Strategic Intelligence: The AI Journey - From Generative to Agentic
Description
Strategic Intelligence: The AI Journey - From Generative to Agentic
Summary
Agentic AI is at a tipping point and will deliver significant value to enterprise users by reducing costs, enhancing the customer experience, and increasing revenue. The barriers that once made enterprise AI agent deployment impossibly complex are being systematically dismantled. Agentic AI is expected to deliver a material increase in labor productivity across the broader economy.
Key Highlights
Summary
Agentic AI is at a tipping point and will deliver significant value to enterprise users by reducing costs, enhancing the customer experience, and increasing revenue. The barriers that once made enterprise AI agent deployment impossibly complex are being systematically dismantled. Agentic AI is expected to deliver a material increase in labor productivity across the broader economy.
Key Highlights
- AI agents are fueling a token explosion. Agents are becoming increasingly complex; they can reason and make decisions autonomously, and this capability is further enhanced via reasoning LLMs. Autonomous, multi-step agents are resource-intensive and consume huge amounts of data, resulting in massive increases in token consumption (both input and output volume) for LLMs.
- The rapid growth in token consumption exposes the challenges faced by LLMs and end users. If token consumption rises faster than cost efficiencies can reduce per-token costs, enterprise adopters will face significantly higher costs, and model builders will be under pressure to deliver cost efficiencies. This challenge will only grow as enterprise adoption scales across sectors.
- GlobalData has built a financial model to better understand the relationship between revenue, depreciation, operating expenses, and net profit before tax in the LLM business model. The key assumptions are consumer subscription growth and average revenue per user (ARPU), enterprise adoption, token-per-prompt growth, and cost-performance improvements in GPU technology.
- In our model, prompt inputs generate tokens, and this is used to generate daily usage for both consumers and enterprises. This is where agentic AI comes into play: the bulk of LLM as a service calls from enterprises will be generated by autonomous AI agents.
- We ran different scenarios using our model and determined that the $5.0 trillion of required AI capex between 2025 and 2030 might be less than $5.0 trillion if enterprise take-up of LLM and agentic AI is slow. The $5.0 trillion figure comes from the gigawatt power forecast for 2025-2030. Using our model, we then ran scenarios to understand what level of demand growth would require that much capacity.
- AI is driving huge tech investment, but the scale of that investment raises concerns. Cumulative AI infrastructure spending between 2025 and 2030 is estimated at $5 trillion. The absolute amount of capital expenditure (capex), the increase in debt issuance to fund it, and numerous circular funding deals raise concerns over the outlook for returns on these investments. This report will help you understand how agentic AI can help large language model (LLM) vendors achieve profitability and identifies the likely winners from the first phase of the AI investment cycle.
Table of Contents
46 Pages
- Executive Summary
- The Promise of Agentic AI
- The AI Infrastructure Investment Boom
- Understanding the LLM Business Model
- Insights From Our AI Market Model
- Glossary
- Further Reading
- Our Thematic Research Methodology
Search Inside Report
Pricing
Currency Rates
Questions or Comments?
Our team has the ability to search within reports to verify it suits your needs. We can also help maximize your budget by finding sections of reports you can purchase.
