The End of the AI Subsidy Is Hitting Enterprise Budgets
Enterprise AI is shifting from a subsidised experiment to a fully priced utility, and the impact on budgets is immediate. Anthropic’s recent move to unbundle its enterprise pricing—charging a base seat fee plus separate token consumption—signals that generous usage allowances are disappearing. OpenAI leaders have openly compared unlimited AI plans to unlimited electricity, arguing they no longer make economic sense. Because HR and business systems are built on top of these frontier models, repricing at the model layer flows straight into the renewals of HCM, collaboration and project management platforms. Early data is stark: the average enterprise AI budget has climbed from USD 1.2 million (approx. RM5.5 million) per year in 2024 to USD 7 million (approx. RM32.2 million) in 2026, while 65% of IT leaders report unexpected charges from consumption-based AI pricing. This is no longer a technology experiment—it is a structural budget problem.
From Per-Seat to Consumption-Based: Why SaaS Pricing Models Are Changing
The traditional per-seat SaaS pricing model is under pressure as AI agents start doing work that used to require human users. In project and work management, for example, an autonomous agent can draft briefs, triage backlogs and generate status updates without a human ever logging in. That means a product that truly automates work can actually reduce the number of seats a customer needs, undermining the vendor’s revenue model. To avoid this trap, platforms like monday.com are pivoting to hybrid structures that combine seats with credits tied to AI usage, tying part of their revenue to consumption rather than headcount. This shift toward consumption-based pricing introduces volatility for buyers: AI costs no longer scale neatly with employee counts but with how intensively agents run in the background. For finance leaders, that means traditional user-count-driven benchmarks are becoming unreliable guides for AI budget planning.
Agentic HR Platforms and the Hidden Cost of ‘Invisible’ Automation
HR technology stacks are rapidly becoming networks of AI agents rather than static systems of record. Major HCM platforms such as ADP, Workday and SAP SuccessFactors are building agentic AI directly into their core architecture, not as optional add-ons. These agents span recruiting, payroll, workforce administration, performance and talent development, and increasingly connect to external agents like Microsoft Copilot, Claude and Gemini. A single agentic workflow can trigger 10 to 20 large language model calls; multiplied across the HR lifecycle, that quickly compounds token consumption. Employees are also bringing their own AI assistants into daily work, adding more invisible usage. The result is a steep rise in enterprise AI costs that may not be obvious from license counts alone. For CHROs and CFOs, the critical task is to understand where agents are running, how often, and which HR workflows are driving the bulk of consumption-based spending.
AI Project Management Everywhere, Clear Productivity Gains Nowhere
Project and task management has become a test case for the disconnect between AI hype and measurable productivity. Access to AI project management tools has grown 50% year on year, yet only 1% of companies describe themselves as mature in AI deployment. Among senior leaders, just 19% report revenue growth above 5% from AI, and only 23% see any cost improvements. Meanwhile, vendors are racing to ship agentic features: monday.com has rebuilt its platform around native agents that join meetings and create action items; Asana offers AI Teammates; ClickUp promotes Super Agents; Adobe Workfront lets managers assign AI agents like human resources. Microsoft is using Copilot licensing as the gateway to more advanced automation in its work management suite. For finance and HR leaders, this proliferation makes ROI harder to pin down: AI features are everywhere in licenses, but demonstrable, trackable productivity gains often lag behind.
Budgeting for Enterprise AI: Governance, Workforce Design and Vendor Strategy
Rising enterprise AI costs demand a joint response from finance and HR. First, budgeting needs to move from static, per-seat expectations to dynamic AI budget planning based on consumption patterns, clear guardrails and usage thresholds. CHROs and CFOs should co-own dashboards that track AI utilisation across HR and project workflows, linking spending to outcomes such as reduced cycle times or improved service levels. Second, workforce planning must recognise that AI automates tasks, not entire jobs; HR should focus on redesigning roles around judgment, context and accountability rather than assuming whole functions will disappear. Finally, procurement strategies need to anticipate both incumbent HCM vendors and AI-native challengers. New entrants are targeting painful manual workflows around compensation, reorganisation and coordination, promising streamlined experiences. Incumbents still excel at managing complexity, but they will pass through model costs. Leaders who treat AI as an operational utility—with governance, metering and clear ROI expectations—will be better positioned as the AI bill comes due.