3 Top Artificial Intelligence Stocks to Buy Right Now

In This Article:

Although DeepSeek gave investors a scare when its R1 model impressed everyone with its performance at its price point, it's becoming clearer that this breakthrough wasn't a death blow to the domestic artificial intelligence (AI) industry. Instead, it challenged the notion that the U.S. is the only place where AI innovation is happening and could be an even bigger catalyst for AI over the long term.

Still, some stocks remain on sale after the AI sell-off, and I think a few are fantastic buys right now. Here's a look at three of them

1. Taiwan Semiconductor Manufacturing

DeepSeek's training was done on H800 Nvidia GPUs, which are downgraded versions from the more popular H100 GPUs in order to meet export restrictions placed on China. Inside these GPUs are chips from Taiwan, specifically Taiwan Semiconductor Manufacturing (NYSE: TSM).

TSMC is the world's top contract chip manufacturer, making chips for giants like Apple, Nvidia, and AMD. Its chips go into products all across the world, making it a benefactor of the AI movement regardless of which country is leading the race. Additionally, its chips go into other products like smartphones and cars, making it one of the most vital companies for global commerce.

All of this adds to management's bold projections that revenue will grow at a nearly 20% compound annual growth rate over the next five years. That's a strong projection, and if TSMC delivers on that projection, the stock would easily outperform the market over that time frame, as long as the stock is priced right.

Following the tech sell-off, TSMC's stock trades for about 22.5 times forward earnings, which is a great price to pay for one of the world's most important businesses. Despite fears of AI investing decreasing, TSMC is likely to continue growing over the next decade, and any marketwide sell-off is a great opportunity to scoop up more shares.

2. Meta Platforms

Meta Platforms (NASDAQ: META) may seem like an odd pick here. DeepSeek's R1 training efficiencies directly challenged Meta's Llama generative AI model, causing some investors to believe Meta may fail in the AI arms race. While DeepSeek may be more efficient, Llama is becoming more powerful.

In 2025, Meta CEO Mark Zuckerberg thinks it will be possible to build an AI engineering agent with the same prowess as a good mid-level software engineer. That's an incredible projection and shows that Meta is still at the forefront of AI innovation, even if it's spending a lot of money to achieve its goal.

Meta isn't going for the most efficient training method right now (unlike DeepSeek). Instead, it's focusing on how powerful it can make its AI, which is the better long-term strategy. Still, AI is only a cost center for Meta; it isn't making it any money.