Axe Compute Inc. has blown past its initial 2026 target for AI infrastructure deals with a headline-grabbing $1.5 billion five-year agreement alone, pushing its total contract value to over $1.3 billion well before July. The company now aims to add $2 billion more before the year ends, raising expectations for a record-setting $3 billion in deals for 2026.
Building Massive Nvidia Blackwell GPU Clusters
At the heart of these contracts is deployment of GPU clusters driven by Nvidia’s cutting-edge Blackwell architecture. Axe Compute plans to roll out more than 9,200 GPUs designed for AI training and inference workloads. The company collaborates with Aethir to tap a distributed GPU network capable of scaling capacity within 48 hours, blending traditional GPU infrastructure with decentralized computing.
Financials and Contract Pipeline
The expected annual recurring revenue from these new agreements tops $384 million, with revenue recognition scheduled to commence in late Q4 2026. The standout $1.5 billion contract alone implies about $300 million in yearly revenue from a single client over five years. Previously, in April 2026, Axe secured a $260 million three-year deal for deploying 2,304 Nvidia B300 GPUs, laying the foundation for scaling their Blackwell deployments.
Expanding AI Infrastructure with Crypto-Linked Networks
The partnership with Aethir introduces an intriguing crossover with crypto infrastructure: Aethir’s decentralized GPU network and its actively traded ATH token blur boundaries between traditional tech and blockchain ecosystems. While Axe Compute’s growth might indirectly benefit this ecosystem, their ties remain collaborative rather than financially dependent.
However, executives recognize the risks as supply chain issues persist, especially around GPU availability and data center capacity. For investors tracking AGPU, key indicators to watch are the speed of contract conversions, timely infrastructure deployments, and whether that $384 million in recurring revenue materializes as forecasted in Q4.
