Bitcoin commands more than 40,000x the compute power of the world’s most powerful supercomputer, yet devotes it all to one task: securing a ledger. In doing so, it has demonstrated a new model for coordinating global resources to create money free from centralized control.

Many of the same ideas that inspired Bitcoin are now being used to decentralize AI. By leveraging the core principles behind Bitcoin’s success, we can build collective-intelligence networks that grow smarter and stronger with each new participant and resource onboarded. Building collective-intelligence systems depends on four core principles:

1. Horizontal Scaling – Each new node increases the overall performance.
   

2. Emergent Outputs – Aggregated resources create a single product that scales in value faster than the sum of its parts.
   

3. Verifiable Fault Tolerance – The network has robust mechanisms to remain reliable despite faulty or malicious nodes.
   

4. Mission-Aligned Incentives – Tokens align contributors toward shared goals.
   
   

Apply the four principles that made Bitcoin successful, and we can build collective-intelligence networks that match the scale and diversity of today’s largest labs in resources, remain open to anyone, and run new workloads, from agent swarms to fully open foundation models. The next sections unpack each principle and show how they can be combined into a flywheel that lets intelligence compound across the network.

Principle 1: Horizontal Scaling

Effective collective-intelligence systems improve their core output with each additional node. For example, adding compute to Bitcoin increases the network's overall value by enhancing monetary soundness and security. Similarly, scaling compute for AI models has demonstrated improved performance.

Source: https://newhedge.io/bitcoin/hashrate-vs-price

This differs from blockspace-focused chains like Ethereum or Solana, where adding validators doesn't proportionally increase value since factors like fees, latency, and liquidity become more important.

Crypto uniquely enables global resource coordination at unprecedented scale. As each node joins, the collective strengthens: more compute yields better models, more specialized agents enhance outputs, and more data deepens insights.

Principle 2: Emergent Outputs

Horizontal scaling is the input; emergent performance is the output. Aggregating resources into one emergent output defines collective intelligence. Bitcoin's security and value are meaningless at the individual node level; its value emerges at the network level, creating something greater than the sum of its parts. Similarly, AI emerges from outputs of compute and data.

On marketplaces like Uber or Airbnb, rides and rooms remain discrete units; supply merely satisfies current demand. Conversely, collective-intelligence networks aggregate resources into inseparable outputs. Adding more GPUs to a compute marketplace may lower prices or expand capacity, but it doesn’t make any one GPU better for the end user. In contrast, collective-intelligence systems ensure that each contributor or resource enhances the whole system, enabling applications that would be impossible in traditional settings.

While horizontal scaling and emergent output unlock huge gains in capacity and product quality, they introduce new risks. More nodes raise coordination complexity, enlarge the fault surface, and demand upfront capital to scale the network. Crypto-economic mechanisms solve these challenges: verifiable fault tolerance keeps the system correct when nodes fail or cheat, while mission-aligned incentives attract and retain the resources needed for continued growth.

Principle 3: Verifiable Fault Tolerance

As scale amplifies the risk of failure, collective-intelligence networks rely on robust fault tolerance and cryptographic verification—just as independent validation by decentralized Bitcoin miners turns unreliable nodes into a resilient whole.

Until recently, decentralized AI training seemed impossible due to communication bottlenecks and the unreliability of GPUs outside of centralized clusters. Breakthrough research in decentralized training can enable Bitcoin-like redundancy for AI systems, maintaining performance even when individual contributors fail.

Trustless collective-intelligence networks need outputs that machines can verify onchain, rejecting bad work without human gatekeepers. Tasks with provable computations fit first, because crypto’s battle-tested consensus already rewards honest nodes and slashes bad behavior.

Principle 4: Mission-Aligned Incentives

Tokens provide the incentive layer that draws compute, data, and talent into one mission, powering the horizontal scaling and emergent output that collective-intelligence systems require. Permissionless networks scale by uniting contributors around missions that centralized actors are unwilling or unable to pursue, whether due to cost, censorship, or coordination limits. Bitcoin proved this: incentives and the mission of financial sovereignty turned a fringe idea in 2008 into the world’s 13th-largest currency.

AI faces a similar barrier. Training a frontier model costs $100 million+ and may rise to billions, limiting most researchers and startups from creating new models. A decentralized alternative lets anyone route spare GPUs, data, or expertise toward a shared intelligence. Prime Intellect showed this in practice when 30 independent parties pooled resources to train an open 10B model, each earning token stakes tied to future upside.

Tokens solve both the bootstrapping and sustainability challenges: early participants fund critical infrastructure while transparent, programmatic rewards align individual incentives with collective outcomes.

Collective Intelligence represents one of the most exciting opportunities at the intersection of crypto and AI. By applying Bitcoin's foundational principles to artificial intelligence, we can create systems that potentially surpass centralized approaches through unprecedented resource coordination. Here are the most promising frontiers for collective intelligence:

• Decentralized Training: Projects like Pluralis, Prime Intellect, Nous, and Ambient demonstrate how new models can be bootstrapped with compute across the internet permissionlessly. As more compute joins the network, output quality improves. These networks also enable hundreds of specialized models to create superior collective intelligence through mixture-of-experts architectures.

• Data Generation: Nodes can both gather real-world data (like Grass for web scraping valuable datasets) and generate synthetic data through distributed simulation in fields like drug discovery, robotics, and materials science. As more data is collected, the overall dataset and IP become more valuable and can be used by the collective or sold to other companies.

• Agentic Networks: Swarms of AI agents collaborate on unified outputs, with each additional specialized agent enhancing collective capabilities. Consider a hedge fund collective where agents analyze different market aspects 24/7. More agents mean broader market coverage, deeper analysis, and ultimately better trading decisions. As more specialized agents (and compute) join the network, the system generates increasingly more valuable predictions and products. 

One of crypto’s greatest innovations is proving that global-scale resource coordination is possible through well-designed incentive systems. The same principles that made Bitcoin a trillion-dollar asset can now bootstrap open intelligence, removing AI’s scaling ceiling and enabling apps no central lab could build. Bitcoin showed we can crowd-secure money; the next wave will show we can crowd-build intelligence, so long as we follow the same blueprint.

We're looking for teams applying this blueprint to collective-intelligence networks that transform distributed resources into powerful emergent outputs in AI. If you are building something like this, let's talk.

Disclaimer: All information contained herein is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. You should consult your own advisers as to legal, business, tax, and other related matters concerning any investment. None of the opinions or positions provided herein are intended to be treated as legal advice or to create an attorney-client relationship. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by Variant. While taken from sources believed to be reliable, Variant has not independently verified such information. Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by Variant, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results. A list of investments made by funds managed by Variant (excluding investments for which the issuer has not provided permission for Variant to disclose publicly as well as unannounced investments in publicly traded digital assets) is available at https://variant.fund/portfolio. Variant makes no representations about the enduring accuracy of the information or its appropriateness for a given situation. This post reflects the current opinions of the authors and is not made on behalf of Variant or its Clients and does not necessarily reflect the opinions of Variant, its General Partners, its affiliates, advisors or individuals associated with Variant. The opinions reflected herein are subject to change without being updated. All liability with respect to actions taken or not taken based on the contents of the information contained herein are hereby expressly disclaimed. The content of this post is provided "as is;" no representations are made that the content is error-free.

Variant is an investor in Pluralis Research.