Blog • May 27, 2026
Agentic Workflow ROI: Uber's $400M AI Budget Lesson
Agentic workflow ROI explained through Uber's $400M AI spend admission. What B2B leaders in UK, SA & USA must do before the board asks the same question.
Agentic Workflow ROI: What Uber’s $400M AI Admission Tells Every B2B Leader
Uber burned through their entire 2026 AI budget by April. If a company with $400M, world-class engineers, and global scale can’t draw a straight line from AI spend to business value — your board is about to ask the same question yours can’t answer.
Uber’s Admission Is the Rule, Not the Exception
On May 26, 2026, Uber’s President Andrew Macdonald told The Verge something most executives only say in closed rooms: Uber had already burned through its full 2026 AI budget by April. The figure cited: roughly $400M, spread across machine learning, dispatch optimisation, fraud detection, and delivery workflow automation.
Macdonald’s exact words: “There is a big increase in AI spending this year. We are asking tough return questions. What are we getting out of all this spending?”
That question — publicly, from a company with more data and engineering depth than virtually any B2B org you compete with — should stop every tech leader cold. Google, Meta, and Samsung have all absorbed billions in AI capex without a clear public reckoning. Uber actually said it out loud. That makes it useful.
Why AI Spending Is Hard to Justify — Even at Scale
Here is the structural problem: most AI investment cycles are built around technology acquisition, not outcome definition. Teams select a platform, stand up a pilot, and report progress in features delivered — not revenue moved or cost removed. By the time the board asks for numbers, the budget is gone and the answers aren’t ready.
This is exactly why AI spending is hard to justify for B2B organisations. It is not that the technology doesn’t work. It is that the measurement framework is absent before the first invoice is paid.
In interviews across European and US listed tech governance teams, not one company could point to a standard return formula for generative AI projects. Not one. Measuring ROI of generative AI projects remains one of the least-solved operational problems in B2B tech right now. That doesn’t mean the metrics don’t exist — it means most teams haven’t built the loop that captures them.
The Right Questions for AI Budget Optimisation
Before your next board presentation, test your current AI initiative against these:
Can you show a before-and-after revenue number directly attributable to a specific AI workflow? Not efficiency gains. Revenue.
Do you have a traceable method for measuring operational impact of that model in production?
Can your team point to a specific workflow in live operations — not a demo, not a roadmap item?
If you pulled the plug today, would anything break in your revenue operations? If not, you have a pilot, not a workflow.
These aren’t gotcha questions. They are the exact questions any board in London, Johannesburg, or New York will ask in the next twelve months. Knowing how to justify AI budget to the board is no longer a nice-to-have skill — it is a survival requirement for every AI programme lead.
Agentic Workflows for Revenue Operations: What Actual ROI Looks Like
The distinction that matters here is not between “AI” and “no AI”. It is between deployed agentic workflows that touch revenue operations and experiments that never leave the sandbox.
Agentic workflows for revenue operations are specific: an AI agent that qualifies inbound leads against CRM data and routes them without human handoff; an orchestrated sequence that triggers outbound actions based on product usage signals; a RevOps automation layer that surfaces forecast variance before the weekly pipeline review. These are not features on a roadmap. They are operational assets with measurable outputs.
AI agents for GTM automation are already delivering measurable outcomes for B2B teams who built the measurement loop first. The pattern is consistent: define the operational problem, map the handoffs where revenue leaks, deploy an agent that closes the gap, measure within the existing CRM or data layer. Then — and only then — scale.
Custom AI workflow development for B2B organisations fails when the build precedes the business case. Uber is a cautionary tale not because they spent too much, but because the spend outpaced the measurement framework.
How to Move Beyond AI Pilots: A Practical Starting Point
If your organisation is currently running AI pilots without a clear path to production, these are the moves that change the calculus:
Start with the problem, not the tool. Map where revenue or operational capacity is actually leaking. AI workflow integration should solve a named problem, not demonstrate a capability.
Define success before you build. What does this workflow change in your pipeline, your cost-per-acquisition, or your support volume? Set the baseline now.
Deploy in increments with rollback built in. Agentic workflow automation is not a big-bang migration. Each agent should be independently valuable and independently measurable.
Tie every agent to a RevOps KPI. Not a project milestone. A live operational metric your board already tracks.
Audit before you scale. If you are considering whether to hire an AI workflow automation consultant in the UK, South Africa, or USA, the first engagement should always be an audit — not a build. You need to know where the leaks are before you plug them.
The Board Is Already Asking
Uber triggered a conversation that is now happening in every boardroom that has approved an AI budget in the last two years. The question is not “should we invest in AI?” anymore. It is “What did we actually buy?”
The organisations that will answer that question confidently are not the ones with the largest AI budgets. They are the ones who built agentic workflows directly into their revenue operations — with measurement built in from day one, not retrofitted after the budget is gone.
AI agents for RevOps consulting is not a new category. But the demand for consultants who can both build the workflow and defend the ROI at board level is accelerating sharply. In the UK, South Africa, and USA, we are seeing the same pattern: boards that approved AI budgets in 2024 and 2025 are now demanding transparent ROI dashboards before approving anything further.
Funnnl is called in when the numbers won’t add up and the board is asking why. As a consultancy that builds, audits, and deploys agentic workflow automation across the UK, South Africa, and USA, we work in production — not in demos. We don’t offer strategy decks that sit on a shelf. We build workflows that tie directly to the KPIs your board already tracks.
If you are facing the same question Andrew Macdonald voiced publicly — “What are we actually getting out of this?” — the answer starts with building the right loop, not buying more tools.
---
Source: Uber's AI investment is hard to justify — The Verge, May 2026