Is Excel Dying? Finance AI Changed Who Builds It

Is Excel dying? Finance AI changed what builds it, not whether we use it. Excel isn’t going anywhere.

Is Excel dying? Finance AI changed what builds it

Seven percent.

That’s the smallest bucket in Paul Barnhurst’s poll on AI use in Excel. 503 votes. Nearly half said less than 25% of the time. Another 15% said never. Six in ten are barely in the door. And 7% said they build their Excel files with code.

I voted “more than 50%.” Nearly all of that time is on the build side. The workbook architecture, the model logic, the Python that replaced the workbook entirely. Once it’s built, the math stays deterministic. The formulas compute the same way every time. The output is mine.

I said it on air earlier this year: I’m not using AI for forecasting. I’m using AI to create the tools for forecasting and reporting.

Barnhurst’s other finding: fewer than 1 in 5 use Excel less than they did a year ago. Excel isn’t going anywhere. What builds it is changing, and almost nobody’s crossed into the building side yet.

The cockroach of enterprise software

Excel is the cockroach of enterprise software that survived all nuclear winter. Neither Python, Tableau, Power BI, nor even AI could kill it.

We’ve all tried the “AI replaces spreadsheets” experiment. You prompt ChatGPT to adjust a model. It rebuilds everything. Nothing ties. You fight it for two hours… Then reopen Excel and fix it in 20 minutes.

That’s not failure. That’s judgment.

My team now uses AI to write formulas, prep data, even draft dashboards.

But we still deliver in Excel because when your CFO asks “Where did this number come from?” you need to click through the logic, not re-run a prompt.

This is how we actually work in 2025.

Nobody taught me financial modeling

Nobody taught me financial modeling.

I learned Excel by trial and error. Built hundreds of models. Broke hundreds more. Eventually became good enough to teach it to others.

And here’s what irritates me most: the models that look the most sophisticated are usually the worst.

50 tabs. Color-coded everything. Formulas that stretch across the entire alphabet.

Beautiful. Impressive. Completely f*cked.

You try to trace precedents? Good luck. You try to understand the logic? It’s buried under layers of links nobody documented.

Hours later, you finally figure out what it’s doing. And then you find the broken pieces.

• The circular reference someone “fixed” by hardcoding.
• The hidden tab that was supposed to update but doesn’t.
• The assumption that changed 6 months ago but lives on in cell P47.

Here’s what I learned after thousands of models: the best financial models are boring.

• One clear logic flow (no treasure hunts)
• Assumptions visible on page 1 (no mystery inputs)
• Can explain it on a whiteboard (if you can’t, rebuild it)
• Someone else can audit it in under 30 minutes (the real test)

Complexity isn’t sophistication. It’s technical debt waiting to blow up your forecast.

The CFOs I respect most? They build models a summer intern could follow.

Not because they lack skill. Because they understand the real risk isn’t being wrong. It’s being wrong and not knowing why.

AI proficiency is the new Excel proficiency

Twenty years ago, “proficient in Excel” was a nice-to-have on resumes.

Ten years ago, it was table stakes.

Now? Nobody even lists it. It’s assumed.

AI proficiency is following the same curve. Except faster.

Shopify’s CEO told his team that if you can’t prove you’re becoming proficient with AI, you won’t keep working here.

Everyone thought: “Well, that’s a tech company.”

Then every other tech CEO said the same thing within 48 hours.

Now I’m seeing it in non-tech companies. A women’s fashion brand requiring “baseline AI proficiency.” A regional healthcare group. Same expectation.

Here’s what most people miss: you don’t need to build AI agents or automate workflows to be “proficient.”

Start with brute force augmentation. Use it to get a little more accuracy. A little more creativity. A little more efficiency. Five minutes here, ten minutes there.

That’s crawling. And crawling is enough for most knowledge workers right now.

The CFOs who figure this out first won’t replace their teams with AI.

They’ll make their teams unreplaceable.

The spreadsheet doesn’t care what OS it lives on

Finance people treat Excel like a religion.

You know the type. Opinions about pivot table formatting. Keyboard shortcuts memorized like scripture. I was that person for 20 years.

So when I switched to Mac, it felt like going through a divorce. I’ve been working with Excel on Windows longer than I’ve been married. I’ve probably spent more time with Microsoft than with my wife and kids. I don’t want to do that math.

Here’s what actually happened.

Excel works. Different shortcuts. No Power Pivot. VBA is limited. That’s the real list of what you lose. And honestly, if VBA macros are still holding your workflow together in 2026, the problem isn’t your operating system.

The stuff I thought I’d miss from Microsoft? I found better alternatives. Email clients that don’t feel like they were designed in 2006.

A trackpad that actually works. And the worst offender: closing your laptop, throwing it in your bag, then pulling it out an hour later to find it’s been running the whole time, burning hot, battery at 12%. That alone was worth the switch.

I didn’t realize how much time I spent fighting my computer until I stopped fighting it.

I don’t want to sound like an Apple evangelist. I resisted this for years. But here’s what I keep coming back to.

In finance, we’re conservative by nature. We resist change. We’re used to being the last department to get system upgrades because we’re back office. Everyone else gets the new tools first. We get them when IT finally has time. So we hold onto what we know. We build our workflows around whatever we were handed and we don’t question it.

That’s the same instinct that kept me on Windows for 20 years. And it’s the same instinct that keeps finance teams from adopting anything new, whether it’s a laptop, a tool, or a way of working.

If a 20-year Windows convert can make this switch and not miss a day, we can embrace change in other places too.

But of course, with controls.

The spreadsheet doesn’t care what OS it lives on.

Related

• More on forecasts, models and board reporting
• When a financial model is too complex
• Excel dashboard vs Power BI for finance
• Claude for Excel