AI stocks winners, laggards and losers for 2025
AI stocks flew high in 2025 despite year-end volatility. Palantir and Lam look overvalued, but opportunities remain.
Key takeaways
- AI stocks gained ground in the fourth quarter, outperforming both the US market and the tech sector despite volatility.
- Oracle and Nvidia had a bad quarter amid fears of an AI bubble.
- AI stocks currently look fairly valued, with a few attractive exceptions.
2025 was another great year for artificial intelligence stocks, even as some names hit air pockets in the fourth quarter as concerns spread about a potential bubble. In the fourth quarter, many AI-related stocks, including Oracle ORCL, suffered. Still, as a group, they managed to outperform the broader market during the period.
These stocks’ strength was evident in 2025. A basket of AI names selected by Morningstar analysts rose 50.8% over the year, while the overall stock market gained 17.3% as measured by the Morningstar US Market Index. The AI basket also beat the broader tech sector, which rose 21.4%.
Of the 34 stocks in Morningstar’s AI basket, 21 (62%) outperformed the market in 2025, with five returning more than 100%. All but one of the outperformers came from the tech sector. The exception was communication services giant Alphabet GOOG.
How AI Stocks Performed in Q4 2025
AI stocks weathered volatility in the fourth quarter as investors worried about the possibility of a bubble, but many still knocked it out of the park. The AI basket was led by semiconductor companies SK Hynix 000660 and Micron Technology MU, which each rose by over 70% in the quarter to end the year with gains well above 200%. Meanwhile, Nvidia faltered as its chips were offered at a discount despite a shortage.
The biggest loser was enterprise software company Oracle ORCL, which lost 30.5% in the quarter due to investor concerns about its AI strategy and its giant deal with OpenAI. Morningstar equity analyst Luke Yang says weakness at OpenAI also contributed to the losses: “Google came up with Gemini 3, which posed a big threat to OpenAI, Oracle’s main customer.”
Yang expects Oracle’s data center buildout to remain on time and for the company to achieve its revenue goals for Oracle Cloud Infrastructure. He says the stock is undervalued but cautions that it also carries a Very High Uncertainty Rating. “For risk-tolerant investors who have a strong conviction in the future of AI, Oracle looks like an attractive investment at its current price. But for most, it’s worth it to wait until we have more clarity about Oracle’s role in the AI ecosystem,” he says.
Which stocks are part of the AI trade?
The Morningstar equal-weighted basket of 34 stocks aims to represent a cross-section of the AI universe. The list includes companies across the AI value chain, from software to semiconductors. The largest portion comprises hardware firms involved in building or supporting AI server infrastructure, such as Taiwan Semiconductor Manufacturing TSM. Next are the so-called “hyperscalers,” such as Amazon AMZN and Alphabet, which are committing billions to the AI buildout. There are also select software firms that could benefit from advancements in AI, such as Salesforce CRM.
Best-performing AI stocks of the year
Among the best performers in 2025 were Micron, up 240%, and semiconductor equipment and materials producer Lam Research LRCX, which rose 138%. Senior equity analyst William Kerwin says both stocks benefitted from surging memory prices last year (memory chips provide storage for data centers). The shortage, which stemmed from AI infrastructure demand, allowed Micron to benefit from higher market pricing. Lam also indirectly benefitted, supplying the equipment necessary to build more memory production facilities. “Both [Lam and Micron] look overvalued after their immense runups in 2025,” Kerwin says.
Software infrastructure firm Palantir Technology PLTR also led the charge with a gain of 135% in 2025. Morningstar equity analyst Mark Giarelli points to the company’s product and execution. “[AI] is a groundbreaking technology that can increase efficiency, and Palantir is in the right place at the right time with the right product,” he says. “Their execution has been superb, and we’re seeing deeper penetration in commercial [sectors].”
Going into 2026, Giarelli expects Palantir to continue its growth undeterred. “The company does not have a clear competitor,” he says. However, despite its impressive performance and lack of competition, Giarelli believes Palantir trades in overvalued territory: “The risk-reward ratio looks tough, given the premium valuation multiple the stock trades at.”
The worst-performing AI stocks of the year
A handful of AI names ended the year in the red, including high-profile software stocks Adobe ADBE and Salesforce, each down about 20%. Morningstar equity analyst Dan Romanoff points to “fears of AI-led disruptions of the per-seat licensing model. AI is likely becoming consumption-based, meaning revenue streams will be less predictable and certain products will be disrupted.”
Despite market skepticism, Romanoff believes the companies’ fundamental performance has been solid. Plus, “If the AI trade unwinds, then more investment dollars would be available to flow back into software … I view software as generally attractively valued.”
The steepest loss came from Marvell Technology MRVL, but Kerwin sees it as undervalued. “Fundamentals look incredibly strong,” he says. “We have high expectations for growth, resulting mainly from AI infrastructure demand. Marvell has key customers like Amazon, Google, and Microsoft, and should see a strong revenue ramp across 2026-27.”
AI stocks trade within fair value, but opportunities remain
After the fourth quarter’s volatility, AI stocks looked less pricey than they did for much of the year, trading at a 4% premium to their average fair value estimate—down from an 18% premium mid-quarter.
Early in the year, the AI basket fell into discount territory following the trade uncertainty caused by President Donald Trump’s tariffs. At its lowest, these stood at a 29% discount to their average fair value estimate on April 8. But the group quickly rebounded and ended the third quarter looking expensive again, with an average premium of 13%.
Marvell is one of Morningstar’s top semiconductor picks for 2026, according to Kerwin. It’s currently trading 30% below its fair value estimate. Wide-moat software company Adobe ended the year at 39% below its fair value estimate, while wide-moat Salesforce was 20% undervalued.
