- Gartner, a leading business and technology insights company, forecasts a dramatic reduction in the cost of running large language models (LLMs) by 2030.
- Specifically, inference costs for trillion-parameter models are projected to drop by over 90% compared to 2025 for GenAI providers.
- This significant cost efficiency will be driven by advancements in semiconductors, infrastructure, and innovative model design.
- However, these substantial savings may not fully translate to enterprise customers, a nuance highlighted by IT Voice.
- This is because the increasing demand for more advanced AI capabilities is expected to lead to higher token consumption, potentially offsetting some cost benefits for businesses.
- As noted by Gartner Senior Director Analyst Will Sommer, while lower token unit costs enable advanced GenAI, these advancements will drive disproportionately higher token demand, causing overall inference costs to increase.
Gartner: LLM Inference Costs Plummet by 2030
Prepare for a revolution in AI economics: Gartner forecasts a dramatic 90% drop in large language model inference costs by 2030, fueled by breakthroughs in technology. Yet, this massive saving might not fully reach enterprise customers, as their growing appetite for advanced AI is expected to drive up token consumption, potentially negating the cost benefits.
Sources
3
Report an issue with this article
Please sign in to report issues with this article.
Live discussion about this article
Comments
Comments are disabled for this article.
