The enterprise AI billing spree is coming. For a lot of companies, it's bigger than expected — and harder to justify than anyone planned.

When the leaderboard became the goal

Earlier this year, Uber built an internal leaderboard tracking which engineering teams used AI tools the most. It was meant to drive adoption.

It worked. Within four months, the company had burned through its entire 2026 AI budget. When asked if all that usage was translating to better products, the COO was candid: "It's very hard to draw a line between one of those stats and 'Okay now we're actually producing like 25% more useful consumer features.'"

The leaderboard tracked usage. What that usage was delivering never made it onto the scoreboard.

Most AI programs were designed to get people using the tools. Very few were designed around what those tools were supposed to change.

The vendors already changed the question

The platforms aren't waiting for enterprises to catch up.  

When Salesforce reported earnings last week, the CEO highlighted both revenue growth and a measure of actual work completed: 3.8 billion Agentic Work Units delivered to date — discrete tasks completed by AI agents.

HubSpot moved two of its flagship AI agents to outcome-based pricing in April — $0.50 per resolved conversation, $1 per qualified lead. You pay when it works.

Vendors are already betting on outcomes. Most of their customers are still running activity reports.

Start asking the right questions now

Budget reviews are coming. The question worth building toward: can you draw a direct line from a specific AI investment to a business result?

Start with a baseline. Before the next review, document where things stand today — sales cycle length, cost per qualified lead, conversion rate from MQL to close.

Without that starting point, any improvement stays invisible.

Then pick one metric per tool. The single number that would tell you whether a specific AI investment is earning its place. For an AI prospecting tool, that's qualified leads generated. For an AI sales coach, it's quota attainment. One tool, one number.

Finally, take a cue from the vendors themselves. Salesforce reports tasks completed. HubSpot charges per resolved conversation, per qualified lead. Both are asking the same thing: what did AI actually deliver? Your internal measurement should follow the same logic.

The only question is whether your team defines that answer before the CFO asks for it.