On November 17, 2025, Acurast officially launched its Genesis Mainnet, introducing what it describes as the world’s first smartphone-powered decentralized confidential computing network. The launch coincided with the debut of the project’s native token, $ACU, and follows an $11 million fundraising effort that includes backing from Ethereum co-founder and Polkadot founder Gavin Wood, MN Capital founder Michael van de Poppe, and other prominent investors in the blockchain and artificial intelligence space. The project represents a significant milestone in the convergence of decentralized physical infrastructure networks, known as DePIN, and AI-driven compute requirements, two of the most active narratives in the cryptocurrency market as Bitcoin trades above $92,000 and Ethereum near $3,025.
The Synergy
Acurast’s core proposition brings together two technological trends that have been developing along parallel tracks. On one side, the exponential growth in AI compute demand has strained traditional data center capacity, driving up costs and creating bottlenecks for developers building machine learning and inference workloads. On the other side, the DePIN movement has demonstrated that decentralized networks can effectively coordinate real-world hardware resources, from wireless coverage to GPU computing, using blockchain-based incentive structures. Acurast bridges these domains by leveraging the most ubiquitous computing device on the planet: the smartphone.
The project’s founder, Alessandro De Carli, frames the opportunity in direct terms. Billions of smartphones already exist worldwide, each containing increasingly powerful processors that spend most of their time idle. By turning these devices into verified, confidential compute nodes, Acurast creates a distributed compute fabric that requires no new data center construction, no massive capital expenditure for server farms, and no centralized intermediary controlling access. The economic model aligns device owners, who earn tokens for contributing compute power, with developers who gain access to affordable, censorship-resistant computation.
AI Use Cases in Web3
The timing of Acurast’s mainnet launch aligns with a surge of interest in the intersection of AI and blockchain technology. The broader Robotics and Crypto narrative, sometimes labeled DePAI for Decentralized Physical AI, has seen its total market capitalization reach approximately $1.4 billion in November 2025, with a remarkable 197 percent surge in a single 24-hour period. This sector encompasses AI agent protocols on platforms like Base’s Virtuals Protocol, decentralized compute networks like Acurast, and projects building the infrastructure for autonomous machine-to-machine transactions.
Acurast specifically targets confidential computing workloads, a category of AI and data processing that requires the data being processed to remain encrypted and inaccessible even to the hardware operator. This is achieved through Trusted Execution Environments, or TEEs, which provide hardware-backed isolation for sensitive computations. In practical terms, a developer can send an AI inference task to the Acurast network, and it will be processed on a smartphone somewhere in the world without that phone’s owner, or anyone else, being able to see the input data or the results. This capability has immediate applications in healthcare data processing, financial analytics, and privacy-preserving machine learning training.
Data Privacy Implications
Privacy is not merely a feature of Acurast’s architecture but its foundational principle. Each smartphone participating in the network must pass cryptographic verification proving it runs approved, manufacturer-certified software. Hardware-backed key pairs confirm device authenticity, and any device that fails verification is rejected from the network. This attestation process ensures that only genuine, uncompromised hardware contributes compute power, addressing one of the primary concerns about decentralized compute networks, which is the risk of malicious nodes tampering with computation results or exfiltrating data.
The privacy guarantees extend beyond the compute layer. Because workloads are distributed across a global network of individual devices rather than concentrated in identifiable data centers, there is no single point of failure or surveillance. This architectural choice has significant implications for users in jurisdictions with restrictive data sovereignty requirements, as computations can be constrained to run only on devices within specific geographic regions while maintaining the economic benefits of a global network.
The Innovation Frontier
Prior to mainnet launch, Acurast had already demonstrated substantial traction. More than 146,000 smartphones were onboarded as active compute nodes during the testnet phase. The network processed nearly 490 million on-chain transactions and enabled over 85,000 workload deployments. These figures suggest genuine demand for decentralized compute capacity beyond speculative interest in the token itself. The $5.4 million oversubscribed public token sale on CoinList in May 2025 further validated market interest.
Investor confidence reflects the thesis that Acurast is building infrastructure at the intersection of multiple growth curves. Vineet Budki, CEO of Sigma Capital, described the project as creating a new user-owned compute layer that aligns with the direction AI is heading. Michael van de Poppe emphasized that the team had shipped real product, grown actual usage, and proven demand before the mainnet launch, a rarity in a market where many projects launch tokens before building functional technology.
Concluding Thoughts
Acurast’s mainnet launch represents a meaningful step forward for the DePIN sector and for decentralized computing more broadly. By leveraging the massive installed base of smartphones, the project bypasses the capital-intensive model of traditional cloud providers and offers a genuinely distributed alternative. The use of hardware-attested Trusted Execution Environments for confidential computing addresses real market demand from AI developers and enterprises with privacy-sensitive workloads. Whether Acurast can scale to meet enterprise-grade performance requirements and compete with the reliability of centralized cloud providers remains to be seen, but the project’s testnet metrics and investor backing suggest it has the foundation to make the attempt. In a market where the DePIN narrative is still young and total sector capitalization remains below $2 billion, projects with real hardware deployment and verifiable usage stand apart from the majority of speculative tokens.
Disclaimer: This article is for informational purposes only and does not constitute financial or investment advice. Always conduct your own research before making any financial decisions.
billions of idle smartphones worldwide. if even 1% opt in to compute networks thats more distributed power than AWS
Leah Thompson 1 percent of smartphones is 150M devices but battery swelling and thermal throttling will limit real uptime
Leah Thompson 1% of global smartphones is still 150M devices. the battery drain and thermal throttling issues are the real bottleneck not adoption
thermals_ok battery swelling from sustained compute load is a real fire risk. phones werent designed for 24/7 number crunching
thermals_ok battery degradation is real but have you seen the benchmarks on snapdragon 8 gen 3? efficiency improved a lot. the real issue is whether acurast pays enough to justify the hardware wear
thermals_ok 8 gen 3 efficiency doesnt matter when ambient temp hits 35C in mumbai or lagos. phones throttle and output drops. compute farms need climate control that phones werent designed for
The best projects are the ones quietly shipping during bear markets
Isabella shipping during a bear market is right. Gavin Wood backing smartphone compute nodes is the kind of weird bet that pays off
Gavin Wood backing smartphone compute is the kind of contrarian bet that either looks genius or delusional in 3 years
depin_scan gavin wood backing smartphone compute is weird until you realize old phones have better uptime than half the data centers in southeast asia
Education is still the biggest barrier to mainstream adoption
smartphone compute for AI inference is clever until you realize mobile RAM is the bottleneck. 8GB on a flagship cant hold a 7B parameter model for quantized inference without aggressive swapping
thermal_cliff_ 8GB RAM cant hold a 7B model even quantized. but 4B parameter models like Phi-3 run fine on mobile NPU. the question is whether anyone wants to pay for phone-grade inference
Mass adoption is happening incrementally — people just don’t notice
Gavin Wood backing this gives it credibility but 11M raise is thin for a hardware dependent DePIN network. Helium raised 350M and still struggled with real world deployment
Priya V. 11M raise vs Heliums 350M is misleading. Helium built physical hotspots from scratch. Acurast uses phones people already own. completely different capital requirements
smartphone compute works for lightweight inference workloads. anything memory bound dies instantly on mobile silicon. the 11M raise wont fix physics
The fundamental value proposition of crypto keeps getting stronger
This is exactly the kind of development the space needs
gavin wood keeps finding weird edges of crypto that actually work. smartphone compute sounds dumb until you see the idle processing power sitting in drawers across southeast asia
phone_farm_88 the idle phone thesis sounds great until you realize the logistics of collecting and maintaining 150M devices. acurast needs a deployment story and a supply story
ACU tokenomics matter here. if compute providers need to stake ACU to participate then token value depends on actual network usage not speculation
Lieselotte F. staking ACU to provide compute means token value is tied to actual usage. thats the DePIN thesis done right if it works
old phones doing decentralized compute is clever but the thermal throttling on a 3 year old phone running inference workloads will be brutal
11M raised with Gavin Wood backing. DePIN plus AI is the 2025 narrative but smartphone compute feels like a stretch vs actual data centers