Goldman Sachs: AI investment scale is still underestimated, token consumption may increase 24 times by 2030
Goldman Sachs predicts that spending by hyperscale data center operators on AI will significantly exceed current market expectations, potentially reaching $1.4 trillion by 2027. This surge is driven by an anticipated 24-fold increase in token consumption by 2030, fueled by the rise of enterprise agents. The increased demand for tokens will consequently boost the need for data centers, chips, networking equipment, and power infrastructure. AI
IMPACT This forecast suggests a substantial increase in demand for AI infrastructure, impacting hardware, networking, and power sectors.