On May 6, 2025, Acurast announced it had surpassed 60,000 smartphones onboarded as compute nodes across more than 130 countries, marking a significant milestone for decentralized physical infrastructure networks. The achievement positions Acurast as what it calls the world’s most decentralized compute network, leveraging the unused processing power of everyday mobile devices to power Web3 workloads without relying on centralized cloud providers.
The Synergy
The convergence of mobile computing and decentralized networks represents one of the most compelling synergies in the AI-crypto intersection. Traditional cloud computing relies on massive data centers operated by a handful of corporations, creating bottlenecks in cost, latency, and geographic distribution. Acurast flips this model by distributing compute tasks across tens of thousands of smartphones, each acting as an independent processing node in a global mesh network.
This approach directly addresses the growing demand for compute resources driven by AI model training and inference. As large language models and generative AI systems require ever-increasing computational power, decentralized networks offer an alternative path that avoids the capital-intensive buildout of new data centers. The smartphone-as-node model also naturally distributes computation closer to end users, reducing latency for location-sensitive applications.
AI Use Cases in Web3
Acurast’s architecture enables several high-value AI use cases within the Web3 ecosystem. Distributed inference allows AI models to run predictions across multiple nodes without sending sensitive data to a central server, preserving privacy while maintaining performance. Decentralized data processing pipelines can aggregate and transform on-chain and off-chain data from diverse geographic sources, enabling real-time analytics for DeFi protocols and prediction markets.
The network also supports oracle computation, where smartphone nodes independently fetch and verify real-world data before submitting it on-chain. This creates a more resilient and tamper-resistant oracle layer compared to centralized alternatives, as compromising the system would require simultaneously corrupting thousands of geographically dispersed devices.
With Bitcoin trading at approximately $96,800 and the broader crypto market cap exceeding $3 trillion, the demand for reliable, decentralized computation continues to grow alongside the increasing sophistication of on-chain applications.
Data Privacy Implications
The distributed nature of smartphone-based compute networks offers inherent privacy advantages over centralized alternatives. When computation is split across thousands of independent devices, no single node processes the complete dataset, making it significantly harder for any one party to reconstruct sensitive information. This architectural privacy-by-design aligns with growing regulatory requirements around data sovereignty and user consent.
However, the model also introduces new privacy considerations. Participants who contribute their smartphones must trust that the workloads they execute do not compromise their personal data. Acurast addresses this through sandboxed execution environments that isolate compute tasks from the host device’s personal data and applications.
The Innovation Frontier
Looking ahead, the convergence of DePIN and AI opens several frontiers. Federated learning across smartphone networks could enable model training on real-world data without centralizing that data, preserving privacy while improving model accuracy. Edge AI inference on mobile devices could power decentralized applications that respond in real-time to local conditions, from supply chain monitoring to environmental sensing.
The economic model is equally innovative. Smartphone owners earn token rewards for contributing compute resources, creating a new form of passive income that democratizes access to the AI economy. With over 60,000 nodes already active, the network demonstrates that decentralized compute can scale organically through economic incentives rather than capital-intensive infrastructure investment.
Concluding Thoughts
Acurast’s 60,000-node milestone proves that decentralized compute is not a theoretical proposition but an operational reality. As AI workloads continue to grow exponentially and concerns about centralized cloud dominance intensify, the smartphone-as-node model offers a viable alternative that aligns economic incentives with privacy preservation and geographic distribution. The next challenge will be demonstrating that this distributed architecture can handle the most demanding AI training workloads, not just inference and data processing tasks.
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Always conduct your own research before making investment decisions.
Bear markets are for building — and builders are delivering
Every cycle the infrastructure gets more robust
The gap between crypto and TradFi is narrowing fast
60,000 smartphones across 130 countries as compute nodes. The latency distribution alone is fascinating. Each device processes fragments of AI workloads without central coordination. This is what decentralized compute actually looks like at scale.
130 countries and still counting. the geographic distribution alone makes this more resilient than any single cloud provider
The pace of innovation in crypto continues to surprise me
Using idle smartphone processing for distributed AI inference is clever. No new data centers needed. The oracle computation use case for DeFi protocols is where I see real traction. Smartphones are everywhere, GPU clusters are not.
smartphone compute is fine for inference but you still need GPUs for training. the 60k number is impressive but scope matters