Nvidia's jensen huang: AGI milestones are no longer meaningful as AI advances

In a recent earnings call, NVIDIA CEO Jensen Huang stated that the concept of achieving artificial general intelligence (AGI) has become less significant as AI systems now perform tasks that match or exceed human capabilities. Huang emphasized that many tasks are already being handled by AI, making the milestone of AGI largely irrelevant. He noted that AI is no longer just waiting for human input but is now running autonomously and improving through recursive processes. Huang highlighted three key priorities for the industry: AI completing work with production value, generating profitable tokens, and increasing profitability through greater computational power. His remarks may also reflect a response to concerns about an AI bubble, as the sector invests heavily in AI infrastructure. Meanwhile, OpenAI CEO Sam Altman recently stated that OpenAI aims to internally achieve AGI by the end of 2026, defining AGI as a highly autonomous system that outperforms humans in most economically valuable tasks. Despite these advancements, critics argue that large language models cannot truly achieve AGI due to limitations in memory, logical understanding, and the tendency to generate false information. However, NVIDIA has benefited significantly from the AI boom, reporting $962 billion in revenue for its fiscal second quarter of 2026, a 106% increase compared to the previous year. The company attributed this growth to the AI-driven demand for memory, which has led to shortages and inflated prices, with the CFO, Collette Kress, noting that memory prices are expected to rise further in 2024. NVIDIA has also faced challenges due to rising memory costs, which have led to price increases for its graphics cards. Huang previously expressed similar views, stating that the tech industry has, in a way, already achieved AGI. He used the example of AI systems developing popular applications that quickly gain massive user bases, though these applications often lose traction rapidly. Huang acknowledged AI's limitations, stating that even with 10,000 such agents, the probability of successfully building NVIDIA remains zero.