Conference Presentation, Panel, Fireside Chat
State of AI Infrastructure: What's Powering AI Today | IREN, Mirantis & NVIDIA | RAISE Summit 2026
- The competition for AI dominance is defined by the ability to efficiently convert power and electrons into high-speed compute and usable tokens, with current demand for AI infrastructure significantly outpacing supply.
- Future demand is projected to accelerate alongside technological breakthroughs in GPU efficiency and cost reduction, though the physical integration of power and data center access is expected to become increasingly difficult.
- IREN intends to leverage vertical integration to eliminate third-party dependencies for co-location and construction, aiming to secure a time-to-power advantage and rapidly bring capacity online.
- Significant demand is anticipated in Europe driven by hyperscalers and major North American labs, with data sovereignty and security requirements likely to foster growth in sovereign compute.
- Enterprise and smaller AI lab customers are expected to transition from bare metal consumption toward fully orchestrated platforms, a shift IREN will support alongside bare metal options for its largest clients.
- Mirantis aims to remove execution roadblocks and codify best practices to deploy software capable of keeping pace with the rapid advancement of both closed and open-source AI models.
- NVIDIA's AI-ready cloud ISV program is expected to establish a revolutionary open standard for AI consumption similar to the impact of APIs in the AWS era.
- The industry's pace requires open standards and open source to manage innovation, leading Mirantis to apply an open ecosystem approach across tens of thousands of GPUs to create a replicable reference implementation.
- IREN targets establishing itself as a major independent entity for flexible compute consumption, planning to optimize for specific use cases once the foundational infrastructure is operational.
- A six-gigawatt capacity pipeline is described as an initial phase intended to provide widespread availability, underlining the ecosystem-wide necessity for collaboration across energy, chips, infrastructure, and application layers to succeed.