The rise of agentic artificial intelligence is fundamentally changing how IT services are billed, pushing traditional hourly and per-seat models towards outcome-based and usage-based pricing. This shift has led to a significant decline in IT service contract values globally. Total contract value for typical five-year deals shrank by 20% to 35% in 2025 and is projected to fall further by 32% to 44% in 2026.[m]
Old Billing Models Under Pressure
For years, IT service providers charged clients based on the number of human workers, or "seats," and the hours they spent on projects. This predictable model is now under direct attack. Agentic AI tools automate tasks like code generation, sales work, legal reviews, and data analysis. These tools allow a smaller number of human workers to deliver much higher output.As a result, clients are questioning why they should pay for thousands of seats when AI can handle many workflows.[communicationstoday+1]
Blended billing rates for global IT services contracts have been steadily declining for the past 24 months in both the United States and India. This affects roles such as web and mobile developers, full-stack engineers, remote desktop support, and cyber defense professionals. For example, US billing rates for intermediate web/mobile developers with four to six years of experience dropped to $75-$91 per hour in late 2025, down from $77-$94 in late 2024. India rates for similar roles fell to $21-$29 per hour from $22-$29 in the same period. Even premium technology consulting rates, which typically resist price pressure, have softened, with US ranges slipping to $138-$181 per hour from $141-$182.[m]
Clients are increasingly demanding large productivity-linked discounts from IT service firms. These discounts must then be reinvested into expanding project scope or building new AI use cases. This pressure has caused a sharp compression in the total value of long-term contracts.Industry analysis indicates that traditional per-seat models become obsolete when one AI agent can operate with "less than half as many people" needed for the same tasks.[m+1]
The Power of Autonomous Agents
Agentic AI systems are different from earlier AI. They do not just respond to prompts. Instead, they can observe their environment, plan multi-step actions, make independent decisions, and adapt based on feedback.They act like digital operators, executing complex goals with minimal human oversight.[blog+7]
These autonomous agents are being deployed across many IT functions. They handle intelligent process automation, cybersecurity, IT support, autonomous software development, and smart cloud infrastructure management.For instance, in IT operations, agentic AI can automatically triage incidents, infer root causes, and deploy fixes, significantly reducing the time it takes to resolve issues.They can also predict infrastructure failures and dynamically allocate cloud resources to match demand.[itechnotion+6]
In financial IT operations, agentic AI automates decision-making by analyzing transactions, ensuring compliance, and generating reports.For billing and payments, these agents can analyze payment trends, predict cash flow, and recommend adjustments to billing cycles. They can also resolve complex customer inquiries and disputes by gathering data from various systems.Infosys BPM, a major outsourcing provider, now uses AI agents to process invoices end-to-end without human intervention, completing tasks in seconds that humans need hours to do.[itechnotion+2]
New Pricing Models Emerge
The shift from human time to machine execution forces the service industry to reconsider what it charges for.Clients now care more about outcomes than how many tokens or API calls an agent uses.This has led to the emergence of new pricing models:[medium+1]
- Outcome-Based Pricing: This model ties payment directly to specific business results. Examples include revenue generated, cost savings achieved, productivity improvements, or resolved issues. Salesforce, for example, prices its Agentforce product at about $2 per conversation beyond included allowances. Intercom charges $0.99 for each successfully resolved customer support conversation by its AI agent, Fin. This aligns vendor and customer incentives, as customers only pay when the AI delivers results.
- Usage-Based (Consumption-Based) Pricing: Here, customers pay for the actual resources consumed by the AI, such as API calls, compute time, data processed, tokens used, or tasks completed. This model offers transparency and scales costs with actual utilization.
- Per Agent Pricing: Some models charge for the agent's availability, similar to hiring an employee. Nullify, a security vulnerability fixing service, charges $800 per agent per year.
- Per Activity/Output Pricing: This involves paying for specific actions taken by the AI or for a tangible output it produces. Replit, for instance, charges $0.25 per "checkpoint" or meaningful change in code.
- Hybrid Models: Many companies are adopting hybrid approaches, combining a predictable base platform fee or subscription with variable usage-based components. This offers customers cost predictability while capturing value from high-intensity users.
These new models address the fact that a single customer might deploy hundreds of agents, creating variable workloads and consuming significant infrastructure without necessarily increasing "seat" count.[zenskar]
Market Growth and Future Outlook
The agentic AI market is experiencing rapid growth. It was valued at $6.96 billion in 2025 and is projected to reach $57.42 billion by 2031, growing at a compound annual growth rate of 42.14%.Spending on agentic AI is expected to reach $1.3 trillion by 2029, with a year-over-year growth of 31.9% between 2025 and 2029.This surge signals a transformation in enterprise IT budgets, shifting investment strategies towards AI-based products and services.[mordorintelligence+2]
Gartner predicts that by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention. This could lead to a 30% reduction in operational costs.Most enterprises are achieving positive return on investment within four to six weeks of well-designed pilot programs.Subscription-based agentic AI models in Business Process Outsourcing (BPO) typically deliver 70% operational cost reduction while maintaining service quality, with break-even usually occurring within four to six weeks.[gartner+2]
However, the transition is not without challenges. Some buyers struggle to define clear, measurable outcomes for AI, worry about cost predictability, and find it difficult to align on value attribution with vendors.Over 40% of agentic AI projects are predicted to fail by 2027 because legacy systems cannot support modern AI execution demands.[bcg+1]
Despite these hurdles, the momentum for agentic AI is undeniable. Organizations that master these new pricing dynamics will capture disproportionate value in the evolving AI economy.The shift from traditional software licensing to value-based, outcome-driven models is not just an incremental update; it is a fundamental paradigm shift for the IT services industry.[scholarspace+1]
