The rise of agentic artificial intelligence (AI) is set to transform computing, with projections suggesting that token consumption will soar to 120 quadrillion per month by 2030. This surge, particularly related to large language models (LLMs), could significantly boost cash flow for major tech companies, according to Goldman Sachs Research.
Token Consumption Set for Explosive Growth
Jim Schneider, a senior equity analyst at Goldman Sachs, noted that the anticipated rise in token usage will coincide with a pivotal moment in the ongoing AI expansion story. The integration of autonomous agents—capable of executing sequential tasks rather than just responding to queries—will drive enterprise adoption ahead of consumer applications. This shift indicates a growing demand for the complex interactions that agentic AI can provide.
The increase in token consumption is especially remarkable; Schneider predicts a staggering 24-fold growth from current levels by the decade's end. This leap reflects the increasing sophistication and deployment of AI solutions across various industries, where each interaction with these agents requires multiple tokens for processing.
Implications for Hyperscalers
Amid concerns about the sustainability of capital expenditures—especially given the financial pressures on hyperscalers—Schneider stresses that improving gross margins could ease some of these challenges. He pointed out, "The concern in the generalist investor community is the sustainability of capex because the free cash flows of hyperscalers have been compressed. What fixes that? The answer lies in the underlying economics of the problem." By enhancing operating cash flow through higher margins, tech companies can gain more flexibility in their spending.
As computing costs continue to decline, the environment seems primed for what Schneider refers to as a period of 'margin inflection.' This could enable companies to manage their investments in infrastructure and technology more effectively, fostering a sustainable operational model in an increasingly competitive market.
Challenges and Opportunities in Agentic AI
Despite the excitement surrounding agentic AI, quantifying its potential impact remains challenging. Schneider remarked that while there is significant enthusiasm for these autonomous agents, concrete metrics to measure their business outcomes are still being developed. To better grasp their implications, Goldman Sachs has modeled common use cases, such as travel bookings and customer service interactions, to simulate the real-world application of agentic AI.
In practical terms, tokens—considered units of compute or information—are essential to the operation of these AI systems. Schneider explained, "Agentic AI requires a lot of tokens because many queries are repeated in sequence. It’s like taking a simple chatbot request and blowing it up 10-fold, 20-fold, 50-fold." This exponential increase in token usage highlights the necessity for advanced AI infrastructure capable of meeting such demands.
The Road Ahead
With the anticipated surge in token consumption, tech companies and investors will need to address the complexities of integrating agentic AI into existing frameworks. As enterprises increasingly adopt these advanced AI solutions, the focus will shift to ensuring that infrastructure can support the expected growth in demand. While the excitement around agentic AI is promising, it requires a clear understanding of its economic implications and operational feasibility.
As the AI field evolves, the relationship between technological advancements and financial strategies will be critical for hyperscalers looking to take advantage of the emerging potential of agentic AI.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
