Andrew Ng Thinks AI Apps are “Underinvested”—Here are the Stocks I’d Buy in Response

While others see an inflating AI bubble that could burst and lose investors a boatload of money, others see underinvestment in some specific parts of the theme. Undoubtedly, it’s hard to know with any degree of certainty where the AI trade goes next, especially as other volatility drivers (think Trump tariffs) are thrown into the mix.

Either way, I do think that value investors who insist on modest multiples and discounted long-term growth stories might still be able to score pretty good gains over time, as most others begin to throw in the towel on the broad basket of AI names.

Of course, the Magnificent Seven remain the gold standards as far as long-term AI strategies are concerned (and yes, that includes Apple (NASDAQ:AAPL | AAPL Price Prediction), a misunderstood name which many still view as lagging behind in AI). But with significant selling pressure on about half of the Mag Seven names, including Apple, Microsoft (NASDAQ:MSFT), and Meta Platforms (NASDAQ:META), which are down around 13%, 17%, and 18%, respectively, from their all-time highs, one could argue that the lagging half of the Mag Seven are now undervalued bargains, even if the AI trade is poised to experience bigger bumps in 2026, a year that could get rougher.

Deutsche Bank said that the “honeymoon is over” for the AI trade. And that could spell trouble for the broader basket of names, including the Mag Seven heroes that have more recently fallen out of favor. In a climate where the bears might be starting to take control, there are still several bullish voices that stand out. Andrew Ng is one of the pundits who believes that there’s still opportunity to be had in the AI trade.

Looking to the application layer for value

Specifically, Mr. Ng pointed to the “application layer” as one that could be underinvested by investors at this point in time. Of course, with so much hype behind the frontier model makers, the chip makers (storage stocks have really taken off of late), and the infrastructure plays (think the data center top dogs), there’s a good chance that the beaten-down software plays might be skewing towards the undervalued side.

At the end of the day, I do think that Ng is on the right track. AI-native apps seem necessary if AI tech is to become more useful (applied), such that customers become more willing to spend. If such AI apps can add more value than they cost firms, it’s a pretty easy sell.

With Claude Code and Cowork sparking agentic AI disruption fears that pummelled software stocks, including those that are well-equipped to benefit from gains to be had in the “application layer,” I think there’s an opportunity for contrarians to step in at a reasonable multiple into what could be the most intriguing GARP (growth at a reasonable price) plays within the AI trade.

Of course, it’s easy to hit the panic button over the potential for agents to code software platforms similar (or even better) than the popular SaaS ones that exist today. But could it be that it’s not so simple for agents to erode the software moats, given the data advantages and everything else that goes behind a pretty user interface? Perhaps. Either way, I think AI-native software firms with distinct advantages could be worth consideration right here.

You don’t have to look far for AI software bargains

Salesforce (NYSE:CRM) and ServiceNow (NASDAQ:NOW) stand out as software firms caught in the downdraft that may deserve to experience a swift recovery as investors better recognize what they bring to the table. Notably, Salesforce and ServiceNow have pivoted towards agents in a big way.

With BNP Paribas recently highlighting ServiceNow as one of the most “resilient” software firms, I do think value seekers might have a chance to pick up a few shares on the way up. As agents take off, perhaps ServiceNow is the firm that can make such technologies better, as the firm looks to act as more of a platform for end-to-end automation. 

Either way, shares of both companies have fallen under pressure and could be compelling buys if the application layer is where the puck could be headed next in the AI trade.