Veteran analyst reveals 2 ‘must-own’ AI stocks

4 min read

If you want to play AI, Dan Niles says, stop overthinking it.

In a recent Fox Business interview, the veteran tech analyst boiled it down to two stocks. For those seeking the strongest exposure to the massive AI spending boom, he recommends Nvidia (NVDA) and Broadcom (AVGO).

Own them both, and you’re “pretty much covered,” Niles said.

For perspective, Dan Niles is a bona fide Wall Street veteran, with over 35 years in the game.

Consequently, he’s lived through several tech booms and busts, watching earnings and capex numbers like a hawk.

He currently serves as portfolio manager at Niles Investment Management, a company he founded back in 2004.

Moreover, many of the largest fund managers have continued buying Nvidia and Broadcom, despite the broader market weakness.

For instance, Cathie Wood’s ARK Invest Feb. 17 daily disclosure showed $8 million in Broadcom buys.

Also, billionaire Ken Griffin’s Citadel doubled its stake in AI bellwether Nvidia, building the position to roughly $4 billion.

On top of that, Ray Dalio’s Bridgewater invested $253 million in Nvidia stock, which now accounts for nearly 2.63% of the fund’s portfolio. 

As Niles puts it, if companies are effectively doubling down on AI infrastructure, Nvidia and Broadcom are the standouts.

Dan Niles says owning two key AI infrastructure stocks covers investors.

Photo by PATRICK T. FALLON on Getty Images

Nvidia vs. Broadcom returns comparison

  • 1M: Nvidia +0.90% versus Broadcom -5.04%
  • 6M: Nvidia +6.98% versus Broadcom +13.25%
  • YTD: Nvidia +0.75% versus Broadcom -3.50%
  • 1Y: Nvidia +34.96% versus Broadcom +46.02%
  • 3Y: Nvidia +778.53% versus Broadcom +460.77%
  • 5Y: Nvidia +1,158.83% versus Broadcom +581.67%
    Source: Seeking Alpha

Nvidia and Broadcom sit at the center of the AI boom

As Niles told Fox Business, the case for owning Nvidia and Broadcom comes down to the capital-spending math.

He pointed to a major shift in market expectations. 

Niles argued that just 45 days ago, hyperscaler AI capex growth was estimated to be at nearly 25% to 30%. After earnings dropped, he says the number now looks much closer to 60%.

And as companies effectively double down on AI infrastructure, Niles says there are only two clear options to capture that spending.

Nvidia is the dominant supplier of high-end GPUs that power AI training and inference. On the other hand, Broadcom designs tailor-made ASICs for hyperscalers building AI systems in their own data centers. Collectively, they cover the two most crucial hardware paths in the AI ecosystem.

Valuation is at the heart of Niles’ thesis.

He notes that Nvidia stock currently trades at 25 times forward earnings (46-47x trailing P/E), which is only slightly higher than the S&P 500, trading near 23 times. 

More AI Stocks:

That’s a compelling price point, and Niles says Nvidia offers a far more attractive risk-reward profile than software names that trade closer to 40 times earnings.

His broader point is that when capex is rising and valuations remain reasonable, he prefers owning businesses that sell picks and shovels, not those that are just hoping to monetize AI indirectly.

For Niles, the AI trade isn’t complicated.

Niles explains why Meta Platforms can spend huge and still deliver

Niles laid out a split-screen on another major tech giant in Meta Platforms.

He argues that even though the company has been great at using AI to make money, it still isn’t one of the companies that’s selling AI to everyone else.

On a more bullish note, Niles says that Meta’s core strength of turning AI into engagement and monetization continues to trickle down to its bottom-line profits. 

Related: Morgan Stanley issues sharp take on the stock market

That’s backed up by Meta’s mighty impressive recent numbers. 

It posted almost $60 billion in Q4 sales (up 24% year over year), with EPS of $8.88 (+11%), spearheaded by a massive 41% operating margin. The ad machine also remains hot, with ad impressions and price per ad numbers up 18% and 6% respectively, in Q4.

However, the big problem, according to Niles, is that Meta currently has neither a large, public-facing language model like Gemini/Grok/ChatGPT, nor a mature public cloud.

That’s exactly why Mr. Market gets nervous about the company, fearing it has moved from generating cash to burning cash as it goes toe-to-toe with cloud giants.

A $115-$135 billion capex number feeds those fears. But Meta’s roughly $50 billion partnership with Nvidia, noted by Reuters, could serve as validation for the spending ramp. 

Related: Legendary fund manager drops $2.52 billion on mega-cap tech stock