Key Takeaways
- • Grayscale Research projects AI token consumption could expand roughly 24-fold by 2030, driven by falling model costs and the rise of autonomous agents.
- • Agentic workflows can require five to fifty times as many tokens as a basic chatbot exchange, depending on task complexity.
- • Grayscale concludes that expanding AI activity favors owners of scarce compute resources, including powered data centers and grid-connected facilities.
- • Physical constraints in power generation, data center construction, and GPU supply mean infrastructure lags digital demand, supporting higher pricing for compute capacity.
- • Goldman Sachs projections cited in the discussion attribute much of the expected 24x rise largely to agent-driven activity.
Summary
Grayscale Research, in a recent analysis, projects that overall AI token consumption could expand roughly 24-fold by 2030. The report attributes this growth not only to broader usage but to structural shifts in how AI systems operate, particularly the move from single-pass chatbots to autonomous agents that pursue multi-step objectives by sequencing tasks, gathering information, evaluating options and refining outputs.
According to the analysis, these longer workflows can require five to fifty times as many tokens as a basic chatbot exchange, depending on complexity. Preparatory work for a client meeting, such as reviewing prior notes, assessing portfolio shifts and composing an agenda, illustrates how token usage multiplies quickly. Always-on systems operating continuously toward user goals could extend this pattern further, generating sustained compute loads rather than isolated sessions. OpenAI's recently introduced "dots," described as always-on agents capable of working around the clock, along with new lower-cost models from various providers, are cited as additional drivers.
Falling per-token prices, partly driven by competition from open-weight alternatives, make more automated workflows economically practical. Goldman Sachs projections cited in the discussion attribute much of the expected 24x rise largely to agent-driven activity, with inference already growing rapidly. Zach Pandl, Grayscale's Head of Research, has said AI token demand could rise 24x by 2030 on the back of falling costs and broader use.
The physical infrastructure delivering this computation faces real constraints. Power generation and transmission, data center construction and GPU supply all require extended timelines for planning, permitting, building and deployment, while digital demand can scale almost immediately once software is available. Some specialized providers have recently signed new contracts at higher prices for compute capacity, consistent with scarcity amid rising usage.
Grayscale Research concludes that expanding AI activity is positioned to favor owners of scarce compute resources. Efficiency gains at the software layer may not fully offset the volume increase from broader adoption, and owners of powered data centers and grid-connected facilities capable of converting electricity into reliable computing capacity stand to benefit as demand outpaces physical build-out.
According to the analysis, these longer workflows can require five to fifty times as many tokens as a basic chatbot exchange, depending on complexity. Preparatory work for a client meeting, such as reviewing prior notes, assessing portfolio shifts and composing an agenda, illustrates how token usage multiplies quickly. Always-on systems operating continuously toward user goals could extend this pattern further, generating sustained compute loads rather than isolated sessions. OpenAI's recently introduced "dots," described as always-on agents capable of working around the clock, along with new lower-cost models from various providers, are cited as additional drivers.
Falling per-token prices, partly driven by competition from open-weight alternatives, make more automated workflows economically practical. Goldman Sachs projections cited in the discussion attribute much of the expected 24x rise largely to agent-driven activity, with inference already growing rapidly. Zach Pandl, Grayscale's Head of Research, has said AI token demand could rise 24x by 2030 on the back of falling costs and broader use.
The physical infrastructure delivering this computation faces real constraints. Power generation and transmission, data center construction and GPU supply all require extended timelines for planning, permitting, building and deployment, while digital demand can scale almost immediately once software is available. Some specialized providers have recently signed new contracts at higher prices for compute capacity, consistent with scarcity amid rising usage.
Grayscale Research concludes that expanding AI activity is positioned to favor owners of scarce compute resources. Efficiency gains at the software layer may not fully offset the volume increase from broader adoption, and owners of powered data centers and grid-connected facilities capable of converting electricity into reliable computing capacity stand to benefit as demand outpaces physical build-out.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.