There are many stories here. First, there’s the story of what Microsoft is trying to do, and how it’s doing it—by investing heavily in OpenAI to create something that’s disruptive, in a business that it’s held on to for years despite a tiny market share. There’s also the story of Google, which is facing its biggest challenge to its core business. Then, there’s the story of why investors are watching Google’s moves very closely and punishing them for minor mistakes. There’s also the story of who will ultimately prevail in this war between Google and Microsoft, and how.

Today’s edition is about all of these stories.

But above all, it’s the story of Google’s dance. 

Let’s dive in.

You don’t need me to tell you that companies typically don’t lose market value because their platform has weird, inaccurate, or cringe content. (See: Reddit, Twitter, and LinkedIn, respectively). 

And yet, for some reason, everyone seems paranoid about Google. They believe that ChatGPT is a fundamental change for Microsoft, and that this poses a serious threat to Google’s search business. There are many ideas and theories about why this is the case. Some of these are real, but the vast majority of them are misconceptions because most people don’t fully understand how search and advertising work. 

Today, I’m going to break these down. 

Google is not being disrupted technologically by ChatGPT

I’ve written about ChatGPT before, and how far it can go to fool humans. For many people, for good reason, ChatGPT looks and feels smarter than the usual chatbots out there. Plus, as Microsoft highlighted multiple times in their presentation on Tuesday, search technology hasn’t changed for a long time, and ChatGPT could represent a paradigm shift—as disruptive as the iPhone or the internet. 

Sure, it may seem imperfect today, but that’s the thing about disruption. It creeps up on you, and over time, it takes over. This has been famously explained (at length) by Clayton Christensen, the academic and business consultant in his landmark work on disruptive innovation.

At first glance, it appears that this is exactly what ChatGPT is about to do to Google. 

Except, it isn’t. Because Google’s problem isn’t about technology at all. In fact, by all accounts, Google has been working on AI models for years now. Among all the companies out there, Google is the one that has vast data sets that includes everything that has ever existed on the internet. It even has access to training sets gained by digitisation of millions of books. Last year, one of Google’s engineers was convinced that its chatbot has “achieved sentience”. Of all the companies out there, Google has probably spent billions of dollars to build AI into its products. Sure, ChatGPT is impressive, but if you asked me whose technology is more likely to be better in the medium-term, I’d pick Google any day.

Plus, there’s something that isn’t that well-known. 

ChatGPT is built using the Transformer research project—a research paper which Google released in 2017. The T in ChatGPT stands for Transformer. 

I’m not saying that ChatGPT’s technology isn’t great, but if Google gave Transformer away for free, I can only begin to imagine what they didn’t, and how sophisticated it must be by now. 

Google’s distribution is far superior to Microsoft and Open AI

During the Microsoft event, the company touted its unfair advantages from a distribution standpoint. Primarily, it went to great lengths to stress that it has both a browser (which is how most users go online) and a search engine (which is how users search once they go online). Per Microsoft, this is what gives them the perfect opportunity to integrate ChatGPT. They control the stack. This is the bedrock of their disruption.

But, err… you know who else owns both a browser and a search engine?

And not just any browser or search engine, but the world’s most popular ones. 

A lot is generally written about Microsoft’s power of distribution. On the enterprise side, that is what makes Microsoft formidable. And I foresee Microsoft making applications on its products like Teams, and even Office, supported by ChatGPT, which can lead to some interesting features. 

However, on the consumer side, I find it hard to believe that Microsoft will win because it has Edge and Bing. Again, put them on one side and put Google Search and Chrome on the other, and I think it’s pretty evident which of these will be more likely to distribute their technology faster through consumer applications. 

Google is not being complacent

Back in December, when ChatGPT broke into the public sphere, Google’s management team saw its potential and  immediately issued a “code-red”—essentially setting off an internal fire alarm. The company did multiple high-level meetings, trying to figure out how to respond and protect their business. 

Then a month later, they called in reinforcements.

Think about this for a moment. Google’s AI technology and its resources surpass Open AI, and arguably even Microsoft. Its product is competitive. It owns all the key distribution channels, i.e., the search engine, browser, and operating system. It’s not even a slim lead—in all three categories, Google leads by a massive margin. And finally, it has the organisational will. It wants to beat ChatGPT. It’s taking this threat seriously. 

Why then are people betting against it? 

The answer lies in how search advertising works. 

As you can imagine, the biggest worry that analysts and observers have with Google is that once it transitions to a ChatGPT-like search result, how would it make money off it?

Search advertising has two unique characteristics. 

First, it captures the highest form of intent, which makes it much more valuable for advertisers. If I’m an advertiser, I’m looking to show my product to people who are likely to buy it, or at least try it out. I can find these users either through inference—like working out that people who are likely to consume my product read a certain publication, or have certain demographic characteristics. So I put an ad in a news website, or target them on Facebook. Again, this is advanced guesswork. However, when you search for something, you’re essentially broadcasting a signal about what you are explicitly looking for. If I put an ad at that moment, right in front of your face, at the point where you are looking for it, chances are you’ll be interested. This is what makes search advertising so incredibly effective as a product. 

Second, it’s the most profitable software category out there. Most software is profitable at the gross level because of near-zero variable costs, but it’s hard to think of any category that’s more profitable than online search. The costs are minor, and you get to charge advertisers a premium, because it’s much more effective. 

Imagine owning a company that controls over 93% of online search globally, combined with a business model that gives them a 55% gross margin. No wonder it’s one of the mightiest companies in the world.

So no, ChatGPT does not attack Google’s technology, distribution, or organisation. It attacks the business model. In a world where search results are delivered as text and not links, it’s unclear how Google will show ads.

Satya Nadella has stated that he believes that the high-profit-margin search business is coming to an end. In his words, “From now on, the [gross margin] of search is going to drop forever”. 

Microsoft’s search revenue from Bing is somewhere around US$11 billion dollars—a paltry sum if you compare it to Google’s figure of over US$200 billion. 

For Microsoft, search may be a side-hustle, but Nadella sees it differently. 

Here’s what he said in that same interview with Nilay Patel at The Verge.

In other words, Microsoft and Open AI do not care about the business model. They do not care about how to make search ads highly profitable. Here’s Sam Altman, CEO of Open AI, answering a similar question from Ben Thompson of Stratechery. 

Notice how he answers the question of the business model.

This is the summary of Google’s position. It wants to act and it has all the tools it needs to succeed, but it would have preferred to transition slowly and deliberately to the new world order. 

Except, OpenAI and Microsoft have forced its hand, forcing it to move fast, and break things. Google would have wanted to wait, taking classes, followed by closed-door performances and then a few special appearances, before finally transitioning to a massive public display. Expectations are high, and Google would have preferred to make its mistakes at home.

But now, Google must learn how to dance in full public view.