Unpacking The AI Ecosystem’s Bottlenecks
Leaders from across the AI supply chain outlined key constraints shaping industry development during a panel at the Milken Global Conference in Beverly Hills. Discussion focused on semiconductor supply, energy requirements for data centres and limitations in current AI architectures.
Silicon Shortages And The Race For Chips
Christophe Fouquet, Chief Executive Officer of ASML, said chip production remains constrained despite increased investment. He noted that demand from companies such as Google, Microsoft, Amazon and Meta continues to exceed supply, with shortages expected to persist for two to five years.
Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram
Data Centers And The Energy Imperative
Francis deSouza, Chief Operating Officer of Google Cloud, pointed to rising demand for computing infrastructure alongside growing energy constraints. He referenced exploration of alternative data centre models, including concepts for space-based infrastructure, while noting engineering challenges such as heat management and system efficiency.
Beyond Silicon: The Data And Real World For Autonomous Systems
Qasar Younis, Co-Founder and Chief Executive Officer of Applied Intuition, highlighted limitations in training data for autonomous systems. Real-world data across sectors such as automotive, defence and industrial applications remains difficult to replicate through simulation, creating gaps between model performance and real-world conditions.
Redefining Intelligence
Eve Bodnia, quantum physicist and founder of Logical Intelligence, presented energy-based models as an alternative to current AI systems. These models aim to capture underlying data structures more efficiently and may be applied in areas requiring modelling of physical systems.
Building Trust With Digital Workers
Dimitry Shevelenko, Chief Business Officer at Perplexity, described the development of AI agents as “digital workers” integrated into enterprise environments. Focus remains on access controls, security layers and monitoring systems to manage how these agents operate within organisations.
The Intersection Of AI And National Sovereignty
Discussion also addressed geopolitical considerations linked to AI deployment. Younis said physical AI systems, including autonomous vehicles and drones, introduce regulatory and security challenges. Fouquet added that semiconductor manufacturing capacity remains a constraint in global competition, including in China.
Optimism And A Path Forward
When asked about the effect of AI on future generations’ critical thinking, panelists pointed to both risks and opportunities. Francis deSouza, Chief Operating Officer of Google Cloud, said advanced tools could support creative problem-solving across areas such as healthcare and infrastructure. Dimitry Shevelenko, Chief Business Officer at Perplexity, noted that entry-level roles are likely to change as automation expands, while broader access to technology enables more people to build and deploy new products.
Taken together, the discussion points to a system shaped by constraints in chips, energy and data, alongside continued changes in how AI models are designed and deployed. How these factors evolve will determine the pace of adoption and the range of practical applications across industries.
