/Technology%20abstract%20by%20Joshua%20Sortino%20via%20Unsplash.jpg)
Stock market sentiment regarding artificial intelligence (AI) has soured significantly in the past few months. The DeepSeek spook damaged the argument that AI GPU and compute demand was going to not only stick around, but continue to grow exponentially. Then, analysts reported that Microsoft (MSFT) was canceling some of its data enter leases, sending AI stocks further into a downturn.
The biggest losers were data center pure-plays and energy companies that were thought to benefit significantly from increased electricity demand. That said, it’s not a good idea to be entirely bearish on all AI stocks since there is still demand for AI hardware, and cloud businesses are still growing.
Amazon (AMZN) and Google’s (GOOGL)cloud businesses are still growing, even if the pace has slowed. Microsoft’s retreat from some leases could simply mean it’s optimizing its footprint or redirecting resources to more efficient setups. In the meantime, others could take advantage of the void left by Microsoft.
Here are two AI stocks that could be winners:
AI Stock #1: Alphabet
Alphabet (GOOGL) has seen its own share of volatility due to there being chatter about the company’s cloud business slowing down and the search segment losing ground to AI competitors. The perception hit hard after DeepSeek’s low-cost AI model spooked the market in January and the market still seems unsure whether or not Google’s heavy investments in AI infrastructure will pay off against leaner rivals.
TD Cowen analyst Michael Elias offers a compelling counterpoint. He notes that Microsoft’s retreat from over 2 gigawatts of data center capacity in the U.S. and Europe over the past six months has opened doors for Alphabet. Elias’s channel checks at Nvidia’s (NVDA) GTC event and DCD Connect show that Google is stepping in to backfill capacity internationally.
There is a “global capacity shortfall” after an August 2024 pullback, the analyst says. Back then, Google shifted focus to squeezing more out of its existing data centers. Now, demand is surging again. Elias ties this ramp to Google’s internal AI push, and I believe he’s likely to be right since Google’s search revenue is still growing and its own AI efforts are picking up steam. The company rolled out Gemini 2.5 Pro in March 2025 and its own AI search model.
It still trades at less than 20 times earnings as of writing.
The mean price target of $217.39 implies 38% upside potential from here.
AI Stock #2: Meta Platforms
Meta Platforms (META) has been among the strongest performers over the past two years, and the recent pullback has barely made a dent in its earlier gains. That’s also why I am less bullish here than on GOOG due to its high valuation.
Elias sees a silver lining amid Microsoft’s lease pullback. He notes that Meta is backfilling capacity in the U.S. that Microsoft abandoned and has snapped up over 1 gigawatt of data center space in the past month alone. Elias ties this to a “material year-over-year ramp” in Meta’s data center demand due to Meta’s Llama AI model.
Regardless, I’m not as bullish on META right now. Part of that is due to valuations, but a big part of that is due to Meta’s Llama AI model itself, which trails most other flagship AI models in performance. The company is still heavily reliant on its “Family of Apps” for cash flow, which it is pouring into AI and VR investments. The current valuation, with sharse trading at nearly 23 times forward earnings, is too rosy due to the AI growth bump, so I believe you should be more careful.
The mean price target of $746 implies 27% upside.
