Now that ChatGPT unleashed a firestorm of interest in artificial intelligence investing, it may be time to urge some calm for investors. Since OpenAI’s ChatGPT was made available to the public, investors have bid up any stocks playing the trend. Microsoft , which announced an AI-powered version of its Bing search engine, has jumped more than 12% to start the year, better than the S & P 500 ‘s more than 7% gain. Meanwhile, competitor Alphabet is about 10% higher. Chip stocks are also riding the wave of interest, with the iShares Semiconductor ETF more than 23% higher this year. And Nvidia ? Up a whopping 55% this year. Not to mention other stocks on the trend such as C3.ai , which has more than doubled, surging 128%. Audio and speech recognition stock SoundHound AI skyrocketed 121%. AI YTD mountain AI software provider C3.ai has seen its stock more than doubled as investors look to hop on the artificial intelligence trend. Still, investors should maintain some caution. While AI is promising, there are pitfalls to navigate as with popular investing trends in the past. While AI has been around for a while, the technology remains costly, and difficult for companies to scale. “It’s still early days and I think betting on individual companies, especially unprofitable companies, is still very risky and speculative,” said Irene Tunkel, chief U.S. equity strategist at BCA Research. Here’s the bear case for artificial intelligence. It’s expensive Users may find ChatGPT’s human-like responses to prompts for poems written in the style of William Shakespeare impressive, but the computing power required for those responses would make generative AI hard to profit from for many companies. Take ChatGPT’s cost per query, which Morgan Stanley’s Brian Nowak last month estimated would amount to 2 cents on average — or anywhere from $0.004 or $0.044 depending on the size or complexity of the query and response. That’s about seven times more than the cost to run a typical Google search, which Nowak estimated to be $0.003, or not even half of one cent. That will make the cost of building out generative AI capabilities “significant” for Google-parent Alphabet, let alone any other company trying to build out language processing models. “The extent to which AI/natural language models like ChapGPT penetrate search, we estimate every 10% of GOOGL search queries that are replaced with a natural language query (similar to ChatGPT) in ’25 would add ~$0.6bn-11.5bn to GOOGL’s [operational expenditure] depending on the words per query result,” Nowak wrote. Google’s dominance Even without those costs, companies expanding their language processing capabilities will find it difficult to make money if they fail to take market share from Google. Google has roughly 93% of the search engine market share worldwide , according to data from web traffic analysis firm StatCounter. Microsoft’s Bing, on the other hand? Just 3%. And other competitors such as Yahoo! , Russia’s dominant search engine Yandex, or China’s Baidu have even smaller slices of that global pie. “I think it’s worthwhile for Microsoft to pursue it, because I don’t think the Google function is perfect, by any means. And I kind of feel like the chatbot that’s used for OpenAI is a better mousetrap than the Google search engine,” said Pat Tschosik, senior portfolio strategist at Ned Davis Research. “But outside of Microsoft, I question how many companies have the technology and the resources and the profitability to make money,” Tschosik added. The perils of AI gaffes Meanwhile, there are other challenges. Since generative AI is only considered as good as the large sets of data it’s trained on, there could be limits to the accuracy of its responses. It’s a point Alphabet found to its detriment when an advertisement for its own AI chatbot Bard showed an incorrect answer to a prompt. Separately, the lack of proper citation from generative AI could mean future lawsuits for AI companies. While ChatGPT can scrape websites for data to generate its responses on everything, it would need to properly source its material to avoid plagiarizing copy. “At least Wikipedia is good about citing the source of everything,” said Tschosik. “But the ChatGPT isn’t listing any sources for its responses, and maybe that’ll be something that has to come down the road.” How to invest in A.I. A.I. ETFs are rallying as ChatGPT hysteria heats up. What to know before jumping in Morgan Stanley says these stocks will benefit from an A.I. boom Heavyweight tech battle: How Microsoft and Alphabet stack up against each other in the A.I. war Still, the need for proprietary data could spell opportunity for some smaller players to compete, according to Tschosik. The strategist expects that investors could find companies who capitalize on the trend in business, health care, and others. “There’s a long lead time in developing the technology,” Tschosik said. “But if there’s money to be made, you know, these companies will do it. It’s not impossible.” However, choosing those companies will remain challenging for investors, which is why BCA Research’s Tunkel said investors are better off buying an AI exchange traded fund, or a pooled investment. “I’m absolutely sure about its future,” said Tunkel, but added, “I don’t think that it’s possible to pick a winner here.”