Conference Presentation, Panel
Clarifai, Oracle, TensorWave, Cherry & Vaire: Sovereign AI in a Borderless Tech World
- A significant portion of AI businesses at the current conference are projected to cease operations within the next two years.
- TensorWave anticipates expanding its GPU fleet from tens of thousands to hundreds of thousands within one year.
- Training clusters in the AMD NeoCloud ecosystem are expected to be rapidly repurposed into inference pools as new Blackwell clusters come online.
- Refurbished GPU fleets used for inference are projected to become economically unviable as electricity costs rise.
- Most companies are predicted to develop their own Large Language Models within two years or less following concerns over external security risks.
- Saudi Arabia and the UAE may enter the global AI race within five years contingent on successful infrastructure development.
- Investment narratives in the Middle East are expected to rely on government-level support, distinguishing them from corporate-driven stories in the US.
- US power generation capacity is forecast to be insufficient for current AI infrastructure plans without a new top-down initiative.
- The US lacks a unified government coalition comparable to China's strategic mindset to secure a lead in the AI race by 2030.
- Policymakers are expected to approve hundreds of billions of euros in spending agreements while failing to sustain the operational work required for asset functionality.
- Customers in the open market are predicted to prioritize open-source or lower-cost solutions over sovereign AI if those alternatives prove sufficiently capable.
- The industry is shifting toward a multi-winner model necessitating long-term orchestration strategies that integrate both sovereign and general-purpose models.