Aethir establishes DePIN computing leadership with enterprise-level growth: a new generation of computing infrastructure model driven by real revenue
Aethir is a global leader in decentralized GPU cloud infrastructure, dedicated to providing enterprise-level computing power services for AI, gaming, and next-generation Web3 applications.
Original source: Aethir
Against the backdrop of surging global demand for AI infrastructure, traditional centralized cloud computing systems have gradually revealed capacity bottlenecks and efficiency ceilings. With the rapid penetration of large model training, AI inference, and agent applications, GPUs are transforming from "computing resources" into "strategic infrastructure assets." Amid this structural market transformation, Aethir, through its decentralized physical infrastructure network (DePIN) model, has built the industry's largest and most commercialized enterprise-grade GPU computing network, quickly establishing a leading position in the sector.
Commercial Breakthroughs in Scalable Computing Infrastructure
To date, Aethir has deployed over 435,000 enterprise-grade GPU containers worldwide, covering the latest NVIDIA hardware architectures such as H100, H200, B200, and B300, and has delivered more than 1.4 billion hours of real computing services to enterprise clients. In just the third quarter of 2025, Aethir achieved revenue of $39.8 million, driving the platform's annual recurring revenue (ARR) to surpass $147 million.
Aethir's growth is driven by genuine enterprise-level demand, including AI inference services, model training, large AI Agent platforms, and production-grade workloads from global game publishers. This revenue structure marks the first time in the DePIN sector that a scalable computing platform is driven primarily by enterprise-level paid usage.
Aethir's infrastructure has been integrated into the core production systems of several cutting-edge AI companies.
1. Kluster.ai leverages Aethir's computing network to compress clinical trial patient screening processes, which originally took months, down to minutes, significantly enhancing the commercial viability of medical AI.
2. Attentions.ai builds and deploys enterprise-grade private large models through Aethir, promoting the implementation of no-code AI platforms in traditional industries.
3. Mondrian AI, selected for "KOREA AI STARTUP 100," uses Aethir as the underlying computing support for its enterprise-grade AI services.
At the gaming industry level, Aethir's production-grade delivery capabilities have undergone large-scale commercial validation.
1. SuperScale's test data shows that products based on Aethir's instant cloud gaming architecture have increased user preference by 43%, click-through rates by 35%, and final paid conversion rates by 45% compared to traditional download methods.
2. In Reality+'s "Doctor Who: Worlds Apart" project, Aethir drove a 201% increase in installation conversion rate and a 61% increase in ARPU.
Currently, over 400 games have been integrated for testing via Xsolla, and leading global publishers such as Scopely, Zynga, and Jam City are conducting in-depth evaluations of the system.
Institutional Endorsement: The Birth of Strategic Compute Reserves
In October 2025, Aethir completed a $344 million ATH token targeted investment (NASDAQ: POAI) and officially launched the Aethir Digital Asset Treasury (DAT). This mechanism is positioned as the world's first "Strategic Compute Reserve" (SCR) framework, aiming to incorporate decentralized computing assets into enterprise-level long-term balance sheet systems and explore deep integration paths between computing assets and traditional capital markets.
As of November 10, 2025, the DAT has disclosed holdings of 5.7 billion ATH tokens and plans to provide services to AI companies by deploying GPU resources, using the resulting revenue to buy back ATH, thereby forming a positive cycle of "compute supply—enterprise monetization—ecosystem buyback."
This structural capital move makes Aethir one of the very few platforms in the DePIN sector to receive substantive recognition from traditional capital markets, and it is also the first to advance "decentralized computing power" to the level of institutional asset allocation.
Enterprise-Grade Delivery Capabilities Under a Decentralized Architecture
The decentralized GPU network built by Aethir has already matched or even surpassed traditional centralized cloud providers in terms of performance, stability, and cost structure. H100-grade GPUs currently support over 90% of mainstream large model inference tasks globally, and in the first quarter of 2025, NVIDIA allocated 60% of its production capacity specifically to enterprise AI clients, further highlighting the strategic scarcity of high-end GPUs.
Through a distributed hardware supply system, Aethir bypasses the construction cycles and supply chain bottlenecks of traditional data centers, offering enterprises near bare-metal computing performance with higher resource utilization and more flexible pricing, while significantly reducing overall usage costs.
DePIN Enters the "Real Revenue Driven" Era
According to McKinsey, global data center construction investment will reach $6.7 trillion by 2030. Meanwhile, the DePIN market is expected to grow to $3.5 trillion by 2028. However, only platforms with genuine enterprise revenue capabilities and scalable delivery capacity are qualified to build a long-term moat in this sector.
About Aethir
Aethir is a global leader in decentralized GPU cloud infrastructure platforms, dedicated to providing enterprise-grade computing services for AI, gaming, and next-generation Web3 applications. Through a distributed GPU network architecture, Aethir transforms idle computing power worldwide into cloud-level resources that enterprises can access instantly, building a more open, efficient, and decentralized digital infrastructure.
Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.
You may also like
Before Cook steps down, is Apple playing its final trump card?
Apple has launched new Mac mini models equipped with M6 or M5 Pro chips, as well as an upgraded Mac Studio featuring M5 Ultra or M5 Max chips. The M6 chip is Apple’s first to use the 2-nanometer process, enhancing energy efficiency and on-device AI capabilities; the M5 Ultra focuses on professional-grade high-performance computing. These two new products mark the beginning of a new product cycle, while Tim Cook will step down as CEO on September 1st, to be succeeded by John Ternus.
AI Computing Power Capital Battle Upgrades Again! Nvidia-backed "New Cloud" Lambda Plans to Raise $3 Billion to Fuel Its Computing Power Landscape and Prepare for IPO
AI cloud computing service provider Lambda is in talks for a funding round of up to $3 billion, preparing for a potential initial public offering next year.

Net Profit Decline Can't Hide Core Momentum! Bank of Montreal (BMO.US) Q3 Adjusted Profit Beats Expectations, Plans to Launch Share Buyback Program for Up to 25 Million Shares
Due to the drag from one-time items such as the sale of transportation and supplier financing businesses, Bank of Montreal reported a year-on-year decline in Q3 net profit. However, adjusted earnings per share and revenue both exceeded market expectations, and pre-provision, pre-tax profit in all business segments reached record highs. The bank also announced plans to launch a share repurchase program of up to 25 million shares in the third quarter.

Dick's Sporting Goods Lowers Full-Year Outlook Following Second-Quarter Miss
