and compliance issues, especially in industries like healthcare and finance.

To solve this, Ocean Protocol’s “compute-to-data” technology allows models to be trained where the data sits, without moving or exposing the underlying information. So, instead of giving a third party access to sensitive data, hospitals or businesses can let AI algorithms train on it in a secure, decentralised way.

This same approach is now being integrated into the broader ASI stack, allowing developers to fine-tune and train models while keeping full control over the data. It's especially useful for collaborative projects, where institutions can contribute data without actually sharing it—preserving both privacy and security.

Monetising AI in the Open

Most of today’s AI development is locked away in private labs. If you build a great model, chances are you’ll end up selling it to the highest bidder—or never get it off the ground at all. ASI wants to change that.

Using the Create launchpad and Train tools, developers can not only build AI agents, but also monetise them in open marketplaces. They can offer their models as services, set usage fees, and even receive community funding through token-based crowdfunding.

Users benefit from this too. Rather than relying on black-box algorithms from big tech firms, they’ll have access to open, transparent AI services that anyone can audit, improve, or compete with. It’s a model that promotes innovation—and gives power back to the people actually building these tools.

Decentralised Compute Infrastructure

AI training and deployment needs serious hardware—often high-end GPUs that are expensive and difficult to access. This is where CUDOS, one of the newest members of the alliance, comes in.

CUDOS provides decentralised cloud and edge computing infrastructure, offering affordable access to powerful compute resources, which are essential for running large language models or training advanced agents.

Developers no longer need to rely on Amazon or Google for cloud compute—they can access it directly from the ASI ecosystem, often at a lower cost. And anyone with spare compute capacity can contribute their hardware to the network and earn rewards, creating a more sustainable and distributed alternative to traditional cloud providers.

Roadmap & Upcoming Milestones

The ASI Alliance has outlined a detailed roadmap for 2025, laying out the next phase in its mission to build a decentralised stack for AI development and deployment. The roadmap is extensive, covering infrastructure, AI models, data systems, deployment tools, and community growth. Rather than diving into every detail, we’ll highlight the key areas that provide a clear sense of where things are headed.

At a high level, the roadmap is organised around four pillars: infrastructure and network tools, AI algorithms and models, deployment applications, and ecosystem incentives.

Infrastructure 

Much of the early roadmap focuses on building the foundations—with CUDOS leading the way on decentralised compute. In the first half of the year, we’ll see further rollout of bare metal servers, Infiniband connections, GPU provisioning, and smart contract tooling for compute management. Later in the year, the Alliance will launch ASI: Zero, a custom-built, ledgerless Layer-0 designed specifically for high-performance AI operations across chains like Ethereum, Cosmos, and Cardano.

This infrastructure is critical for powering model training, inference engines, and data marketplaces without relying on centralised cloud providers.

AI Model Development and Learning Systems

In terms of AI models, the roadmap shows steady progress throughout the year. Early development will focus on areas like robotics, biochemistry, and physics-informed neural networks, building on what’s described as Vision-Language-Action (VLA) models. These will later evolve into more scalable systems supported by tools like Hyperon, a neural-symbolic reasoning layer used to support cross-domain AI agents.

Meanwhile, the ASI: Learn platform will introduce tools for converting data into knowledge graphs, indexing large-scale datasets, and enabling transparent AI learning. Combined with Ocean’s “Hyperpredictor,” these tools aim to make model training more interpretable, traceable, and collaborative.

ASI: Create – The Developer Launchpad

One of the most central products on the roadmap is ASI: Create, a developer-focused launchpad for building and deploying AI agents. Starting in Q1 with an MVP that supports inference engines and agent hosting, it will expand each quarter with new features—like agent templates, LLM aggregation, crowdfunding, and finally integration with IDEs like VS Code. The goal is to make it as easy as possible to build useful AI services, with modular tools and decentralised infrastructure available out of the box.

Data Systems and the Marketplace

The ASI: Data platform will allow users to buy, sell, and trade datasets with full provenance and privacy protection, integrating Ocean Protocol’s existing compute-to-data architecture. Data scientists will be able to contribute and monetise datasets while keeping them secure, with support from Ocean’s prediction models and enterprise tooling.

Incentives and Community Growth

To bring users and developers into the ecosystem, the roadmap also includes ongoing community-focused efforts like ASI hackathons, “earn and burn” incentive programs, and Ocean’s Node Booster and Ocean Foam campaigns. These initiatives aim to kickstart usage of the ASI stack across its various layers.

Potential Challenges

While the vision of the ASI Alliance is ambitious, the road to decentralised superintelligence is anything but straightforward. As with any large-scale initiative involving multiple partners, cutting-edge technologies, and a fast-evolving regulatory environment, there are several potential challenges that could shape—or constrain—its long-term success.

Cross-Team Integration

One of the more immediate hurdles for the ASI Alliance is coordination between its founding members. Fetch.ai, SingularityNET, Ocean Protocol, and CUDOS each bring unique technologies and teams—with different ways of working. While their tools complement one another, combining them into a unified stack is a complex task. Aligning roadmaps, development timelines, and integration efforts takes careful planning and constant communication. 

The modular design of the ASI stack is a clear strength, but it can also lead to delays or misalignment if priorities shift across teams. Keeping everything on track will be key to delivering a seamless user experience.

Regulatory Uncertainty

The ASI Alliance is building at the intersection of two rapidly evolving sectors—crypto and AI—both of which have experienced regulatory scrutiny. Token-based incentives for things like data sharing, compute access, or AI training may raise questions from regulators, especially in regions where the rules around digital assets remain unclear. At the same time, new policies around AI transparency, data protection, and algorithmic accountability are beginning to take shape. This means ASI will need to stay flexible and proactive as global rules continue to evolve.

Reaching Beyond the Crypto Community

While ASI’s vision is rooted in Web3, many of its real-world applications—such as healthcare, scientific research, or robotics—depend on adoption by traditional organisations. These groups may be unfamiliar with decentralised tools, and convincing them to use agent-based systems or tokenised infrastructure could take time. For ASI to grow beyond the crypto space, onboarding needs to be simple, reliable, and backed by strong incentives that speak to users outside of the typical blockchain crowd.

Closing Thoughts 

The Artificial Superintelligence Alliance represents one of the most coordinated and technically ambitious collaborations we’ve seen in the decentralised AI space. By merging the strengths of Fetch.ai, SingularityNET, Ocean Protocol, and CUDOS, the Alliance isn’t just building infrastructure—it’s laying the groundwork for a new digital economy built around open AI systems.

Each project brings something essential to the table: autonomous agents, trusted data exchange, decentralised compute, and AI service marketplaces. Together, they form a modular stack that aims to make artificial intelligence more transparent, accessible, and collectively governed.

Of course, the road ahead is not without challenges. Aligning product roadmaps, building adoption, and ensuring scalability will take time. But with real infrastructure already in place—and a roadmap that spans training models, deploying agents, and onboarding developers—ASI appears to be moving with clear intent.

Whether or not it lives up to its long-term ambitions remains to be seen. But in a world where control over AI is rapidly centralising, projects like ASI may offer an important counterbalance—one where value and ownership flow not just to corporations, but to communities.