‘Almost unlimited’: Execs says AI demand remains strong even as enterprises move to ‘valuemaxxing’ (www.cnbc.com)

AI executives reported to CNBC that demand for artificial intelligence remains robust, contradicting concerns of overcapacity as businesses become more cost-conscious about AI investments. Pat Gelsinger, former Intel CEO, emphasized that the potential economic benefits of AI are virtually limitless across industries, marking energy availability as the main constraint. Despite fluctuations in AI chip stocks and speculation about Meta's decision to sell excess computing capacity, many companies continue to experience unprecedented demand for AI compute infrastructure. Notably, firms like Nebius and Lumentum are struggling to meet this surge, with Lumentum's products sold out for the next five years, highlighting a significant supply-demand mismatch in data center resources. As enterprises shift from a "tokenmaxxing" strategy—encouraging indiscriminate use of AI for maximum output—to a focus on value maximization, the AI landscape is evolving. CFOs are urged to prioritize return on investments, making it crucial for companies to wisely evaluate the cost versus benefit of various AI models. This rationalization process, akin to trends seen in previous technology cycles, suggests a future where specific AI workloads will be matched increasingly to the appropriate computational resources, ensuring efficiency and optimization in AI deployments.