There are a significant number of investment opportunities in the artificial intelligence space, but investors will have to be patient to see those investments pay out, according to portfolio managers at Columbia Threadneedle Investments.

As AI captures the imagination of individuals and businesses alike, the Boston-based firm hosted a webinar in mid-July highlighting areas where it made sense to invest in AI. The presenters pointed out that the tech is still in its infancy.

“I think it’s really important for investors to understand that the impact of these investments and opportunities may take years for most companies to monetize and subsequently turn into additional sales opportunities or productivity gains,” said Tiffany Wade, a senior portfolio manager at Columbia Threadneedle.

The reason for that delay is there is a lot of preliminary work firms must conduct. They will have to gather their own data before building and testing AI models. They also need a vetting process to test the security and utility of the models, she said.

That’s not to say that there are no current opportunities to invest in AI. Columbia Threadneedle is focusing on groups that are building the infrastructure for the technology.

“In the short and medium term, we want to follow where we think the dollars are going,” Wade said. “Right now, the biggest opportunity is in the picks and shovels of AI.”

The current opportunities include cloud providers and data centers, she said. The companies looking to launch an AI service will need cloud providers to run those applications. Other companies are creating the building blocks of AI, the different data models that determine how the technology learns.

Another area that investors can focus on is internet security. As more data is transferred and used by AI applications, there will be a growing need to protect potentially sensitive information, according to Wade.

“Companies face traditional security threats such as data breaches but will also face new threats such as malware that could potentially disrupt or alter AI model output or enhanced security threats that are driven by AI,” she said.

Investors could also look into companies from a wide variety of sectors that incorporate AI into their business platforms to improve efficiency and productivity, including companies that have cost-heavy help desks, Wade said.

One of the industries that can benefit from AI is agriculture; farmers can use the technology in smart farming, said Dave Egan, a senior equities analyst at Columbia Threadneedle. AI will know where and how many seeds to plant, as well as how best to use pesticides and fertilizer. It will also help farmers increase their crop yields.

In marketing, AI is being used to generate ads, which brings down the cost of advertising and allows companies to tailor their advertising and marketing to a specific audience, Egan said.

The portfolio managers said, however, that advisors and investors should be cautious before putting their money into a particular company.

“One needs to be wary of pretenders versus contenders,” said Rahul Narang, another senior portfolio manager at the firm. “There are companies that will see the benefit more immediately from AI … and then there are companies that will see it much later in time.”

Columbia Threadneedle has been investing in AI for years, but has done so in a disciplined manner, which is how other investors should approach the technology as well, Narang said. An advisor or investor needs to do their own research on an investment, since many companies make AI references during earnings calls, he explained.

When conducting research on a company, advisors should look into whether it has a competitive moat around its business and if it understands how its product will differ from others in the market, he said. The company should also understand the competitive landscape it’s in.

There are still risks involved when it comes to AI investing, Narang said. The first has to do with regulations. Since this is a new technology, it is unclear what rules will be implemented and how they’ll affect the space.

Data ownership is also an issue, he said. There will be a question about who owns AI data and who is ultimately responsible for it.

Narang cautioned advisors and investors not to get too engrossed by the marketing pitch of a company or to fall for trends when making an investment decision.

“I remember during the late 1990s … a company would add ‘dot-com’ to their company name and the stock would go up multiple times,” he said. “It’s important not to get caught up in the hype of AI marketing.”