Microsoft has become the undisputed leader in AI, thanks to its investment in ChatGPT and the chatbot’s integration into Bing. The company has invested $11 billion in OpenAI, the developer of ChatGPT, and that’s in addition to whatever it's spending internally to develop AI, which is considerable.
So far, though, Microsoft has little revenue to show for it. If you believe the press, the company's chatbot so far has mainly been used for the equivalent of tech parlor tricks, stunts like creating a cover of Beyoncé's "Cuff It" as if it were sung by Rihanna, writing a Seinfeld script in which Jerry and George plot to kill Jeffrey Epstein in prison, and passing a law school exam.
There’s no money in that, of course, and Microsoft knows it. But it's happy to have the chatbot perform these kinds of stunts because they bring plenty of publicity. And ultimately, Microsoft plans to turn that attention into cash.
Is it even possible, or are the billions in hoped-for revenue little more than a pipe dream?
For once, I believe the hype may be justified; AI is likely to pay out for Microsoft — at least in the long run. Ultimately, there will be plenty of ways to cash in, but at the moment the big three are: stealing search revenue from Google via an AI-powered Bing; providing the infrastructure and tools for businesses to build their own chatbots and AIs; and incorporating AI-enhanced productivity tools throughout its product lineup.
Here’s how Microsoft will do it.
Attacking Google’s search dominance
We’ll start with the lowest-hanging fruit, using AI-powered Bing to eat into Google’s search dominance. Google built itself on search, and the company’s $162 billion in search income a year represents the lion’s share of its revenue. Google retains its stranglehold on the market; it had 85% of the worldwide market share of search in the first quarter of 2023, compared to only 9% for Bing.
Thanks to the integration of Microsoft’s chatbot into Bing, that may change in a big way. Microsoft is betting its chatbot will upend the way people search the internet. When you search via a traditional search engine such as Google, you’re not searching for information as much as you’re searching for websites you hope contain the information you want. It’s up to you to click each link, scroll each site, hope you find what you need, and then collate, organize, and summarize what you’ve found.
Bing’s chatbot attempts to turn that on its head. Search for information, and it goes out, finds it, and writes it up into a well-organized paragraph, complete with citations. No need for you to visit any web sites, scroll and read, or do collating or organizing. It can be a spectacular time-saver and offer more finely tuned results than traditional search.
That’s the theory, anyway. In practice, it can be hit-or-miss. Microsoft is betting that over time it will become far more accurate.
Even in its current state, though, it’s powerful enough to get Microsoft a bigger chunk of the search market. Thanks to Bing’s AI chatbot, Samsung is considering replacing Google with Bing as the search engine on its Android phones. That’s a very big deal because the company sold 259 million smartphones in 2022 alone.
That may be just the beginning — other manufacturers including Apple could follow suit. The New York Times notes that Google was thrown into a panic by Samsung possibly switching to Bing and has launched a desperate all-hands-on-deck attempt to build enough new capabilities into its search engine to try and retain the contract.
Cashing in on AI-as-a-Service
Building and running chatbots and other AI apps requires supercomputer-like processing power, storing vast amounts of data, training AI on the data, troubleshooting and iterating the results, and then fine-tuning it and creating an easy-to-use, flexible interface that integrates with a business’ existing infrastructure. That requires capabilities and finances far beyond all but a small handful of companies.
That’s where Microsoft and what I’ll call AI-as-a-Service comes in. Microsoft sees tremendous potential in selling AI capabilities to companies the same way it now sells cloud services. The idea is simple: provide the processing power, storage, framework and other infrastructure a company needs via an AI-focused version of Microsoft’s Azure cloud services. Layer on top the tools and services needed to build AI apps and integrate them into a business. And charge a company based on how much AI services they use.
Microsoft calls it the Azure AI Platform, and it’s already up and running. It allows companies to use their own internal data for AI, and even build their own machine-learning models. A number of companies have already built AI capabilities, using it for a wide range of purposes, including the NBA, CarMax, H&R Block, and others.
Microsoft turned itself from a sleepy, almost has-been company in the early 2010s to the second most valuable company in the world by focusing on the cloud. It’s hoping AI-as-a-Service can juice that growth and possibly make it the most valuable company in the world. It’s too early to tell whether it will pay out, and whether AI-as-a-Service is a real thing or marketing hype.
Either way, the company has bet the farm on it.
A Copilot for productivity
Finally there’s Microsoft 365 Copilot, designed to improve productivity in Office apps, Teams, and potentially other Microsoft products. In Microsoft’s words, Copilot will let you “focus on the most important work and less on the busy work.”
What does that vague-sounding, hypey description mean in practice? Copilot, Microsoft claims, will be able to do things like draft a two-page project proposal based on information found in a Word document and an Excel spreadsheet. The company claims that in Excel it will be able to analyze your data and break down corporate sales based on type and sales channel — and create a chart to visualize all the data. In PowerPoint, according to Microsoft, it will be able to build a five-slide presentation based solely on Word document — and find and include relevant stock photos.
In Teams, it could be even more powerful, if Microsoft is to be believed. The company claims that by listening in on meetings, Copilot can perform tasks such as building a table of pros and cons about a topic being discussed and add what else should be considered before making a decision. It will also, Microsoft says, list all the decisions made and suggest follow-up actions.
Microsoft 365 — what used to be called Microsoft Office — is one of the company's cash cows. In good economic times and bad, businesses and people continue to use it. Microsoft is battling Google over productivity apps, with one recent estimate showing Microsoft 365 with a slight lead over Google Workplace, with a 47.5% market share to Google Workspace’s 44.56%. If Copilot can deliver on even a sliver of what the company promises, that could move the needle significantly towards Microsoft 365.
As for Teams, its revenue has grown from $0.01 billion revenue in 2017, the year of its launch, to $6.8 billion by 2020. The kind of features Microsoft promises could supercharge its growth and make Slack a perennial also-ran.
Keep in mind, though, that Google is working on building AI into Google Workspace, and AI is getting incorporated into Slack. So Microsoft’s AI work may not yield additional revenue. Still, even to just maintain market share, Microsoft needs Copilot.
The upshot
Microsoft hopes that building AI into all its products and launching new AI products will draw new customers into its ecosystem — where they’ll buy even more Microsoft products. The company hopes to forge links via AI between Bing search, a company’s AI infrastructure, and its productivity tools. It envisions enterprises living solely inside its ecosystem.
Of course, Google and other companies have similar plans. But for now Microsoft has a solid head start. It’s been moving at warp speed to try and make sure it stays that way. If it pays off — and that’s still a big if — the company will be well-positioned to become the dominant tech company for years to come.