Generative AI has experienced rapid advances over the past two years, but concerns are now rising in Silicon Valley about a potential slowdown in progress. Key players, including OpenAI and Anthropic, are facing challenges, with OpenAI’s GPT-5 showing limited quality improvements and Anthropic delaying its Opus model. Similarly, Google’s Gemini is reportedly underperforming internal expectations. This slowdown raises questions about the validity of scaling laws, which suggest that more computing power and data yield better models indefinitely. Experts warn of a “data wall,” indicating companies may struggle to find new training data and are increasingly relying on synthetic data.
