If there’s one piece of software that dominates the business world across industries, it’s Microsoft Excel. Can AI finally dethrone the mighty spreadsheet?
Memorable spreadsheet moments
Everyone has memorable spreadsheet moments. I have a few. For example, my then-architecture firm was involved in more than a dozen housing developments abroad. I developed an Excel workbook that took the required number of households as input and automatically generated a breakdown of buildings and their apartment types for AutoCAD. This was urban planning and architectural design done with a spreadsheet.
I also developed business software using Excel for project portfolio management. The prototype was later scaled into a commercial SaaS that is now used globally.
Another memorable moment was when a property owner told me their Excel file grew so large that it ran out of rows and columns. That must have been before 2007, when the maximum number of columns on a sheet was still just 256 and the maximum number of rows was 65,536. The current limits are 1,048,576 rows and 16,384 columns, which I hope no one will exceed.
Why Excel rules in AEC
Everyone in AEC knows how to use Excel. You don’t need a coding degree to build a reasonably smart collection of data and rules. It enables quick, messy “what-if” scenarios that rigid ERP systems don’t allow.
Over the years, companies have created numerous Excel spreadsheets and templates that contain essential business knowledge for bidding, quantity takeoffs, resource planning, scheduling, and more. When we developed the SaaS application, we saw that even large corporations used Excel to manage their IT and development project portfolios and make strategic decisions based on those files.
A technical reason for Excel’s dominance is that practically every business and design software reads and writes Excel or at least Comma-Separated Values (CSV) files. The tool and format have become as important as PDF files in project collaboration. Neither of them is a recent innovation. Excel was first released in 1985, and PDFs came about in 1993.
The case against Excel
The biggest argument against Excel in AEC is that it creates static snapshots of data. If something is in a spreadsheet file, it is probably already outdated in some way.
However, there are ways to incorporate live data in Excel. It has powerful but often underused capabilities for live data connections. You can build a direct connection (pipe) to SQL databases, exported model data, or web APIs using features like Power Query and data connectors. Once the pipe is built, the user just clicks “Refresh” to update the data.
Even if Excel were a window into live data, it’s a separate physical entity. By embedding data in a .xlsx file, we are essentially placing our project’s most valuable information in a document. True digital transformation requires data to be liquid, flowing seamlessly between the design model, the site sensors, and the financial ledger. You can’t build a real-time organization on a foundation of frozen files.
The third major problem is that a complicated Excel file, created by some savvy individual, can easily break or at least not function as expected if you don’t follow the creator’s logic. In addition, files with hundreds or thousands of formulas are sure to contain errors. In one study, professional auditors examined operational spreadsheets in large corporations and found that 94% had errors.
The replacement
Artificial Intelligence is the most significant variable in this transition. Paradoxically, it might be the only thing capable of making Excel functional for the next decade, while simultaneously rendering it obsolete.
Large Language Models have effectively democratized coding power. A site engineer who doesn’t know a line of code can now use AI to write complex macros or Power Query scripts. AI has lowered the “formula barrier,” making the spreadsheet more accessible than ever.
However, the real “Excel killers” are AI-native vertical platforms. While a spreadsheet can tell you that you are over budget, an AI-driven model can run 5,000 permutations of a construction sequence to find the one path that saves the margin.
Headless Data
We are used to staring at rows and columns, but are they the ideal way to present data? Do they answer the questions we might have without extra work?
We are entering the era of Headless Data. In 2026, a Project Director doesn’t need to open a 50-tab workbook. They can simply ask their firm’s fine-tuned AI agent: “Which sub-contractors on the North Tower are trending toward a delay based on the last three weeks of site photos and weather data?”
The AI queries the database, runs the cross-tabulation, and provides the answer in seconds. The “spreadsheet work” happened in the background, but the human never saw a grid.
From cell chaos to object-oriented logic
In Excel, cell A1 has no inherent relationship to cell B1 unless a human manually types a formula. In Object-Oriented platforms, like Airtable, Smartsheet, or bespoke internal apps built via “vibe coding,” a “beam” is an object. That object carries its own weight, cost, vendor, and installation status.
When we move to object-oriented decision-making, we gain traceability. We move away from “Shadow IT” (the mission-critical spreadsheets living on Bob’s desktop) and toward governed, audited data. In the AEC industry, where litigation is always a looming shadow, the ability to see a decision’s history is more valuable than the decision itself.
The role of Excel in the future
Excel has served us well for over 40 years and won’t be abandoned anytime soon. It will remain, and possibly perform quite well, as a sandbox and “digital napkin” for trying out various scenarios. It supports quick-and-dirty, “let’s just see if this works” logic.
Superseding Excel won’t be a sudden event; it will be an evolution of roles. Excel will remain our favorite calculator, data lakes will become our data sources, and AI will be our analyst.
By moving our decision-making into live, AI-enhanced environments, we are finally building on a foundation that doesn’t break when you delete a row.
