AI × DePIN: Can idle phones become a computing network?
AI’s computing needs aren’t limited to training large models.
Email classification, content moderation, data scraping, and automated tasks also require computing resources that run continuously. Meanwhile, many phones still have unused processing capacity every day.
@Acurast aims to connect these distributed phones so developers can use them on demand for specific tasks.
Its model works as follows: phones run a Processor, developers submit computing tasks (deployments), and the network matches devices according to resource availability and execution requirements. Tasks can run in a phone’s secure hardware environment, with the results sent to a designated application or chain upon completion. This gives developers a way to try deploying workloads beyond traditional data centers.
Secure hardware provides mechanisms for execution and verification; it does not mean that every task is free of privacy or reliability risks. Project progress and technical developments should be assessed based on their dates.
On January 20, 2026, Acurast announced the launch of its mainnet and
$ACU . At the time, it said that more than 169,000 phones had connected to the network and that over 364,000 deployments had been completed. These figures include the network’s history before the mainnet launch and should not be treated as paid mainnet orders or revenue.
There are also real-world task examples to try, beyond the mainnet announcement.
In its October 2 introduction to the Laya demo, Acurast showed Android phones on the network running small models to handle requests such as email classification, game decisions, and prompt screening. The post said that more than 320,000 phones had connected to the network.
This shows that phones can handle some lightweight AI inference, but the performance of a single demo does not yet prove the cost-effectiveness or reliability of large-scale commercial tasks. The Laya demo $ACU is Acurast’s native mainnet token, used for network fees, computing deployments, and node incentives.
Its growth potential depends on whether real demand can keep pace with supply:
How many paid tasks run continuously?
What are the actual utilization, completion, and service quality rates for phones?
Can task fees cover the costs of devices, electricity, and maintenance?
Annual token inflation and unlocks should also be monitored.
Growth in the number of phones is progress, not a guarantee of token price appreciation or returns.
This article is for sharing purposes only and does not constitute investment advice.
#Acurast #ACU
AI’s computing needs aren’t limited to training large models.
Email classification, content moderation, data scraping, and automated tasks also require computing resources that run continuously. Meanwhile, many phones still have unused processing capacity every day.
@Acurast aims to connect these distributed phones so developers can use them on demand for specific tasks.
Its model works as follows: phones run a Processor, developers submit computing tasks (deployments), and the network matches devices according to resource availability and execution requirements. Tasks can run in a phone’s secure hardware environment, with the results sent to a designated application or chain upon completion. This gives developers a way to try deploying workloads beyond traditional data centers.
Secure hardware provides mechanisms for execution and verification; it does not mean that every task is free of privacy or reliability risks. Project progress and technical developments should be assessed based on their dates.
On January 20, 2026, Acurast announced the launch of its mainnet and
$ACU . At the time, it said that more than 169,000 phones had connected to the network and that over 364,000 deployments had been completed. These figures include the network’s history before the mainnet launch and should not be treated as paid mainnet orders or revenue.
There are also real-world task examples to try, beyond the mainnet announcement.
In its October 2 introduction to the Laya demo, Acurast showed Android phones on the network running small models to handle requests such as email classification, game decisions, and prompt screening. The post said that more than 320,000 phones had connected to the network.
This shows that phones can handle some lightweight AI inference, but the performance of a single demo does not yet prove the cost-effectiveness or reliability of large-scale commercial tasks. The Laya demo $ACU is Acurast’s native mainnet token, used for network fees, computing deployments, and node incentives.
Its growth potential depends on whether real demand can keep pace with supply:
How many paid tasks run continuously?
What are the actual utilization, completion, and service quality rates for phones?
Can task fees cover the costs of devices, electricity, and maintenance?
Annual token inflation and unlocks should also be monitored.
Growth in the number of phones is progress, not a guarantee of token price appreciation or returns.
This article is for sharing purposes only and does not constitute investment advice.
#Acurast #ACU