Goldman Sachs Sees 24‑Fold AI Token Rise by 2030 Amid Chip Shortage
Goldman Sachs expects autonomous AI agents to drive token consumption up 24‑fold to 120 quadrillion per month by 2030, warns of a 12‑18‑month high‑end chip shortage, and projects agentic AI use among knowledge workers rising from 12% in 2030 to 37% by 2040.
TL;DR
Goldman Sachs predicts autonomous AI agents will drive global token consumption up 24‑fold to 120 quadrillion per month by 2030. Analysts expect a high‑end semiconductor shortage to persist 12‑18 months, with full recovery possibly taking up to two years. They also forecast agentic AI use among knowledge workers will rise from 12% in 2030 to 37% by 2040.
Context
Goldman Sachs released a research note last week arguing that the spread of always‑on AI agents will push computing demand far beyond current levels. The bank says lower chip costs will accompany the surge, creating a "margin inflection" that could free cash for hyperscalers. Still, the note warns that supply constraints for advanced semiconductors could slow the rollout.
Key Facts
Token volume is projected to rise 24‑fold to 120 quadrillion tokens per month by 2030. Analysts expect high‑end semiconductors to remain tight for 12‑18 months, with full supply recovery possibly taking up to two years. Analysts project agentic AI usage among knowledge workers will reach 12% in 2030 and 37% by 2040.
What It Means
The projected token explosion implies data centers will need more processing power, even as chip makers work to expand capacity. If the shortage lingers, cloud providers may face higher costs or delayed deployments, potentially tempering near‑term growth.
Over the longer term, widening use of agentic AI could reshape how knowledge workers allocate time, shifting routine tasks to autonomous systems. Investors should watch whether semiconductor fabs can meet the 2030 demand timeline and how quickly enterprises move from pilot to production.
Continue reading
More in this thread
Conversation
Reader notes
Loading comments...