Newsletter #4- Bard-a-Bing, Bada Boom

Last week, Google announced Bard, their answer to OpenAI's ChatGPT. However, Google lost $100B in market capital after Bard provided a wrong answer. What's even worse is that the Bing demo had several mistakes that no one noticed till recently. Many assume this is a David vs Goliath case, but it is not. Microsoft is a $2 trillion market cap behemoth similar to Google. The market crashed under the assumption that Google's business model will not work in the new era. In this article, I will explain why the market freak-out was unnecessary.

First, let's take a look at Google business model.

Google's business model is centered around its search engine, Google Search, and the advertising revenue generated from it. The company also generates revenue from other products and services, such as Gmail, Google Maps, YouTube, and Google Cloud.

1. Advertising: Google's primary source of revenue is advertising. Advertisers bid for keywords related to their products or services and pay when users click on their advertisements. This is known as the pay-per-click (PPC) model. Google's advertising platform, Google AdWords, allows advertisers to place ads on Google Search, Google Maps, Gmail, YouTube, and other websites that are part of the Google Network.
2. Data Collection and Analysis: Google collects vast amounts of data on users through their searches and other Google services, such as Gmail and Google Maps. This data is used to target advertisements to specific users based on their interests and search history. Additionally, Google's machine learning algorithms analyze user data to continuously improve the relevance and quality of its advertising offerings.
3. Traffic Acquisition Costs (TAC): TAC refers to the cost incurred by Google to deliver traffic to its advertising partners. TAC includes the cost of acquiring traffic from partners who use Google's search services on their websites or the cost of acquiring traffic for Google's own websites, such as Google Search, Gmail, and Google Maps. Did you know Google pays Apple a hefty sum to be the default search in their hardware?
4. Other Products and Services: In addition to advertising and search, Google generates revenue from other products and services, such as Gmail, Google Maps, YouTube, and Google Cloud. These products and services are offered for free or at a low cost and often serve as a way to drive traffic and collect data, which can then be used to inform and improve Google's advertising offerings.

Let's zoom into primary source of revenue- Advertising. The total Digital Ads market for 2023 is estimated to be $703 billion, with Google capturing around 28% of this market. In 2022, Google made $224 billion from ads, which accounted for 80% of its overall revenue. Satya Nadella, the CEO of Microsoft, has stated that he is on a war path to capture a share of this market by leveraging an upgraded version of Bing with the help of GPT-4 infused Bing.

Let's take a closer look at their respective cost model. Google charges around 3 cents per search (i.e., revenue divided by the number of meaningful clicks), with the cost of goods sold (COGS) being around 1.6 cents per search. On the other hand, it is estimated that ChatGPT will cost around 30 cents per query, due to the difference in tokens between traditional Google search and conversational Bing chatbots which output paragraphs. Assuming all things equal(LLM optimization, LLM quality, model related metrics), "Bard infused Google search's Chatbot" will incur similar cost of around 30 cents/query. But here's why Google will make more revenue(margins) than Microsoft in search- chatbot market.

The unique value captured by Google is as follows: Bard is trained and inferenced on Tensor Processing Units (TPUs), specially designed to run machine learning workloads. These specialized silicon devices are designed to be 30x faster than traditional state-of-the-art GPUs, which allows Google to process more queries in the same amount of time, leading to more revenue generated per click compared to Microsoft's Bing. Additionally, TPUs offer reduced latency, meaning faster inferencing and display of results on your screen, which is a crucial metric for customers. The two most dominant players in mobile phone market are iOS and Android. Google is the default search engine across both platforms deepening its moat.

Despite the higher cost of goods sold (COGS) associated with chatbots compared to regular search, Google's use of TPUs provides a differentiated silicon "moat" that Microsoft simply can't catch up on. This advantage has been built over the last decade, making it difficult for competitors to replicate Google's technology and stay competitive in the market. Also Google has boatload of cash to burn while iteratively improving software/hardware lockstep to make it economically viable similar to existing business model. Google also has pre- collected data of users, crawlers which would provide leg-up compared to Bing.

All this being said, I would still agree with market sentiment on Google moving at a crawl pace. They need a wartime CEO, not a peace time CEO.

Given the high cost of Chatbots, it may not be the most effective approach for Microsoft to put all its eggs in one basket to disrupt Google. Instead, Microsoft could consider diversifying by focusing on disrupting other industries with OpenAI relationship, such as the video game industry.

In recent years, the gaming industry has seen a significant shift towards more immersive, interactive experiences that rely heavily on artificial intelligence and natural language processing. By leveraging ChatGPT, Microsoft could potentially develop chatbots that are capable of conversing with players in a more natural and intuitive way, creating a more engaging and personalized gaming experience for its XBOX console.

Microsoft could work with OpenAI to explore new paradigms of AI-generated video games(text to video-video to text Foundational Models), leveraging successful historic franchises. This approach could potentially provide a way to reduce dependency on traditional publishing houses like EA or Activision Blizzard. By creating dedicated games that work in tandem with ChatGPT, Microsoft could redefine our view of gaming industry.

Comment below or DM!

#semiconductor #microsoft #google #chatbots #semiconductorindustry

In this newsletter, I intend to cover various technology trends ranging from Cloud to AI to Quantum, powered by "Semiconductors- The New Oil". Follow me on LinkedIn and subscribe below.