{"id":50163,"date":"2026-05-07T04:11:05","date_gmt":"2026-05-07T08:11:05","guid":{"rendered":"https:\/\/overcentral.com\/en\/ai-industry-leaders-at-milken-conference-warn-of-chip-and-energy-bottlenecks\/"},"modified":"2026-05-07T04:12:27","modified_gmt":"2026-05-07T08:12:27","slug":"ai-industry-chip-energy-bottlenecks-milken","status":"publish","type":"post","link":"https:\/\/overcentral.com\/en\/ai-industry-chip-energy-bottlenecks-milken\/","title":{"rendered":"AI Industry Leaders at Milken Conference Warn of Chip and Energy Bottlenecks"},"content":{"rendered":"

The rapid advancement of artificial intelligence is confronting significant physical and logistical realities, a central theme underscored by industry leaders at the recent Milken Global Conference. While the demand for AI<\/a> capabilities seems limitless, the infrastructure required to support it\u2014specifically semiconductors and energy\u2014faces tangible bottlenecks. Christophe Fouquet, CEO of lithography giant ASML<\/a>, stated that despite massive acceleration in chip manufacturing, the market will remain supply-limited for the next several years, meaning tech giants will not receive all the processors they have ordered. This scarcity is reflected in the explosive growth of commitments; Google<\/a> Cloud’s Francis deSouza noted the company’s backlog of undelivered revenue nearly doubled in a single quarter to $460 billion, highlighting the intense, unmet demand for compute power.<\/p>\n

Beyond silicon, the industry’s voracious energy consumption presents a profound challenge, prompting exploration of unconventional solutions. DeSouza confirmed Google is seriously investigating orbital data centers to access more abundant solar energy, though engineering hurdles like heat dissipation in a vacuum remain. Both he and Fouquet emphasized that efficiency, not just raw power, is critical; Google’s strategy of co-designing its Tensor Processing Units (TPUs) with its AI models yields significant gains in computation per watt. However, the fundamental equation persists: more compute requires more energy, and that carries a substantial and unavoidable cost, a reality the industry must address as it scales.<\/p>\n

Amidst these scaling challenges, some are questioning the foundational architecture of modern AI. Eve Bodnia of Logical Intelligence is championing energy-based models (EBMs), which aim to understand the underlying rules of data rather than predicting linguistic patterns. She argues this approach, which powers her 200-million-parameter model, is more akin to human reasoning and is better suited for domains like chip design or robotics where understanding physical rules is paramount. This represents a fascinating alternative path in a field heavily focused on scaling ever-larger language models.<\/p>\n

The conversation also turned to the future of AI interaction, particularly the rise of AI agents<\/a>. Perplexity<\/a>‘s Dimitry Shevelenko described the company’s evolution into creating “digital workers” that can act on a user’s behalf. This shift raises critical questions about control and security, which Shevelenko addressed by emphasizing granular permission systems and mandatory user approval for agent actions, a necessary friction to build trust. Meanwhile, Applied Intuition’s Qasar Younis highlighted the unique geopolitical stakes of physical AI\u2014such as autonomous vehicles and defense drones\u2014which nations view as issues of sovereignty, leading to stricter controls than those placed on purely digital technologies.<\/p>\n

Finally, the panelists addressed societal impacts, particularly on future generations and the workforce. While optimistic about AI’s potential to solve grand challenges like disease and climate change, they acknowledged shifts in labor. DeSouza and Shevelenko focused on unleashed creativity and entrepreneurial accessibility, while Younis pointed to a more pressing issue: physical AI is poised to fill critical, growing labor shortages in fields like agriculture, mining, and trucking where human workers are increasingly scarce. The consensus suggests that while the path forward is constrained by hardware and energy, and complicated by architectural choices and geopolitical concerns, the transformative potential of AI across both digital and physical realms remains vast, demanding thoughtful navigation of the bottlenecks ahead.<\/p>\n","protected":false},"excerpt":{"rendered":"

The rapid advancement of artificial intelligence is confronting significant physical and logistical realities, a central theme underscored by industry leaders at the recent Milken Global Conference. While the demand for AI capabilities seems limitless, the infrastructure required to support it\u2014specifically semiconductors and energy\u2014faces tangible bottlenecks. Christophe Fouquet, CEO of lithography giant ASML, stated that despite […]<\/p>\n","protected":false},"author":6,"featured_media":85773,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/50163.png","fifu_image_alt":"AI Industry Leaders at Milken Conference Warn of Chip and Energy Bottlenecks","footnotes":""},"categories":[31],"tags":[],"class_list":["post-50163","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology"],"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/50163.png","fifu_image_alt":"AI Industry Leaders at Milken Conference Warn of Chip and Energy Bottlenecks","fifu_redirection_url":"https:\/\/edition.cnn.com\/2025\/04\/03\/africa\/africa-ai-cassava-technologies-nvidia-spc","_links":{"self":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/50163","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/comments?post=50163"}],"version-history":[{"count":0,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/50163\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media\/85773"}],"wp:attachment":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media?parent=50163"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/categories?post=50163"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/tags?post=50163"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}