
Artificial intelligence is transitioning from linear to exponential improvement, fundamentally altering business models across industries. Stephen Byrd, Global Head of Thematic and Sustainability Research at Morgan Stanley, highlights that current scaling laws—where a 10x increase in training compute yields a 2x gain in capability—are driving this nonlinear shift. This rapid advancement creates massive economic value for enterprises, as agentic AI workflows replace human tasks at a fraction of the cost. While compute, memory, and power remain scarce, these constraints grant hyperscalers significant pricing power. Early adopters are already realizing substantial efficiency gains, such as the automation of software coding, signaling an impending revenue inflection by 2026. As agentic AI adoption scales across knowledge-based roles, token demand will continue to explode, reinforcing the necessity for expanded AI infrastructure investment despite current capital expenditure concerns.
Sign in to continue reading, translating and more.
Continue