Artificial intelligence (AI) took some massive steps forward in 2023. Following the release of OpenAI's ChatGPT in late 2022, practically every executive was talking about how they were applying the advancements in technology to their businesses.
While start-ups like OpenAI and Anthropic garnered a lot of attention and massive investments from Microsoft and Amazon, respectively, one company has been building advanced AI applications for well over a decade. And despite a strong stock performance in 2023, its potential may still be underappreciated.
That's why Meta Platforms (META 2.33%) is the smartest AI stock to buy right now.
Image source: Getty Images.
Increasing the investment in AI
Some of the biggest winners in artificial intelligence are big, deep-pocketed companies like Microsoft, Amazon, and Meta. That's because it takes massive capital expenditures and a lot of research and development to improve AI algorithms.
Even after making several cuts to its capital expenditure budget last year, Meta spent around $28 billion in 2023. It expects $30 billion to $35 billion in capital expenditures next year, driven by investments in AI hardware and data centers. Meta's total expenses will approach $99 billion. Few other companies can compete with that level of spending.
Meta ended the third quarter with $61 billion in cash and equivalents on its balance sheet and just $18 billion in long-term debt. On top of that, it's producing huge amounts of free cash flow -- $13.6 billion in the third quarter. That points to the fact that it's taking its huge AI investments and turning them toward generating revenue from its massive user base.
Perhaps Meta's biggest AI advancements in 2023 were in its machine-learning algorithms. These are the algorithms that learn what users want and don't want and surface more engaging content and advertisements.
Meta made a significant push in its recommendation algorithm following the release of Reels on Instagram and Facebook. While Reels increased engagement, the way users interacted with Reels made them harder to monetize. As of the end of the third quarter, though, the impact of Reels became revenue neutral, ahead of management's original forecast.
Meta is also at the forefront of generative AI and large language models. Its Llama 2 model has outperformed OpenAI's model in some instances.
The company is also applying generative AI to advertising products. It makes it easier for marketers to create and test ad designs and campaigns and optimize their results. As such, they're willing to pay more for ads on Facebook and Instagram than other platforms.
In October, Meta introduced its generative AI-powered chatbots for Messenger and WhatsApp. It also promised businesses the opportunity to easily create their own chatbots, a feature OpenAI quickly moved to copy.
But Meta has the advantage of tens of millions of businesses already using its messaging platforms to interact with customers. Bringing them the capability to make their own AI-powered chatbot can boost its WhatsApp for Business revenue and click-to-message advertisements, which is already a $10 billion business.
Meta is a leader in generative AI efficiency
Meta's Llama 2 isn't the most powerful generative AI model available. While it can outperform leading models in some tasks, it often trails by a slim margin on many tests.
Where Llama 2 shines through is in its efficiency. For example, Llama 2 can summarize text at about 1/30th the cost of GPT-4. And when you scale that over an entire enterprise operation with thousands of users, that can add up to a lot of money.
Efficiency will also be key going forward as power consumption becomes a limiting factor for companies developing AI applications. While chipmakers have an opportunity to help reduce the power-consumption cost of cutting-edge AI development, developing more power-efficient AI models is a more sustainable approach. And that's where Meta's models truly shine.
Image source: Getty Images.
As such, there should be ample opportunity for Meta to play a role in the foundation of future generative AI applications with its open-source Llama model. And as the open-source community adopts Llama, it should help Meta further improve its model for the next Llama generation. Not only does that open the door for meaningful licensing revenue (despite calling it open-source, Meta still charges for commercial licenses with over a certain amount of users), it'll also support Meta's own AI applications, fueling its core revenue sources in its family of apps.
With the stock trading for less than 20x analysts' 2024 earnings expectations, Meta is relatively inexpensive. That's especially true when compared to other deep-pocketed companies investing in AI and considering the outlook for its overall earnings growth.
Microsoft, for example, trades for around 30x forward earnings, and analysts don't expect it to grow its earnings nearly as fast. That's why there aren't any better AI stocks you could buy with $500 today.
