How to invest in artificial intelligence

"A lot of this, like self-driving cars, is very far away," says Abhinav Davuluri, a Morningstar equity analyst.

But these AI-driven innovations are coming, and many companies are already implementing AI, mostly in consumer-facing products. Voice activated technology, like  Amazon's (AMZN) Alexa or  Alphabet's (GOOG) Google Home, are good examples of AI plays today. Facebook, Alphabet and Amazon, which collect copious amounts of data on their users, are using AI to help improve searches and interact with customers.

"It's still early innings and most of the applications are for consumers--and those aren't necessarily being monetized to the fullest," says Davuluri. "A lot of companies are still figuring out how to adopt [AI]."

As companies see the value of giving people things they want before they want it, more firms are going to invest in and purchase AI tools.

"AI in general is applying technology, algorithms and programs to existing problems and coming up with better solutions," says Walter Price, a portfolio manager with Allianz Global Investors who specializes in the technology sector. "It's getting a lot easier to program and develop, and people are focused on using it to solve new problems."

Lack of pure-play options

While investing in AI companies may seem like a no-brainer -- it's hard to find anyone who doesn't think AI is the future -- it's also more complicated than it might seem. There aren't many pure-play companies, as most smaller AI-focused operations are bought by other companies, or they're content with collecting venture capital money and staying private.

In most cases, buying into AI means owning Amazon, Google, Facebook,  IBM (IBM), and  Intel (INTC) -- companies that have many other businesses outside of AI. For most, AI is just a small part of their business today. Intel doesn't break its numbers out, though Davuluri thinks that AI would account for less than 5% of the company's revenues. That will rise over the next few years for Intel and others.

"It should be very fast growing," he says.

For comparison, semiconductor company  NVIDIA (NVDA), which is considered one of the more pure-play AI operations, has exploded over the past few years. Its stock price has soared by more than 1,500% since 2012, compared to 78% for the S&P 500. Revenues have also climbed by nearly 125% to US$10 billion in fiscal year 2017 over that same time period, while diluted earnings per share have risen by 270%.

High growth, high valuations

One of the other problems with playing the AI theme right now is valuations. As of this writing, NVIDIA has a P/E ratio of 56 times earnings, while Amazon has a P/E of 249. But, like many technology companies, it's hard to value these companies on current earnings only. They carry astronomical valuations because of the promise of their future earnings. If AI does become as life-altering as many expect, then paying a P/E of 56 for Nvidia will be a steal.

For investors, it may not be a matter of if they play the AI trend, but when. Every technological revolution goes through the same kind of cycle -- mania, depression and then adoption. Scott Burns, Morningstar's head of asset management solutions, says we're likely in the mania period with AI. At some point companies will start missing their estimates and investor sentiment will turn.

"There's a difference between AI having a lot of growth in the future and AI being a great investment right now," he says. "Growth is never in a straight line."

Price also thinks the stocks are overhyped right now and points to NVIDIA as an example. He bought in to the company two years ago and only expected the stock to double in price. It's gone up by more than 500% since.

"It's become one of the most valuable semiconductor companies in the world, but there's a lot of expectation at this point," he says.

For Davuluri, the opportunity is real, but he'd rather investors buy on a price decline, rather than get in now. You don't have to wait for NVIDIA to fall to US$45, but you might want to hold off until the stock drops by 15%.

"Be patient and when there's a sell-off or a bad quarter or we get a bit of a correction, that's where the enticing opportunities arise," he says.

Pick your spots

While it still may be a while before the market dips, investors who are interested in AI should start figuring out what businesses they want to buy. There are two main subsectors with AI: semiconductors and everything else. The semiconductor companies, like Intel and NVIDIA, are creating chips used to power this technology.

The other players, like Google, Amazon, IBM and Facebook, are involved in a number of AI initiatives, but again, they have many other lines of business. They key is figuring out which companies will benefit most from AI, whether that's using AI to further their own operations or to develop products they might sell to others.

"The most capable AI groups are the Googles, Amazons and  Microsofts (MSFT)," says Price. "But you need to look at the fundamentals in the business and see what AI would add it to it."

Both Alphabet and Amazon are interesting, adds Davuluri. Alphabet would be the more AI-centric play of the two; it's working on a number of AI-related products. Amazon could benefit from AI in other ways. For instance, its budding Amazon Web Services division, which is the only profitable part of the company, is the go-to cloud storage for many companies. The more data that gets stored on its servers -- and AI requires copious amounts of data to work -- the more that part of Amazon's business will grow.

On the chip side, Intel presents an opportunity if it falls a bit, says Davuluri. He likes some of the AI-related acquisitions the company has made; it bought deep learning company Nervana in August 2016 for about US$400 million, for instance, and he thinks the company could look quite different in five or 10 years from now as AI becomes a bigger part of its business.

 Qualcomm (QCOM) also has an interesting future, says Davuluri. It has a lot of exposure to smartphones, and it has done poorly because of its legal battle with   Apple (AAPL), but it will likely start developing more AI-related chips.

"It's not a pure play, but it is a noteworthy name," he says. It's trading in 4-star range as of this writing, suggesting that shares are undervalued.

Investors looking to own a basket of IA plays might look at Global X Robotics & Artificial Intelligence ETF (BOTZ), which is up more than 50% for the year-to-date as of this writing. The portfolio includes NVIDIA, Japan's Keyence (a sensor company), and  Intuitive Surgical (ISRG) (a healthcare robotics company). Morningstar analysts don't cover the fund, nor is it rated.

Those investors who feel a bit more adventurous and want to benefit from AI in a different way may want to consider Horizons Active A.I. Global Equity (MIND), an ETF launched earlier this week that is entirely run by an adaptive artificial intelligence system. The ETF's sponsor, Horizons ETFs Management (Canada) Inc., describes it as the first of its kind in the world.

In any case, the message for investors is: Pick your spots, in most places wait for a pullback to get in, and then watch the returns add up.

"We're not at the stage of disappointment yet, but it could be on its way," says Price. "But then you can make more money when the reality stage comes in."