Investment bank Goldman Sachs published a comprehensive report on the future of agentic artificial intelligence technologies that stand out in the world of artificial intelligence. The bank predicts that this technology will lead to a significant increase in token consumption, and that by 2030, consumption will increase 24 times compared to 2026 levels.

Artificial intelligence queries are expected to reach 23 billion in 2030, from 5 billion in 2025. The basis of this increase is the use of agentic artificial intelligence, which is created by software bots that work tirelessly and manage the interaction between machines.

Agentic Artificial Intelligence and Its Effects on the Hardware Industry

According to the Goldman Sachs report, agentic artificial intelligence software can verify data 24 hours a day, 7 days a week, and make transactions using external tools. This creates a new source of optimism for processor manufacturers such as Intel, AMD and Arm.

Huge capital expenditures on AI infrastructure are predicted to become profitable thanks to increased token usage. The use of this technology is rapidly becoming widespread, especially in areas such as supply chain management, software development and law.

The bank predicts that the use of agentic artificial intelligence will increase token consumption by 55 times by 2040. Currently, less than a quarter of businesses use this technology, and the transition to fully autonomous mode is still in its infancy.

Cost Reduction and Data Quality Warning

Thanks to the new generation chips offered by platforms such as NVIDIA, AMD and Trainium, token computation costs are decreasing by 60 to 70 percent on an annual basis. This cost advantage is enabling AI and cloud service providers’ gross margins to turn positive in the first half of this year.

However, Goldman Sachs also draws attention to an important risk factor in addition to this optimistic picture. Poor data qualityThis may lead to inefficient consumption of resources and failure to obtain expected returns. Do you think agentic artificial intelligence technologies will be able to provide the expected efficiency in the business world?