AI industry shifts from model training toward autonomous agents and software

The artificial intelligence industry is undergoing a structural shift from model training to the deployment of autonomous AI agents and software applications. This transition is driving changes across the entire technology ecosystem, from hardware requirements to economic investment patterns.

In the consumer sector, Meta’s Muse app has seen rapid adoption, signaling a move toward AI as a personal agent capable of executing tasks like booking travel and managing emails. Simultaneously, a trend known as ‘AI for AI’ is emerging, where models are used to automate the research, coding, and training of subsequent generations. Major players including OpenAI, Anthropic, and Google are increasingly focusing on self-improving systems and automated research capabilities.

Economically, the investment cycle is expected to pivot. Morgan Stanley forecasts a shift from heavy hardware and chip spending toward software and services by 2028. However, some analysts warn of potential risks; Fitch Ratings suggests that a sharp decline in AI investments could trigger a US recession, while other reports note a growing gap between the rising costs of building AI and the decreasing prices offered to consumers due to market competition.