The impact of AI in Excel on work, jobs, and training.
For two years, AI’s biggest impact was on coding. Developers were the first workforce to feel it. But developers represent a relatively small population globally. In the last few weeks, AI moved into the tool that touches nearly every office worker on the planet: the spreadsheet.
Claude in Excel launched as a beta capability in early 2026. It allows users to describe their requirements in plain language and have the AI build formulas, structure data, and generate models directly within Excel. No formula syntax required. No memorizing VLOOKUP arguments. Just describe the output you want.
This is not a theoretical shift. I used a PDF article on startup funding stages, described to Claude what I needed, and built a complete spreadsheet with interconnected formulas in minutes. That same task, done manually, would have taken a day or more.
I will never enter another formula into Excel manually. I will always describe what I need and let AI build it. I will never create another PowerPoint. I will use AI to create more interactive and engaging work. That line has been crossed.
To understand why this matters, you have to understand how deeply spreadsheets are embedded in the way the world works.
Microsoft has historically described the Office suite as reaching more than 1 billion users. A December 2025 Microsoft 365 blog post reports that more than 430 million people use Microsoft 365 apps. Excel sits at the center of that suite. While a precise Excel-only user count is not publicly reported by Microsoft, the tool is used by hundreds of millions of people across virtually every industry on earth.
The people spending all this time in spreadsheets were never properly trained to use them:
Source: Acuity Training survey of 1,000 office workers, Jan-Feb 2022
This means hundreds of millions of people are spending a third of their working lives using a tool they largely self-taught, building models that drive real business decisions.
The consequence of untrained people building mission-critical spreadsheets by hand is predictable: errors are endemic.
Published academic research on spreadsheet quality has repeatedly found that error rates are uncomfortably high when business spreadsheets are audited. A 2024 peer-reviewed overview of this research literature found that spreadsheets used in business decision-making frequently contain errors, with some field audits reporting rates above 90%. Mainstream industry analysis indicates that approximately half of the spreadsheet models used in large businesses may contain material defects.
| Company | What Happened | Financial Impact |
|---|---|---|
| JPMorgan | Copy-paste errors across spreadsheets contributed to flawed VaR risk models in the “London Whale” incident | $6+ billion in trading losses |
| Fidelity | Accountant omitted a minus sign when transcribing a $1.3B net capital loss, turning it into a gain | $2.6B dividend miscalculation |
| Citigroup | A manual spreadsheet input error caused an interest payment to be sent as a full principal repayment | ~$900M accidentally wired |
| Lazard / SolarCity | Computational error in a spreadsheet used for DCF analysis during Tesla acquisition advisory | $400M valuation error |
| TransAlta | Erroneous spreadsheet entry caused overbidding on power transmission hedging contracts | $24M loss |
The spreadsheet touches every layer of the workforce, from Wall Street to Main Street. Here is how AI in Excel reshapes each tier, ranked by the financial stakes of the spreadsheet work being disrupted.
U.S. workforce: ~368,500 financial and investment analysts (BLS, 2024)
Investment banking analysts routinely spend 6 to 8 hours per day inside spreadsheets, building DCF models, LBO analyses, merger models, and comparable company valuations. The Association for Financial Professionals reports that 96% of FP&A professionals use spreadsheets for planning at least weekly.
A first-year analyst building a three-statement financial model from scratch might spend 20 to 40 hours on it. AI can now draft the structural framework of that model in minutes. The analyst’s role shifts from formula construction to assumption setting, model validation, and strategic interpretation.
What changes: The focus shifts from building the model to understanding what the model should say and whether to trust it.
These are the people behind every budget cycle, quarterly forecast, variance report, and board presentation. They consolidate data across departments, build rolling forecasts, and run scenario analyses. Their spreadsheets are the backbone of corporate decision-making.
When an FP&A team can describe a three-scenario revenue forecast and obtain a working model in seconds rather than spending a week assembling it, the time saved is measured in thousands of hours per company per year.
What changes: Less time on consolidation plumbing. More time on what the numbers mean and what the business should do about them.
U.S. workforce: ~1,579,800 accountants and auditors (BLS, 2024), ~653,400 actively licensed CPAs (NASBA, 2025)
Accounting professionals spend an enormous amount of time on reconciliations, trial balances, consolidations, audit workpapers, and tax preparation. The disruption here goes beyond speed. When AI handles formula construction and data validation, it eliminates the class of manual transcription errors that have cost companies billions.
What changes: The accountant’s role shifts from “build and check the spreadsheet” to “verify the AI’s output and provide professional judgment.”
Inventory models, procurement trackers, project timelines, resource allocation sheets, logistics planning. Every manufacturer, construction firm, and logistics company runs on spreadsheets at the operational layer. These teams manage Gantt charts, vendor costs, and capacity planning in Excel.
What changes: The person who spent hours building a project tracking workbook can now describe what they need and iterate on the output. Operational reporting becomes faster and more consistent.
Pipeline tracking, commission calculations, campaign ROI analysis, customer segmentation, and territory planning. These teams build the reports that drive pricing decisions and revenue forecasts. The person building a commission waterfall model or marketing attribution spreadsheet can now specify their requirements in plain language.
What changes: Reporting cycles compress. More time selling, less time formatting.
Payroll tracking, headcount planning, benefits analysis, employee scheduling, compliance reporting. These spreadsheets are often the messiest in any organization because they have been inherited and patched over the years. The person maintaining a 47-tab employee tracking workbook with broken formulas everywhere is about to have their working life transformed.
What changes: Legacy spreadsheet debt gets resolved. New models get built clean from the start.
This is the bookkeeper in a small shop in Southwest Florida. Industry data indicate that approximately 70% of small businesses lack a dedicated accountant. Many small business owners report feeling unskilled in bookkeeping. These are often part-time roles in which individuals manually enter receipts, reconcile bank statements, and build basic P&L reports in Excel using formulas copied from YouTube tutorials.
This tier gets the most dramatic quality-of-life improvement. The bookkeeper who struggled to build a basic cash flow statement can now describe what they need and get a professional-grade output. The gap between their skill level and the output quality collapses overnight.
What changes: The bookkeeper goes from data entry to data insight. They can advise the business owner rather than merely recording transactions.
The disruption is not about replacing people. It is about what happens when the time allocation shifts.
If an investment banker reallocates 30 of the 50-60 weekly hours currently spent in Excel, those hours can be devoted to deal analysis, client relationships, and strategic thinking. If a bookkeeper who used to spend 15 hours a month maintaining a messy spreadsheet can now get the same output in 2 hours, she can analyze the numbers and advise the business owner.
The Zapier survey found that 76% of workers spend 1 to 3 hours a day just moving data from one place to another, and 83% spend 1 to 3 hours a day fixing errors. That is the category of work that AI in Excel directly eliminates.
The shift is not from spreadsheets to something else. It is from building spreadsheets by hand to describing what the spreadsheet should do. The skill moves from formula syntax to clear thinking.
The corporate training market reached $371 billion globally in 2024, with Excel training representing one of the largest categories of technical skills instruction. That entire training model is about to undergo a significant change.
Instead of teaching people formula syntax, keyboard shortcuts, and VLOOKUP arguments, training becomes about three things:
Microsoft’s documentation for Copilot in Excel includes warnings about using AI outputs for tasks that require high accuracy and reproducibility. This reinforces rather than undermines the training opportunity: the human’s role is now verification, judgment, and decision-making.
For two years, coding got the attention. AI coding assistants have transformed how software is built, affecting roughly 30 million developers worldwide. The spreadsheet population is measured in hundreds of millions. This wave is an order of magnitude larger in the number of people it touches.
The capability is weeks old. Most organizations are not yet aware of it. Most training programs have not adapted. Most workers are unaware of this. That gap between what is now possible and what people currently know is both the challenge and the opportunity.
The people and organizations that move first will compress weeks of spreadsheet work into hours. Those who wait will continue doing it the old way, one formula at a time, and wonder why their competitors seem to have more time for the work that actually matters.