IT costs are not reducing; they are rotating. We are witnessing a “Budget Rotation” where funds are being aggressively moved from traditional “shelfware” SaaS to “active” AI consumption.
Why the “Efficiency Era” Won’t Lower Your IT Bill
The prevailing narrative in boardrooms is that Artificial Intelligence will drive cost savings by automating labor. While AI does reduce labor costs (headcount), it is simultaneously creating a massive new cost center in IT: Compute Consumption.
We are currently in the third major wave of enterprise software spending. We are seeing “repurposing” rather than “reducing” as per Jevons Paradox: as technology increases the efficiency of a resource (intelligence), consumption of that resource rises to meet the new capacity, keeping total spend stable or increasing it.
1. The Three Eras of IT Spending
To understand why costs aren’t dropping, we must look at the structural shift in what is being bought.
| Era | Primary Asset | Payment Model | Key Metric |
| 1. On-Premise | Infrastructure | CapEx (Upfront) | Server Uptime |
| 2. Cloud SaaS | Software Access | OpEx (Subscription) | User Seats (Headcount) |
| 3. AI & Agents | Intelligence | OpEx (Consumption) | Work Outcomes |
The Shift: From “Renting Tools” to “Hiring Machines”
In the SaaS era (2010–2023), you paid for a tool (e.g., Salesforce, Slack) that a human used. If you fired the human, you cancelled the seat.
In the AI era (2024+), you are paying for the work itself. If you fire the human, you must spin up an AI Agent to replace them. The vendor captures the value of that “digital worker,” preventing your total costs from plummeting.
2. The “SaaS Squeeze”: Where the Money Comes From
CFOs are currently engaged in “SaaS Optimization” to fund AI initiatives without raising the total budget. They are cannibalizing the “bloated” middle layer of the SaaS stack.
- The Statistic: According to Zylo’s 2025 SaaS Management Index, organizations utilize only 47-49% of their provisioned SaaS licenses. Roughly 51-53% of paid software seats are “shelfware” (unused), costing companies an estimated $21 million annually in waste.+1
- The Action: Companies are auditing these unused seats to reclaim budget.
- The Result: A company saving $1M by cutting unused Zoom or Asana licenses is immediately reallocating that $1M to purchase Microsoft 365 Copilot or OpenAI Enterprise credits. The net spend remains flat, but the vendor mix changes.
Industry Insight: “SaaS Optimization” is now the top concern for IT leaders in 2025, specifically to free up budget for AI experimentation. (Source: BetterCloud / Gartner)
3. The Pricing Pivot: Why Costs Won’t Drop
The strongest evidence for your “repurposing” theory lies in how vendors are changing their pricing models. They know AI will reduce human headcount, so they are moving away from “Per Seat” pricing to ensure they don’t lose revenue.
A. The “AI Tax” on Seats (The Hybrid Model)
Incumbents like Microsoft and Salesforce are protecting their revenue by adding a premium “AI Tax” to existing seats.
- Example: Microsoft 365 Copilot costs $30 per user/month on top of the standard E3/E5 license. This represents a 60–100% price increase per employee for the privilege of AI access.
- Impact: Even if you reduce headcount by 10% due to efficiency, the remaining 90% of employees are now costing you 60% more per person.
- Math: 100 employees @ $30 = $3,000.
- AI Scenario: 90 employees @ $60 ($30 base + $30 AI) = $5,400. (Total cost INCREASES).
B. The Move to Consumption (The Outcome Economy)
For “Agentic AI” (AI that runs autonomously), vendors are introducing consumption-based billing.
- Salesforce Agentforce: Pricing is moving toward roughly $2 per conversation or outcome, rather than just a flat fee.
- Microsoft Copilot Studio: Uses “Copilot Credits” (metered billing) for autonomous agents.
- The Trap: As agents become successful, you use them more. Unlike a flat SaaS fee, AI costs scale linearly with productivity.
4. Market Statistics: The “Budget Rotation” in Numbers
The data supports the view that we are seeing a rotation, not a reduction.
- IT Spend Growth: Gartner projects worldwide IT spending to grow by roughly 9-10% in 2025, reaching over $5.6 trillion. This growth is driven by “price hikes” and “software,” not just new units sold.
- SaaS Spend per Employee: Despite layoffs in the tech sector, the average SaaS spend per employee has risen to approximately $4,830 (up from ~$4,600). We are paying more for software for fewer people.
- The Shadow AI Spend: Shadow IT (unapproved spending) is rising again, with 15-30% of employees purchasing their own AI tools (e.g., ChatGPT Plus, Midjourney) on corporate cards, creating a hidden cost layer that OpEx budgets must eventually absorb.
5. Strategic Conclusion: The “Labor-Software Merger”
Your assumption is correct. We are not entering an era of “Cheap IT.” We are entering an era where Software Budgets and Labor Budgets are merging.
In the On-Prem and SaaS eras, “Software” was a tool to support the “Labor” budget.
In the AI era, “Software” (Agents) is the Labor budget.
The “Repurposing” Roadmap:
- Phase 1 (Optimization): Cut unused SaaS seats (saving 15-20% of IT budget).
- Phase 2 (Reallocation): Spend those savings on AI “Copilots” for remaining staff (Cost Neutral).
- Phase 3 (Replacement): Reduce human headcount, but transfer those salary funds to “Agent Consumption” credits (Cost Shift).
Final Verdict: Do not forecast a reduction in your IT budget. Forecast a 20-30% increase in “Compute/AI” spend, fully funded by a corresponding decrease in “SaaS Waste” and “Entry-Level Headcount.”