What’s Curious about Microsoft and OpenAI
In a Thursday night bonus issue of The Dispatch, I dig into some non-obvious reasons why a ChatGPT integration with 365 and Bing would be a big win for Microsoft.
It’s more than a flurry: we’re seeing a monsoon of articles about OpenAI’s ChatGPT, Microsoft’s proposed $10B investment in OpenAI, and how Microsoft would recoup that investment by getting 75% of OpenAI’s profits—of which there are currently none. (Nina Schick has a nice summary.)
Even without that 75% of profits provision, this is a great deal for Microsoft because Microsoft plans to integrate ChatGPT into its 365 Suite (a.k.a. Office) and into Bing, its search engine.
For 365, the benefits of ChatGPT are obvious: a writing assistant to help you generate text with just a few prompts, help you create beautiful slides for a presentation, or write the formula for a spreadsheet.
Other benefits aren’t as obvious, particularly for Bing.
Right now, Google dominates Search and therefore also dominates Paid Search, which is how it makes the vast majority of its vast amounts of annual revenue. This chart from Statista shows that global revenue going into Search Marketing this year will be $300B; this chart from IBIS shows that U.S. revenue going into Search Marketing this year will be $122B.
Microsoft owns and operates Bing, the distant #2 search engine after Google. If ChatGPT can bolster Bing’s allure and coax just 3.3% of Global Search or 8% of U.S. over from Google, then Microsoft will make back its $10B outside of any profit sharing.
However, this would be at best a temporary win.
A ChatGPT-powered version of Bing would be more effective for users but less profitable for Microsoft.
In an email exchange, my friend Kevin Lee described it this way:
It’s going to be really interesting because, short term, the SERP being enhanced by ChatGPT would allow for a jump in relevance and usefulness, which would give Bing a new advantage.
However, long term, ChatGPT gives you the answer (doesn’t need a SERP at all) so that should be sold as a subscription (Prime or Office 365-style).
The entire SEM world relies on a SERP. So, things are gonna get interesting.
Unpacking the industry patois a bit, SERP means “Search Engine Results Page” and “SEM” means Search Engine Marketing. What Kevin is saying is that Google makes its money on the long list of sponsored results when you type a sentence like “I want to buy a new Toyota Camry” into the Google Search bar and hit enter. The sponsored results are the ones that say “Ad” next to them. Bing makes its money the same way.
However, if Bing pivots to a ChatGPT-powered approach to search, then instead of getting a lot of paid-for answers you have to scroll through you’ll only get one answer—the right one. That’s not as profitable.
Instead of free search powered by ads, a ChatGPT-powered search might evolve into a paid service, with one implication being that it would deepen the economic divide between Haves and Have-Nots. Have-Nots would rely on free, ad-supported search, while Haves would get ad-free search for a monthly or annual subscription.
The Haves would probably enjoy more privacy than the Have-Nots.
Other Benefits to Microsoft
If Microsoft integrates ChatGPT into Office365, that might create a compelling reason for people to subscribe to those services over free alternatives like the Google Suite that has word processing, spreadsheets, and slide ware. (That Haves vs. Have-Nots thing again.)
Plus, with the 2021 acquisition of Xandr from AT&T, Microsoft re-entered the digital advertising business after selling off its old portfolio of ad-tech to AOL back in 2015. The more people use Microsoft’s digital services (which ChatGPT integration will accelerate), the more data Microsoft will have about the users, which then enables Microsoft to put highly-targeted (programmatic or AI-driven) ads in front of those users, including on Xbox.
Finally—and this is going to get a bit wonky, so please forgive me—all that data would be in a logged-in, first-party ecosystem, so there’s at least a reasonable argument for Microsoft to make that using this data to harness advertising does not violate GDPR (the EU’s General Data Privacy Regulation), CPRA (the California Privacy Rights Act), or any of the other new legislation seeking to contain how much information big tech companies compile about people.
English Translation: Facebook lost a lot of its power and market value after Apple put that “do you want this app to track everything you do on your iPhone forever?” option into your iPhone’s Operating System (iOS). No sensible person has ever clicked “yes.”
If Microsoft can use ChatGPT to get a lot more data about a lot more people across all of its apps and services, that’s probably not something Apple can stop… and advertisers will pay for it.
For all that possible benefit, $10B is a steal.
Thanks for reading. See you Sunday.
The data being first-party and part of a closed ecosystem smacks of "the browser is an inseparable part of the operating system" doesn't it? I guess one of the open questions will be whether or not regulators feel like picking that fight.
I'm still having trouble crossing the 'AI will enhance SERPs' bridge and can't really make Kevin's leap from today's search world to that of fewer but more accurate results. I think more companies are apt to use AI-powered content to try to generate more relevance to certain search terms. Sophisticated advertisers do cost-benefit analyses to figure out how to break down critical search terms and develop a strategy for whether they want to pay for them (SEM) or chase them by developing content (SEO). Makes me wonder what the AI-generated content world will look like. Will the signal-to-noise ratio drop like it did in email? Right now, ChatGPT can feign creation of content, but it takes a lot of verbiage to say very little in terms of substance. What happens when we're flooded with content like that and how does it affect search?