I’ll start with the AI ad problem. At this point in GPT-4’s development, it seems like training data is what will differentiate one large language model (LLM) from another. So, if you have unique sources of data like Tweets or customer reviews, you might receive unique responses from the AI model. One possible future is that these LLMs could be built into the backend of many of our tech services. This outcome could hurt ad models in the long term. In theory, fewer eyeballs at Google means less ad revenue.

While it’s hard to imagine today what AI-powered ads will eventually look like, I believe advertisers will find a way into the AI ad funnel. Simply put, I don’t believe AI will kill advertising. Consumers want free services, and ads are the best way to support those products.

Over the past hundred years, incumbent advertising structures have evolved to work with new tech-enabled paradigms. When radio came out, advertisements built on print shifted to live read spots. When TV launched, the live read spots migrated and were quickly replaced by the 30-second ad format, bringing in an entertainment component to the ad experience. When the internet grew, display ads, search, and the 10-second video ad formats took off. Mobile introduced native ads through social feeds.

Now we’re entering the age of AI powered search and discovery and a new ad model paradigm will emerge over the next decade. Like the past century, economic forces will once again play a central role in paving the way for advertisers to enter the funnel when we’re interacting with a chatbot.

Since 2019, Microsoft has invested about $13B into OpenAI. That level of investment has been a springboard for Microsoft to lead in showcasing how AI can improve its products. The path is most clear for Microsoft given I estimate more than 80% of the business will eventually have some form of AI assistant.

• Microsoft 365. Likely monetization approach: raise prices. This week they announced “Copilot” an AI assistant for Microsoft 365 that will write emails and proposals, build Excel models, and draft PowerPoint presentations.
• Azure. Likely monetization approach: raise prices/charge for models. As Microsoft adds AI models like chat and image processing to Azure’s library, the company will have opportunities to raise prices or charge for models that previously were not available.
• Bing. Likely monetization approach: advertising. In February the company began to test Bing with AI powered results. While the ad model with AI search is to be determined, history would suggest it’s a function of time before advertisers find a way to benefit from the AI theme.

Google has been an AI first company since 2017, and its 90% plus global search market share (excluding China and Russia) will feed its AI models and make them smarter.

• Search & YouTube. Likely monetization approach: advertising. These two segments account for about 80% of overall revenue with Search accounting for about 60% of revenue. Search’s strong suit is answering queries that fall into four categories: internet navigation, maps, shopping, and general queries like daily weather, copywriting, or summarizing records. The use of Google is more weighted toward shopping and general queries. Similar to my thoughts on Bing, while it’s unclear what form search AI ads will take, it’s a function of time before these ad units emerge.
• Google Cloud. Likely monetization approach: raise prices/charge for models. This will follow a similar approach to the one Azure and AWS will take, charging for AI models that developers can plug into applications hosted in the cloud.

One factor that is often missed in the AI conversation is Apple is the only company that owns both mobile and desktop operating systems, and AI in its most useful form will be the fabric of an operating system that enables new features within apps.

• iOS & MacOS. Likely monetization approach: raise hardware prices. Given Apple owns both the mobile and desktop operating systems the opportunities to integrate AI are endless. Today AI can be found in how Photos organize albums, inside of Apple Watch with fall detection, and in iMessage with suggested responses. In the future, I could see using iMessage or Safari as a search substitute to find and order a product, make a recommendation for dinner, or give advice on a book to read. Apple Watch will only get smarter with AI and AI powered apps in the App Store will generate Services revenue.

Lastly, don’t forget about Tim Cook’s 2017 assessment that “clearly, one purpose of autonomous systems is self-driving cars — there are others. And we sort of see it as the mother all AI projects”. The following year Google’s head of AI, John Giannandrea, moved over to Apple as the company’s head of machine learning and AI strategy. As for the probability of an Apple branded car, it’s a coin toss if it sees the light of day, despite the fact the company continues to invest heavily in the project.