Microsoft Excel Powers an AI-Resistant, Multitrillion-Dollar Empire

Before the Apple II, of course, paper spreadsheets had been in use for generations by bookkeepers, who, one can only assume, also had mixed feelings about them. The basic idea is even older. A clay tablet, found in Iraq and dated by archaeologists to around 1800 B.C., shows 15 rows and 4 columns of numbers, calculations that seemed to have been etched by some anonymous Babylonian forerunner to today’s moderately disgruntled marketing consultant.

While the cuneiform digits are unfamiliar, the format is basically identical to what Bricklin came up with at Harvard Business School in 1978. He got the idea while watching a professor fill out a table of financial projections on a chalkboard and struggling to keep up in his notebook. He began to daydream about a software tool that would combine aspects of an electronic calculator and a word processor, allowing him to edit a table without having to erase and redo the whole thing. The idea was a “magic piece of paper,” as he put it in a class project in 1978, with rows and columns you could update by clicking with a mouse.

At this time the Apple II didn’t have a mouse, but it did ship with two twist dials for use with Pong and other primitive games. Bricklin incorporated the dials into the first version of his spreadsheet program, then switched to using the more accountant-friendly arrow keys instead. When he and VisiCalc co-creator Bob Frankston began selling the software in 1979—originally priced at $99 and sold as a 5¼-inch floppy that came inside a slim leatherette portfolio—it transformed the Apple II almost instantly from a niche tool for gamers and teachers into the first mass-market PC.

Before VisiCalc, anyone who wanted to use a computer for financial modeling needed to know a programming language and have access to a mainframe. After VisiCalc, all you needed to know were a few simple functions, such as SUM (which we’ve covered), AVERAGE (what you’d imagine) and VLOOKUP (a search function for large sheets). “It was like this interactive toy, where you can kind of program it without knowing you’re programming,” says Ray Ozzie, who worked at Bricklin and Frankston’s company, Software Arts, and, later, at Microsoft. “It was amazingly seductive.”

Like many early pioneers, Bricklin proved to be too far ahead of his time. In the 1970s software patents were uncommon—the conventional wisdom was that software was math, and therefore unpatentable—so he didn’t bother trying to protect VisiCalc from copycats. “Financially it would have been great if we’d have been able to patent it,” Bricklin says. “And there would be a Bricklin Building at MIT, instead of a Gates Building.”

The comment hints at a second reason—beyond the simple fact that grids of numbers are just kind of dull—for the ambivalence many feel about Excel. It’s not a brilliant invention but a brilliant knockoff, whose popularity will forever be tied to the imperial approach to business favored by Microsoft and its co-founder Bill Gates. After Microsoft’s first attempt to copy VisiCalc fizzled, Gates set about trying to clone the most successful of the VisiCalc copycats. This was Lotus 1-2-3, which had become the early winner on IBM and IBM-compatible PCs by expanding the capacity of each spreadsheet from 255 rows to 2,048 and by adding a few new features, including charts.

Excel was born at a Microsoft offsite in 1983 under the code name Odyssey. The original plan was to create a better version of Lotus for IBM-style PCs. “In many ways it was a troubled project,” says Doug Klunder, the lead developer. From Klunder’s point of view the trouble started in 1984, when Steve Jobs unveiled Apple Inc.’s Macintosh, the first inexpensive computer with a mouse and a graphical user interface. The buzz—along with the fact that Microsoft was itself preparing to release its Windows operating system—led Gates to switch Excel’s target market from PCs running the old DOS operating system to Macs. This delayed Excel’s release significantly, forcing Klunder to spend six months redoing his work.

He was so frustrated, both by the delay and because he’d been excluded from the meeting in which the decision was made, that he quit, telling colleagues his plan was to leave the software industry in favor of migrant farmwork. “I look back on it, and it’s clear I was just totally burnt out,” he says.

Project Odyssey got back on track after Klunder ran out of cash somewhere in the Ojai Valley two months after his departure. “I picked lettuce a couple of times and did some other odd jobs,” he recalls. “Then my backpack was stolen. I’m like, ‘I need money. Let me call Microsoft.’”

While Klunder and a team of 10 programmers—plus a handful of employees assigned to test the software and write the manual—raced to finish, Microsoft’s marketing operation kicked into gear. In May 1985, Gates and Jobs appeared together at a launch event at Tavern on the Green in New York City’s Central Park. By this point the two men were rivals, and though both were excited about the release, it was for different reasons. Jobs saw Excel as a stopgap until Lotus finished its own Mac spreadsheet. Gates viewed it as a key application for Windows.

To help push that point, Microsoft executives planted a question about the future of Excel. During a Q&A portion of the event, a reporter asked if it would be available exclusively on the Mac, or if Microsoft planned to bring it to PCs. Gates answered diplomatically, telling the crowd his plan was to make Excel available to all platforms “in time,” according to Jeff Raikes, then a Microsoft marketing executive who attended the event. “He’s trying to be gentle with Steve Jobs,” Raikes recalls.

Jobs, seeming to sense a threat, parried back. “In time,” he said, “we’ll all be dead.” Gates allowed the audience to laugh at that one and then deadpanned, “Not IBM.” The message was clear: As far as Gates was concerned, the Mac was the stopgap. “Jobs was speechless,” Raikes recalls.

Microsoft had said it would deliver Excel by September, forcing Klunder and his team to pull an all-nighter on Sept. 29 to finish debugging their work. The following day they got a few 3½-inch floppy disks made and drove them and the roughly 550 pages of manuals and reference guides over to a local Egghead Software (a bygone retail chain) in Bellevue, Washington. After that humble start, however, things moved fast. By the spring of 1986, after Klunder left his job again, this time to walk the Pacific Crest Trail, other hikers told him they were using Excel to plot their daily mileage and plan their meals.

Another highlight of early Excel was a feature that allowed users to preview on the screen what their spreadsheet would look like laid out on paper. “That was just thrown in because we thought it was cool,” Klunder recalls. Blumenthal initially tried to cancel the feature, but he relented under pressure from Klunder’s team. Print preview would turn out to be a cornerstone function of pretty much every piece of desktop publishing software created since.

Even so, the defining feature of Excel, at least in the eyes of its users, was how it looked, not how it worked. “Speed of recalc was something people always talked about” but “not the thing that really grabs you,” says Steve Ballmer, who was Microsoft’s head of sales in the mid-’80s and later ran the company. “The real thing was we had a graphical user interface, so almost everything looked better.” Ballmer says that spreadsheets are harder to conceptualize than most business applications and that he urged reps to “demo, demo, demo, demo, demo,” as he puts it. “You’re showing another senior person, ‘See, I use this thing. Isn’t this amazing?’”

Ballmer retired from Microsoft in 2014 and now spends much of his time running the Los Angeles Clippers, but as if to make the point about the power of a spreadsheet demo, he pulls up one of his own Excel files—a detailed accounting, with hundreds of rows, of how he spends each hour of each workday. The spreadsheet includes the number of hours in a given month spent meeting with players, working on his nonprofits, checking emails, exercising, playing golf, getting massages and on and on. Bathroom breaks are included in a row called “free time.” (The sheet seemingly makes no mention of any time spent navigating the National Basketball Association salary cap.) For each category, Ballmer has an annual budget, based on previous years’ spreadsheets. All the times add up to 2,080, the theoretical number of work hours each year. He started this as part of an attempt to spend less time traveling for work. “Most people think it’s pretty weird,” he says with a hint of pride.

It wasn’t just memorable demos or Ballmer’s energetic pitches that sold Excel. In 1988, Raikes, who would eventually become president of the company’s business division, came up with the idea of using the software to push buyers to switch to other Microsoft products. This became known as “the bundle,” and it involved packaging Excel with Microsoft Word as well as a presentation application for the Mac called PowerPoint, which Microsoft bought in 1987 for $14 million. “I said, ‘Hey, you know these companies put their list price at $495 per app,’” Raikes recalls. “‘Why don’t we put three apps into a package and sell it for the price of two?’” This became known as Microsoft Office, sold in a literal package in computer stores and later as a per-user license to the IT departments of large companies and to computer manufacturers that included it with the new machines sold to consumers.

Microsoft used bundling as a competitive strategy, which rivals saw as predatory and unfair. Kapor, the Lotus co-founder, says he didn’t mind the other company’s copying. But “what was painful and difficult were Microsoft’s business practices, which were over the line in many ways,” he says. “I’m hoping that Bill, before he exits the scene, will take more responsibility for that.”

Eventually, another example of Microsoft’s bundling would figure in the US government’s antitrust case against the company in the late 1990s, but by that point Office had effectively taken most of the market for business software. “Lotus tried to catch up,” says Salas, the Brandeis computer science professor. “The Microsoft Office behemoth just locked everybody in.” Salas is famous among spreadsheet aficionados as the inventor of what’s known as the pivot table, a software tool that summarizes complicated datasets. For instance, a pivot table might take a detailed list of transactions for a retail chain, including information about store locations, products sold and prices paid, then generate a report that shows a breakdown of the most popular products by region or that makes clear which stores are the most profitable.

Pivot tables are now seen as a key part of Excel. Oz du Soleil, an influencer who runs the popular YouTube channel Excel on Fire, describes learning about the feature (while working as an analyst at an online education company) as “the crack in the wall” that changed the trajectory of his career. He remembers a feeling of empowerment, a sense “that I can get the shit done.” Pivot tables, he says, are “the turning point for a lot of people.”

Microsoft didn’t invent pivot tables, of course. Salas came up with the idea working at Lotus. But it only took off once Microsoft engineers copied it and stuck pivot tables into a 1994 version of Excel. “We thought it was flattering and amusing,” Salas says of watching his invention get knocked off. “That was the arrogance of youth.”

“I was kind of shocked at first,” says Blumenthal, the former Microsoft executive. “Like, ‘Oh, my God, they just ripped off this idea.’ Which, by the way, I did as well, when we did the first version of Excel.” He says Sheets now includes “99% of the functionality that 99% of Excel users use.”

But a spreadsheet that anyone can edit isn’t necessarily what you want when you’re working with data that’s complex, sensitive or complexly sensitive. Those use cases include a lot of the data that workers put into spreadsheets. There was also the lock-in factor from Microsoft’s dominance of PC software. “We thought Google Sheets was going to kill Excel, but it didn’t,” says Ozzie, the onetime VisiCalc (and Lotus) employee who was Microsoft’s chief software architect from 2006 to 2010.

By the late 2000s most companies were already paying huge sums each year for Microsoft’s bundle, meaning they’d also have to toss Outlook, PowerPoint and Word if they wanted to save money by switching to Sheets. This sense of inertia had a psychological component too. We’d all just gotten used to Microsoft products, even if we didn’t like them. O’Mara, the Silicon Valley historian, attributes Excel’s staying power partly to “this managerial hangover where the people making purchasing decisions are living in a world that Microsoft made.”

“I’m a Gen Xer,” O’Mara says. “We’re all comically attached to the Microsoft suite. All my students use Google Docs. All my younger editors and collaborators use Google. I’m still using Word.”

The second serious challenge to Excel’s dominance has come more recently, with the rise of ChatGPT and other AI chatbots. These bots promise to instantly make sense of enormous amounts of information, suggest new strategies, spot trends and even spit out fully formed business plans, without all the hours of staring at and manipulating a grid. (Their track record of actually fulfilling these promises is, for now at least, spotty.) There are dozens of startups vying to replicate parts of Excel or replace the entire thing. This summer, tech news site the Information reported that OpenAI is working on an office suite that will compete directly with Microsoft’s bundle.

None of these efforts has panned out quite yet. “We are meeting with a bunch of AI spreadsheet companies,” says Alex Immerman, a partner at Andreessen Horowitz, the venture capital firm. “They’re all built on top of Excel.” By this he means that most Excel killers are, for now, either copycats or add-ons that help you make an Excel-compatible spreadsheet.

This happens to be Microsoft’s strategy too. Its main chatbot, Copilot, runs inside of Excel and promises to make it easy to add more sophisticated calculations to existing spreadsheets. The idea, says Rajesh Jha, the Microsoft executive who oversees Excel (as well as Windows and the other Office apps), is to bring the level of expertise commonly seen at the Excel World Championship to the rest of us.

There are reasons to be skeptical of claims that AI is killing off the spreadsheet. The chatbots’ large language models typically struggle with simple calculations, and they’re notoriously bad at showing their work. “In my day to day, I spend more time fighting off people’s vibe-coded projects than actual useful AI integrations,” says Nick Frescas, a data analytics manager who serves as a moderator for the Excel forum on Reddit under the handle Frescani. “There’s something about processing data yourself that helps you internalize the information.”

For now, Excel is doing just fine. In the most recent quarter, commercial revenue for Microsoft 365 increased 17% from the same period a year earlier; revenue from consumers for the bundle of software and services that includes Excel was up 28%. Even Immerman, who as an investor says he sees “tremendous opportunity” in the chance to invest in Excel competitors, remains enamored of Microsoft’s 40-year-old spreadsheet app. He first used Excel to do simple calculations as a grade-schooler and eventually moved on to bar charts. In college he found his way to iterative calculations, conditional equations, macros and plug-ins. “You can go deeper and deeper,” he says. “Even though I spend time in Excel every day, I don’t harness the full power.”

Before his career in venture capital, Immerman worked as an analyst at the investment banking firm Allen & Co. He says he was embarrassed back then by the number of hours a day he spent staring at Excel. He no longer feels that way. “It’s open all day, every day for me, alongside Slack and email,” he says. “But those are communication mechanisms. Excel is where actual work happens.”

This distinction between communication and work may hold the key to understanding why in some quarters, Excel isn’t just tolerated but, weirdly, loved. “Excel is often used as the symbol of the soul-crushing office atmosphere,” says Blumenthal. But, he says, it can also be “the most inspiring and independent thing people are doing within their soul-crushing environment. It can actually be a source of creativity and independence.” It is, in other words, both the definition of what makes certain kinds of work unbearable and a big part of what makes them fulfilling.