News

Pi Network has over 421,000 Nodes, representing more than 1 million CPUs, that could support opt-in AI computing tasks. In the OpenMind pilot, 7 Pi Node operators returned image-inference results within 4 seconds, including labels such as “bus” and “person”. Pi Network is testing a new use for its global node network through a proof-of-concept tied to AI training and computing. The project centers on spare computing capacity across more than 421,000 Pi Nodes, which together represent over 1 million CPUs. This unused capacity could support external AI workloads beyond blockchain functions. The effort frames Pi Network’s latest AI move around distributed computing and paid participation by node operators. The project is in response to two broader issues in the AI sector. One is the strain linked to centralized computing, including data center limits and concentrated energy use. The other is the rising demand for computing power as AI models, agents, and services expand. Pi also pointed out that its distributed networks may help coordinate scattered and unused resources that would otherwise remain idle. Pi released a deep-dive case study into the recent proof-of-concept project for a new Pi Node utility that supports decentralized AI training and computing tasks for third-parties using the spare computing capacity of over 421,000 Pi Nodes. In collaboration with OpenMind, a… — Pi Network (@PiCoreTeam) March 6, 2026 The AI roadmap was announced as part of Pi Network’s updated Mainnet strategy during the first anniversary of its Open Network. As we previously covered, the plan placed artificial intelligence among the network’s top priorities alongside ecosystem tokens and identity services. Pi Network Tests Distributed Image Recognition Tasks  The recent proof-of-concept was completed with OpenMind, a robotics startup backed by Pi Network Ventures. OpenMind is building an operating system and open-source protocol for robots. To support that work, it needs computing power for training, evaluation, and model execution. The pilot tested whether Pi’s distributed Node network could handle AI-related tasks outside blockchain activity. For the test, OpenMind built a container that could send computing tasks to individual computers. Volunteer Pi Node operators downloaded the container and ran it on their own machines. OpenMind then sent image recognition tasks through the system. The computers processed images using OpenMind’s model, with the goal of identifying as many discrete objects as possible. Pi reported that the pipeline worked from end to end. Seven volunteer Pi Node operators joined the pilot, and job acknowledgments came back from all seven within one second. Inference results were returned from multiple workers within four seconds. The results included expected object labels such as bus and person, along with bounding boxes. Pi Nodes can accept external computing jobs and return valid results to a third-party client. Pi added that distributed AI training remains at a research stage, and more work is still needed across the sector. Still, the test offers an early example of how spare Node capacity could be packaged for AI companies seeking alternative computing resources. Recently, CNF noted that Pi Network tested AI image recognition tasks on its nodes with OpenMind, using idle CPU capacity while its Mainnet upgrade path continued. The test showed how unused node resources could support artificial intelligence workloads across the network. Additionally, Pi Network started Phase 2 of its mainnet protocol upgrades after completing the Protocol v19.9 migration. CNF reported the project now targets Protocol v20.2 before Pi Day 2026. Pi traded at $0.2285 , up 13.77% in 24 hours, with a $2.2 billion market cap and $65.38 million in daily volume.

According to Odaily, Peter Schroeder, Circle's Global Head of Markets, recently cited an Enterprise Onchain article on the X platform, stating that over the past nine months, 140 million payments have been completed between AI agents, with a total transaction volume of $43 million, 98.6% of which were settled in USDC. The average transaction amount was $0.31. There are more than 400,000 AI agents with purchasing power.

Starting from March 8, Daylight Saving Time will be implemented in North America. The trading hours for financial markets in the United States and Canada (including gold, silver, US oil, US stocks, etc.) and the release times for economic data will be moved one hour earlier compared to Standard Time. Beginning tomorrow (March 9), gold, silver, and US oil will open at 6:00 AM (GMT+8), and US stocks will open at 9:30 PM (GMT+8). Investors are advised to pay attention to these changes.