Nvidia (NVDA -1.10%) has become the poster child of the artificial intelligence (AI) revolution. Its AI data center chips were in hot demand in 2023, sending the company's revenue surging by triple-digit percentages throughout the year.
But Cathie Wood, who runs Ark Investment Management, believes hardware is just the beginning of the AI opportunity. She thinks software companies using AI will generate eight dollars in revenue for every dollar worth of hardware Nvidia sells, and it makes sense because software can be developed once and sold an unlimited number of times.
In fact, Ark's analysis suggests AI software could generate $14 trillion in revenue per year by 2030.
Social media giant Meta Platforms (META -2.31%) might become one of the greatest value creators in this space. The company is currently worth $985 billion -- but here's why it could eventually eclipse Nvidia's $1.4 trillion market cap.
Image source: Getty Images.
Facebook and Instagram are increasingly powered by AI
First, let's explore Meta's core business. The company generates nearly all of its revenue by selling digital advertising space to businesses on its leading social media platforms Facebook, Instagram, and WhatsApp. They attract 3.1 billion users every day, so Meta is the go-to place for any merchant trying to reach a large audience.
If you're a regular user like me, you've probably noticed an increasing amount of content showing up in your feeds that isn't posted by anybody you follow. It's because AI now curates over 40% of the content we see on Instagram.In other words, AI is learning what we enjoy viewing so it can show us more posts we would (theoretically) like to see.
It appears to be working. In the recent third quarter of 2023 (ended Sept. 30), CEO Mark Zuckerberg said AI drove a 6% increase in the time users were spending on Instagram, and a 7% increase on Facebook.
Meta continues to develop generative AI tools for businesses, too, which can help them create more engaging ads. It can craft their text-based content, help them design images, and test different variations of an ad to optimize performance. This creates a better experience for the user because they will see higher quality ad content, which is also more likely to be relevant to them. In turn, that drives better results for businesses and entices them to spend more money.
Meta is also developing large language models
Outside of Meta's core platforms, the company is also developing large language models (LLMs), which are the foundation of every AI application. The quality of an LLM determines the accuracy of an AI-powered chatbot, and the depth of its knowledge.
Meta released its Llama 2 LLM in July last year, but it's very different from other popular LLMs like those developed by ChatGPT creator OpenAI -- Llama 2 is entirely open source (with some restrictions in place for safety purposes).
That means Meta provides the code, the license, and everything else a developer needs to use Llama 2 to build their own AI applications, even for commercial purposes. By offering the model for free, Meta can observe how it performs in a number of different environments, which is valuable information for the development of its own AI applications.
After all, Meta is a giant tech company with over $100 billion in annual revenue already. The value it creates from an LLM will come from how it applies the model on its own platforms to enhance the experience of users and advertisers.
American users can already access a new chatbot called Meta AI. It's a virtual assistant available within Messenger, and it's capable of generating images on request, answering questions, and even sending you Reels (videos) when you ask which places you should visit.
As Meta AI becomes more accurate over time, it could drive a surge in engagement on the company's social media platforms. Eventually, it might even reduce the need for users to visit third-party search engines for the information they seek. The more time each user spends on Facebook, Instagram, or WhatsApp, the more opportunities Meta has to show them ads to generate revenue.
Why Meta could eventually be worth more than Nvidia
Mark Zuckerberg just announced Meta will be purchasing 350,000 of Nvidia's H100 data center GPUs, which are the go-to chips for any company training LLMs (until the H200 becomes widely available later this year). It's going to cost Meta an estimated $9 billion, and it will increase the company's total compute capacity to around 600,000 H100 equivalents.
That will accelerate Meta's progress in the AI sphere so it can bring new products to market more quickly. Remember, since Meta has 3.1 billion people using its social networks each day, it has more data with which to develop AI than almost any other company -- except maybe Alphabet, the parent company of Google.
But one of the best reasons to own Meta stock is its valuation. Wall Street expects the company to deliver a record $17.55 in earnings per share in 2024, placing its stock at a forward price to earnings (P/E) ratio of just 21.9. If that estimate proves accurate, Meta stock will have to rise 38% this year just to trade in line with the Nasdaq-100 index, which trades at a P/E of 30.3 today.
Based on Wall Street's estimate for Nvidia's earnings in fiscal 2025 (ending Jan. 31, 2025), its stock already trades at a forward P/E of 28.9. The company has made a habit of beating forecasts lately, but the big question is how long it can keep up its impressive performance. The semiconductor industry is cyclical by nature, and once all the largest tech companies have fitted their data centers with the latest chips, attention will turn to the upgrade cycle, which tends to drive slower growth.
To be clear, I'm not suggesting Nvidia stock is a bad investment. However, over a time horizon of five years or more, I think Meta will be the larger company, and its stock presents a great opportunity at the current price.
