Earlier this week, many technology stocks sold off sharply on news of DeepSeek R1, the groundbreaking, low-cost AI model from China that materially lowered the cost of deploying AI.
But has that sell-off opened up a buying opportunity? Here are three tech stocks that should not only weather the current AI disruption, but also thrive on the other side.
Amazon
Amazon (AMZN -0.45%) founder Jeff Bezos once said, “Your margin is my opportunity.” So if artificial intelligence becomes more commoditized and the largest-scale, lowest-cost player wins, there’s perhaps no company better positioned than Amazon.
If the cost of artificial intelligence comes far down, that will level the playing field for a lot of digital businesses. However, it’s still hard to challenge large-scale players in the physical world. In other words, it may be easy to use AI to spin up a digital website and offer services, but it’s another thing to deliver millions of items to people’s doorstep within a day. So Amazon’s e-commerce moat stands to endure the AI revolution. And Amazon should also be able to implement AI robotics to further optimize its delivery ecosystem and lower costs even further.
Meanwhile, Amazon Web Services may also have an advantage. AWS is the oldest and largest cloud platform and has committed to giving customers as much choice as possible at the lowest possible costs.
Since Amazon’s rival Microsoft (MSFT -1.09%) leapt out and invested early in OpenAI, it had been thought by some that AWS was behind in the AI races. However, if large models become more commoditized, that Microsoft-OpenAI first-mover advantage won’t mean much.
But rather than tying itself to one AI leader, Amazon has committed to hosting a variety of large language models on its Amazon Bedrock platform. Meanwhile, Amazon has also invested in OpenAI rival Anthropic, based on the results and usage it was seeing for Anthropic’s Claude models on AWS.
Amazon is also committed to drive down the cost of training and inference with its homemade Trainium and Inferentia chips. By designing its own hardware in-house, Amazon offers dramatically lower costs than renting Nvidia GPUs. In fact, even Antrhopic, as a leading frontier model builder, has committed to training future models on Trainium. That could give it a cost advantage over other model-builders.
On a big-picture level, Amazon’s history shows it has continually invested in new ideas and innovations, perhaps at the expense of near-term profits. It also usually vertically integrates, solidifying a cost advantage and increasing its moats versus its rivals in large industries.
This all makes Amazon a stock with staying power, which means long-term investors can still feel comfortable buying, even at recent all-time highs.
Tencent
There are a few issues that DeepSeek brought to the market’s attention. One is that if the cost of training AI models comes down materially, AI will be embedded into more and more existing services. The second is that China’s tech community can also pioneer groundbreaking innovations.
Both issues should be a benefit to Tencent (TCEHY 0.08%), which is not only developing its own AI services, but also has several massive existing businesses that should benefit from AI.
After all, Tencent owns WeChat, the largest social network in China, with 1.38 billion monthly active users. That is probably the largest trove of proprietary data available in China, or probably anywhere in the world. Given the democratization of large language models, companies with large distribution systems and proprietary data like Tencent will be able to leverage AI to the greatest advantage.
Image source: Getty Images.
Not to be outdone, Tencent’s other main businesses in video game publishing and fintech should also see benefits from AI. Video game studios can use AI to more efficiently create better games with more monetization opportunities, and Tencent’s fintech arm, Tenpay, can leverage AI to better underwrite loans and detect fraud in digital payments.
Finally, Tencent also writes huge amounts of software across its businesses, and one of AI’s early main use cases is for software coding. If Tencent is able to lower the number of software engineers it needs, there is a huge opportunity to lower its costs.
Meanwhile, China’s stock valuations, despite a rally last year, are still well off their highs and below those of their U.S. counterparts. Thus, this could still be a good time to buy best of breed Chinese stocks like Tencent, for those willing to take the geopolitical risk.
Applied Materials
The first reaction for the semiconductor industry to the DeepSeek news was a violent sell-off. This included most hardware names, from chipmakers to server-makers to semiconductor equipment stocks such as industry leader Applied Materials (AMAT 1.41%).
However, that could be short-sighted. If the cost of AI training comes down, that should spark an explosion of use cases for AI, which is likely to increase computing at the edge. That means edge servers, phones, and PCs. So even if the most advanced AI training chips see a deceleration from expectations, inference chips in these other devices should then take off. Overall, the pie is likely to grow.
Applied Materials is a great choice for those looking for broad-based exposure to the chip industry, as the most diversified equipment-maker across both market segments and types of chip equipment. The company makes etch and deposition, metrology, ion implantation, and advanced packaging machines, among others, serving the leading-edge logic, lagging-edge logic, DRAM and NAND memory markets.
This diversified exposure makes Applied a safe choice to benefit from the overall growth of computing. Add in a services business at around 23% of revenue and largely tied to Applied’s growing installed base of equipment, and Applied is a lower-risk and profitable way to play the growth of computing over the long term.
The large semiconductor equipment companies also form an oligopoly, with only one or two choices at each process step. That enables all players to generate high margins and return on capital. That in turn allows Applied to repurchase its stock consistently, even through downcycles, and pay a rising dividend.
So even if investment in the most leading-edge training GPUs moderates, investment in edge computing segments, which have recently lagged, should pick up in a big way. Either way, Applied should grow along with the overall computing intensity of the economy, all while returning lots of cash to shareholders over the long term.
