Grayscale’s Zach Pandl says AI token consumption could increase 24-fold by 2030 as lower costs expand adoption and usage. AI agents may consume five to 50 times more tokens than standard chat
AI token consumption could increase 24-fold by 2030 as cheaper models and autonomous agents expand usage, according to Grayscale Head of Research Zach Pandl. However, the physical infrastructure needed to run these systems remains constrained. Grayscale says rising demand for computing power could outpace efficiency gains, increasing pressure on available data centers, electricity, and GPUs.
https://twitter.com/Grayscale/status/2108952119294480810?s=20Declining costs are making artificial intelligence more accessible for applications and automated workflows. Anthropic’s latest Sonnet model costs up to 30% less per task, according to the company.
Meanwhile, competition from lower-priced open-weight models is putting pressure on token prices. Tokens measure the text AI models process and generate, with developers paying for their consumption.
As prices fall, businesses may find more uses for AI that previously cost too much. Grayscale’s research report, Investing for the Compute Bottleneck, examines how cheaper intelligence could support wider adoption.
OpenAI’s recently introduced “dots” are always-on agents designed to work toward users’ goals around the clock. Unlike chatbots that answer individual questions, agents can complete multiple steps within a single workflow.
For example, preparing for a client meeting could require reviewing notes, analyzing portfolio changes, and drafting an agenda. According to Grayscale, these tasks can consume five to 50 times more tokens than a chatbot interaction, depending on complexity.
Consequently, agents could drive substantial growth in inference, the computation required to run AI models. Goldman Sachs projects token consumption could rise 24-fold by 2030, largely because of agent adoption.
Although AI models are becoming more efficient, total computing demand can still rise when usage grows faster. However, supplying the necessary infrastructure takes time. Power availability, data centers, and GPUs remain essential to supporting expanding workloads.
Their limited supply could constrain how quickly computing capacity grows alongside AI adoption. Grayscale also cited CoreWeave, which recently reported signing new contracts at higher compute prices. The example highlights continued pricing pressure despite declining costs for AI tasks.
Pandl’s analysis argues that cheaper models and more capable agents could increase usage enough to outweigh efficiency gains. Under this scenario, demand could benefit owners of scarce physical computing infrastructure.
