The debate over the future of AI scaling has ignited intense discussion among the field’s most influential figures, with leading minds divided on whether the era of exponential growth through larger models and more compute is nearing its limits. Geoffrey Hinton, widely regarded as the "Godfather of AI," remains cautiously optimistic about scaling’s continued relevance. Speaking to Business Insider, he acknowledged concerns raised by peers but argued that scaling isn’t obsolete. “I'm not convinced it's completely over,” he said, emphasizing that demand for data will persist. He pointed to the potential for advanced chatbots to generate their own training data—similar to how Google DeepMind’s AlphaGo and AlphaZero mastered Go by playing against themselves. “The equivalent for a language model is when it starts reasoning and saying, ‘Look, I believe these things and these things imply that thing, but I don’t believe that thing, so I’d better change something somewhere,’” Hinton explained. By using internal reasoning to detect inconsistencies in its own beliefs, a model could produce vast amounts of new, high-quality training data—effectively overcoming the bottleneck of limited human-curated data. This idea stands in contrast to views expressed by Ilya Sutskever, co-founder of OpenAI and one of Hinton’s former students. At the “Dwarkesh Podcast,” Sutskever declared that the AI industry is shifting away from scaling as the primary path forward. “Is the belief really: ‘Oh, it's so big, but if you had 100x more, everything would be so different?’ It would be different, for sure. But is the belief that if you just 100x the scale, everything would be transformed? I don't think that's true,” he said. He described the current moment as a return to the age of research—now powered by big computers. Sutskever also highlighted scaling’s appeal as a low-risk strategy for companies: rather than betting on uncertain breakthroughs, firms could simply invest in more compute and data to achieve incremental improvements. Yet not all AI pioneers agree with this pivot. Yann LeCun, another foundational figure in AI and former Meta chief AI scientist, has questioned the assumption that more data and compute automatically lead to smarter systems. “You cannot just assume that more data and more compute means smarter AI,” he stated in April. Like Sutskever, LeCun has since left Meta to launch his own startup. Alexandr Wang, now leading Meta’s superintelligence division, echoed these concerns in 2024, calling scaling “the biggest question in the industry.” His perspective reflects growing unease among leaders about relying solely on scale as a growth strategy. Meanwhile, Google DeepMind CEO Demis Hassabis maintains a more optimistic stance. At the Axios AI+ Summit in December, he argued that scaling remains essential for achieving artificial general intelligence (AGI). “The scaling of the current systems, we must push that to the maximum, because at the minimum, it will be a key component of the final AGI system,” Hassabis said. “It could be the entirety of the AGI system.” As the AI race intensifies, the debate over scaling continues to shape the direction of research, investment, and innovation across the industry.