Five architects of the AI economy explain where the wheels are coming off

This week, five influential figures from various sectors of the AI supply chain gathered at the Milken Global Conference in Beverly Hills. They engaged in a thought-provoking discussion with our editor, addressing critical issues like chip shortages and the potential inadequacies of the existing technological architecture. The panel featured notable leaders including Christophe Fouquet, CEO of ASML, a key player in chip manufacturing; Francis deSouza, COO of Google Cloud, who oversees significant infrastructure projects; Qasar Younis, co-founder and CEO of Applied Intuition, a company specializing in physical AI; Dimitry Shevelenko, chief business officer of Perplexity, an AI-driven search and agent company; and Eve Bodnia, a quantum physicist challenging the conventional frameworks in AI with her startup, Logical Intelligence. Fouquet opened the discussion by highlighting the ongoing challenges in chip production, emphasizing a projected supply limitation for the next few years. Despite advances in manufacturing, major tech companies like Google and Microsoft will face difficulties in acquiring the chips they need. DeSouza added to the conversation by underscoring the rapid growth of Google Cloud, noting a staggering revenue increase and a backlog of committed revenue that has doubled. This situation showcases the genuine demand for AI technologies. Younis shifted the focus from silicon shortages to data constraints, emphasizing that the real challenge lies in gathering sufficient real-world data for training AI models. He pointed out that synthetic simulations cannot fully replicate the complexities of the physical environment. Energy consumption emerged as another pressing issue, with DeSouza revealing Google's exploration of space-based data centers to address energy constraints. However, he acknowledged the engineering complexities involved in such a venture. Bodnia presented a fresh perspective by discussing energy-based models (EBMs) that her company is developing, which differ fundamentally from traditional AI models. Her approach aims to mimic human cognitive processes more accurately and allows for real-time updates as new data becomes available. The panel also delved into the evolution of AI applications, with Shevelenko describing Perplexity's transition from a search tool to a digital workforce assistant, raising important questions about control and permission management. Younis brought a geopolitical dimension to the conversation, discussing the implications of physical AI technologies on national sovereignty. He noted that countries are increasingly concerned about foreign control over AI systems that operate within their borders. As the discussion wrapped up, the panel addressed concerns about the impact of AI on future generations' critical thinking skills. Overall, the consensus was cautiously optimistic, with experts highlighting the potential of AI to solve complex global issues while also acknowledging the challenges it presents in various sectors.