Investors continue to scramble for new ways to play the mania over artificial intelligence software. From here, they’re going to have to think a little more outside the box.
The obvious bets are so well known that it seems everyone owns them. The consensus big winner is
(ticker: NVDA), which makes graphics processors used to train the large language models at the heart of generative AI. Nvidia shares have more than doubled this year already, and the company now ranks as the fifth most valuable tech stock, ahead of
(MSFT) was the market’s first pick on AI software, given its large stake in ChatGPT creator OpenAI and the introduction of AI features into a host of Microsoft applications. The stock is up 33% this year, and has a market value of $2.4 trillion, trailing only
(GOOGL) shares were shunned earlier this year after Microsoft unveiled an AI-powered version of its Bing search engine. But Alphabet has been investing in AI for at least a decade and recently announced AI advancements of its own. The stock is up 39% in 2023. Also creeping into the discussion:
(ORCL), which has an agreement to host Nvidia-powered supercomputing services on the Oracle Cloud. Oracle shares are pennies from an all-time high, up 26% for the year.
Then there are the crowd favorite but speculative AI plays: enterprise software maker
(PLTR) was up 23% on the week, while voice-enabled AI play
(SOUN) rallied 10%.
But if AI is going to be world-changing technology along the lines of the internet, the cloud, the smartphone, electricity, and air travel, there have to be other ways to play it, right?
I raised that question with Brook Dane, a Goldman Sachs tech portfolio manager, who was a guest this past week on Barron’s Live, our daily webcast and podcast.
“I’ve been a tech investor for more than 30 years. This is one of, if not the most exciting, developments I’ve ever seen,” he said. “The power of these models and how they will change knowledge-worker productivity…is profound…we are fully focused on finding the next AI winners.”
That said, Dane points out that it is “super early.” Microsoft said on its most recent earnings call that generative AI related workloads could add 1% to the quarterly growth of its Azure cloud business in the June quarter. That implies $150 million of incremental revenue, for a company generating about $55 billion a quarter in revenue overall. It’s barely a rounding error.
Dane sees four basic ways to play the opportunity in AI. There are pick-and-shovel plays, which enable data centers to run these AI workloads. There are infrastructure needs around AI, “given that data is the key to all this.” There are security firms. And there are applications that will benefit from the addition of AI.
In the pick-and-shovel category, Dane’s favorite pick is
(MRVL), which makes chips used in data center connectivity. Their chips, he says, assure workloads are distributed fast and efficiently.
He also thinks the AI trend will be a boost for software companies focused on electronic design automation, or EDA. In particular, Dane is bullish on
Cadence Design Systems
(CDNS). He also notes that AI workloads are memory intensive, which is bullish for DRAM and flash memory giant
As for data technology plays, Dane says he has been talking to lots of companies about how they think about deploying AI, and that a couple common themes have emerged. He says companies want to train models on their own internal data, without sharing intelligence with the world. That requires “cleaning and sandboxing” data, he says, which is a boost to companies like
(DDOG) that help companies warehouse and analyze information.
In security: “Any time you get a big change in the threat landscape, it is very good for cybersecurity companies,” Dane says. “This will introduce a whole new realm of mayhem into the world as bad actors take advantage of this technology and do things that you and I can’t even think of right now. But it’s coming…We’re going to need new forms of protection.”
His top pick to protect against the new threat is
Palo Alto Networks
(PANW). Dane says Palo Alto has data on “threat vectors” that dwarf anything their customers could compile. “They are going to run AI models across that data to identify threats earlier, faster, and better than any other companies.” He’s also bullish on
(ZS), a cloud security software company.
Finally, there are applications: “This is the earliest area, but will be the biggest over time,” Dane says. “We’re looking for the obvious places these tools can get deployed in a responsible and safe way.” One stock he likes is
(HUBS), which provides customer relationship software for small and medium-size businesses. He thinks HubSpot can use AI to make its customers more productive and efficient in their marketing spend. If HubSpot can do that, Dane says, customers will pay more. And the stock should ratchet higher.
To be clear, this is hardly an exhaustive list. I’ve written before about other companies deploying AI software, including
(NOW) just laid out a new partnership with Nvidia to drive better workflow management software for the enterprise.
The internet has changed every business. And so will AI.
Write to Eric J. Savitz at email@example.com