May 16, 2025

AI Accounting: Are You Throwing Money Into the Void?

SPM Mythbusters Series

3 min read

Key Takeaways

Discover how teams in many organizations are adopting AI, without understanding the true costs or potential ROI.
See how AI accounting requires understanding usage metering, negotiated rates, and appropriate allocation methods.
Decide whether to treat AI as a non-labor resource (like electricity) or a labor resource (like a virtual employee).

Remember the dot-com boom? Or the mobile revolution? Companies poured millions into those technologies, often without a clear understanding of the value they'd deliver. Now, we're seeing the same frenzy with artificial intelligence (AI).

The mad rush to AI

Leaders in most organizations are scrambling to adopt AI, often driven by the fear of missing out (FOMO). Too often, they aren’t asking the tough questions: What's the return on investment (ROI)? What are we really spending on this?

In many organizations, executives don't have a clear handle on their AI costs. They might know the vendor and the contract, but they don't understand the true cost of AI.

The myth of free AI

Take OpenAI’s ChatGPT as an example. People used the free version, and got great results, without spending any money. Consequently, a lot of people started thinking AI was free. The problem is that AI is not free and if you have enterprise requirements, it’ll cost you … a lot.

Tokens are the new currency

The costs of AI can quickly add up.

Consider this: At a recent internal workshop, we used 24 million tokens in a single day. At common pricing rates, that translates to roughly $7.20. If you extrapolate that across 260 working days and 500 employees, the total cost quickly balloons to over $1 million. That’s money someone in your organization will want to see a return on. The question then becomes, how do you allocate that million-dollar investment to the value it's supposed to be delivering?

Allocation is key

One simple allocation method is to distribute costs based on department headcount.

For example, if your department has 5% of the company's employees, you'd be allocated 5% of the company’s total AI cost.

However, as AI spending grows, this simplistic approach may not provide sufficient insight. More thoughtful accounting and allocation methods become critical.

Treating AI like people vs. treating AI like a utility

A key decision for any organization investing in AI is determining how to account for it.

There are essentially two paths: Treat AI as a utility or treat it as a worker.

Should it be viewed as a general operating expense, like electricity?
Or should it be measured like the output of an employee and the results its labor accounted for?

Time to get serious about AI accounting

Track your AI usage and costs meticulously.
Negotiate favorable rates with your AI vendors.
Develop a clear allocation method that reflects the value AI delivers.

Don't let your AI investments become a black hole. Take control of your AI accounting and ensure you're getting a real return on your investment.

Tag(s): ValueOps , Clarity , Strategic Portfolio Management , AI , SPM

Brian Nathanson

Brian Nathanson is a recovering certified Project Management Professional now serving as the Head of Product Management Clarity at Broadcom. He is the host of several popular Clarity-related customer webcasts (Office Hours, Release Previews, and the End-to-End Modern UX Demos) and has conducted many hours of both...

Other resources you might be interested in

Blog December 22, 2025

Top 3 Trends Defining Network Observability in 2026

Discover the three specific trends that will define network observability in 2026. See how unified observability and predictive AI will shape the landscape.

Read Blog

Office Hours December 18, 2025

Rally Office Hours: December 18, 2025

Rally Office Hours features an AI-driven artifact breakdown tool that creates child items from features/stories. The Q&A covers capacity planning, forecasting and burnup charts.

View Recording

Blog December 17, 2025

Why 2025 Shattered the Old Rules of Network Management

This post reveals the five key lessons network operations leaders learned in 2025—and how they need to respond to be successful in 2026.

Read Blog

Blog December 17, 2025

The 2026 VMUG Report: Why Network Observability is the Heart of the New VCF Era

Get the top takeaways from the VMUG Cloud Operations and VCF User Experience Report 2026. See why network observability is key to successful VCF 9 migrations.

Read Blog

Video December 12, 2025

Automic Automation Cloud Integration: SAP S/4 HANA Application Jobs Integration

Simplify your SAP S/4HANA job management. Integrate with Automic Automation for central configuration, monitoring, and orchestration of all your enterprise jobs.

View Video

Video December 12, 2025

Automic Automation Cloud Integration: OpenSSH Integration

Master Open SSH automation. Use Automic Automation for centralized control, secure file transfer, command execution, and full job monitoring.

View Video

Office Hours December 11, 2025

Rally Office Hours: December 11, 2025

Discover Rally's new Ancestors field, static query box deprecation, non-conflicting saves, plus a dashboard demo and query writing tips.

View Recording

Blog December 10, 2025

3 Questions I Expect You to Ask Me

Ask these questions to gain a deeper understanding of a vendor. Find a partner who can solve today’s challenges and prepare you for what’s next.

Read Blog

Solution and Capabilities Briefs December 5, 2025

Carrier-Grade Network Observability: A Technology Brief for Telco Network Operations

Network Observability by Broadcom unifies data to provide contextual, AI-enabled insights for superior service availability, accelerated MTTR and improved MTTI, reduced operational costs, and the...

Read Technology Brief