Decades after the database wars ended, after the tech press stopped covering every new Microsoft and Oracle database feature, databases are the cool new software again. Microsoft and Google each recently launched a new database app. Zapier built a database into its automation tool. Monday.com went from being a project management tool to a database to manage every bit of data in your company. Notion, Coda, and Airtable (the app that started the new database wars) are all core parts of modern workflows.
Every app wants to own and manage your data. And yet, it’s information, “order wrenched from disorder,” as James Gleick defined it in his homonymous book, that’s truly valuable.
Data is that your last three customers paid $100, $90, and $80 each. Information is surmising that your average sale price is trending downwards. Data is the 570GB of text used to train GPT. Information is the ChatGPT output that increasingly speeds up our work today.
There’s a reason the database wars are back, why every app and software giant is building database tools again. And it’s a very different reason from why IBM, Oracle, and Microsoft fought the database wars over the first decades of computing.
The First Database Wars
“You can't see the big picture because it's been sliced and diced and stored in so many different locations.” That’s Oracle founder Larry Ellison, fretting about data fragmentation, in 1999, right before web apps and Software as a Service took off. “It's impossible to know what's going on.”
The original database wars centered on “whether most computing should take place on big centralized servers or on smaller machines located near their users,” as the New York Times framed it in 1998.
IBM’s Db2 ran on mainframes, the room-sized computers that process financial transactions and flight bookings to this day. Oracle’s eponymous database ran on servers. Microsoft’s SQL server ran on desktops.
Any database could store data. But to get information, to “know what’s going on,” you needed all of the data in one place. Which is why, over time, the Oracle server-centric approach won out. Oracle’s databases powered some of the earliest web apps like Salesforce and SAP, Microsoft’s SQL Server morphed into a core part of the Azure cloud, and open source databases like MySQL quickly became the new, free standard.
The web app industry sprung up to put a friendly face on top of databases. Your average business software is a CRUD app, a database wrapped in forms to enter data and dashboards to draw insights from it. Companies quit caring about which database to buy, and started choosing software designed for each task and category of data.
Contact data in Salesforce. Payment data in Stripe. Issues in GitHub. Support emails about those issues in Zendesk. Each tool is powered by databases that are largely hidden from their users.
The database wars had ended. But the information wars had just begun.
Fragmentation in the cloud
It turned out, some web apps were far more sticky than others.
A tool built for a single task—sending invoices, say—was easy to drop if you weren’t sending all that many invoices. The CRM that stored all of your customer data, though, you’d keep using as long as you were in business.
The software that was your system of record became integral to your business. The more data you added, the more insights you could generate, and the less likely you were to quit using it.
“Companies ... will amass more data than ever in history - and for the first time be able to do something productive with it,” predicted IBM CEO Lou Gerstner in 1998. And doing productive things requires having as much data in one place as possible.
It’s still copying and pasting, only this time the computer’s doing it for you. Turning data into information requires automation platforms like Zapier to copy every bit of data into one central app, or CSV exports and manual spreadsheet workflows to patch together disparate pools of data. If the data’s even a bit different between apps, the whole workflow breaks.
Now imagine a single database that had all of that info in one place. If your project, customer, pricing, and template data is all in the same app, your tasks will be completed faster, and will be far less likely to break. With all that data in one place, it’s now possible to train AI on your company’s data and turn bits into information.
Which means the largest databases that are closest to their end application will win.
“The companies best positioned to take advantage of AI are ... those with the most data already on their service,” wrote Stratechery founder Ben Thompson.
And that’s why everyone’s building a database.
Databases on the edge
It’s easier to automate and draw artificial intelligence from data when it’s all together in one place. Every major app now wants to store as much of your data as possible.
Airtable was one of the first to reemphasize databases, launched in 2012 as a web-based rethinking of the desktop-centric databases like Access that Microsoft had sold in the 90’s. Where most other web apps promised to store one type of data—contacts, tasks, orders, and so on—Airtable was a flexible database to store anything.
More structured than a spreadsheet, to keep things orderly. More flexible than your average web app, to store any type of data you wanted.
That made it a perfect automation companion. A single Airtable database might have contact and project and pricing data all in one place, ready to merge into a report in a click. It could be any SaaS app you wanted; you could build a CRUD app without coding MySQL queries. It now promises to make AI more accessible, with all your data centralized.
And now, every other SaaS app wants that same data-backed power.
Zapier, alongside its automated workflows, has added a new Zapier Tables tool to store your data. It’s everything to run automations and GPT-powered AI queries in one place, without relying on spreadsheet lookups to patch together data.
Monday.com is looking beyond projects, with a new mondayDB built to power a “Work OS” with 100k+ items per board—far more data than a mere project would require.
Notion’s notes are also databases to capture everything—and feed into Notion AI to turn those data points into summaries and takeaways. Same for Coda, with documents that can reference database entries inline and AI chat that gets better the more data you put into Coda.
Microsoft and Amazon have both expressed similar sentiments in recent earnings calls, bullish that customers are likely to store more data in their clouds to take advantage of AI and machine learning. And both Microsoft and Google have recently launched new user-friendly database apps, built to match Airtable’s use-cases.
At some level, almost every business app has always been powered by a database. It’s just that now, the benefits of positioning as a database suddenly outweigh the benefits of being a single purpose app.
The new everything apps
You never got fired for buying IBM, went the business logic of the 90’s. Web apps for decades have flipped the equation, as startup software with nicer interfaces and unique, category-specific features won out over more corporate software.
In an AI powered, automated future, interfaces may suddenly matter less. You’ll never see a nicer UI if you only use the software through a chat interface, or if it generates PDFs for you automatically in the background.
What’s going to matter far more is your system of record and your organizing principle for data. An automation centric workflow might find Zapier tables or Airtable to be the best store of data. A notes centered workflow may fit Notion or Coda better. A project centric one might end up on Monday, a contact one on Salesforce. Or, equally likely, Google Tables or Microsoft Lists—both newer database apps—could, with enough marketing power behind them, become the new all-purpose database that everyone uses for everything.
Much like you chose an operating system in the past, businesses will increasingly pick one database tool to store all of their core data—not only a single category of data.
Then you’ll want to push as much data into that one system as possible. You’ll rely on AI inside that product, ignoring it for the most part in other tools, as the larger mass of data will make it so much more valuable. And you’ll choose the remaining software in your stack, your form builders and email services and other companion tools, based on how well they integrate with that system of record.
The computing model of the future might see that one application with your largest database as the center of your workflows, much like a server would have been in a previous workflow. This time, it’s a single database surrounded by apps that feed data in and fetch data out, rather than a single server with individual PCs pushing and pulling data.
Which is why databases are the new must-have feature. It’s likely that more business software will add or emphasize database features, now that it’s more valuable to be a database than just a CRM or project management app. And the tools that will be most valuable to your work, going forward, will be those that you easily move data around between the software you use.