2023, 2024, and 2025 have all been great years to invest in the artificial intelligence (AI) realm. With 2025 nearly over, the question becomes: Will 2026 hold more of the same? The concerns about whether the vast sums of money being spent to build out AI computing capacity are going to pay off are intensifying. Investors are starting to clamor for real returns on those investments, and there haven't been any to date.
That hasn't stopped the AI hyperscalers from continuing to add massive amounts of computing power to their footprints. In 2025, the hyperscalers set records for capital expenditures, and most of that money went toward data centers. All of them also are guiding for even greater capex in 2026.
While investors may see some risks in the amount of money being spent on artificial intelligence infrastructure, there are several ways for them to profit from the trend.
Image source: Getty Images.
AI hardware suppliers are set to deliver impressive returns in 2026
Chip designers such as Nvidia (NVDA 1.36%) and AMD (AMD 3.20%), which are supplying high-end processors to power AI, have been among the top stock choices in the sector. These two make graphics processing units (GPUs), which excel in handling AI workloads rapidly due to their ability to break down certain types of complex calculations into many smaller ones and handle those pieces in parallel.
Nvidia has been the leading AI stock ever since the infrastructure spending surge began in early 2023, and the success of its best-in-class technology stack has turned it into the largest company in the world.
NASDAQ: NVDA
Key Data Points
AMD was playing second fiddle to Nvidia before the AI megatrend took off, and it still is. Certainly it hasn't had nearly the same level of success with its wares. But the tide could be shifting as its offerings are becoming more competitive, and AI hyperscalers are hunting more earnestly for cheaper alternatives to Nvidia's chips. If the AI hyperscalers decide they want to be more budget-conscious on the infrastructure front, they could spend less money for the same amount of computing power, or the same amount of money for more, by switching to AMD's chips.
That's one reason why AMD's products could become more popular in the coming years, and management recently informed investors that they anticipate a 60% compound annual growth rate for data center revenue over the next five years.
Another niche computing provider is Broadcom (AVGO 1.54%). Broadcom doesn't market its own computing units. Instead, it goes directly to the AI hyperscalers and designs computing units in collaboration with them to meet their needs. Because it's designing each chip with a specific type of workload in mind, it can maximize performance and decrease cost by decreasing its flexibility. Several AI hyperscalers have partnered with Broadcom for this work, including Alphabet (GOOG 1.29%) (GOOGL 1.27%).
NASDAQ: AVGO
Key Data Points
Alphabet's custom AI accelerator is the Tensor Processing Unit (TPU), which it has used for internal processing and also made available to clients for lease via its cloud computing service, Google Cloud. However, Alphabet may be emerging as a force to be reckoned with in this landscape. Alphabet is reportedly considering a deal to sell a quantity of TPUs directly to Meta Platforms (META 2.43%), which would be a shift from its current model of installing them only in its own data centers. Should it decide to sell TPUs, Meta might be only the first interested buyer.
Time will tell what type of waves Alphabet makes in this industry, but all four of these stocks look like excellent buys right now, considering how much money is being spent on AI infrastructure. Buying all four would be a smart strategy because it's impossible to know which company will perform the best over the next few years. However, at least one company is likely to perform well, regardless of which company is leading in the AI chip space a few years from now.
Most top chips are made by Taiwan Semiconductor
All of these companies are "fabless" chipmakers, which means they design chips, but they don't manufacture them. Instead, they outsource that work to several different companies. However, a majority of the high-end chips being made today come from one foundry operator: Taiwan Semiconductor Manufacturing (TSM 4.74%). Taiwan Semiconductor is the world's largest chip manufacturer by revenue, and is a neutral player in the AI chip competition.
NYSE: TSM
Key Data Points
So, as long as there is increased spending on AI computing power, TSMC is bound to benefit. That makes it a no-brainer investment. It may not be the best performer of this group of five in 2026, but it should easily be the second or third best.
