The artificial intelligence revolution is upon us, and it doesn’t show signs yet that it’s slowing down. The spending boom on AI chatbots and agents has enterprise and cloud data center providers rushing to refresh equipment, replacing older servers powered by CPUs with next-generation racks powered by chips and memory specially designed for AI.
We haven’t seen this level of excitement — and such a massive retooling of IT budgets — since the dawn of the internet. Only time will tell if AI’s impact will exceed that of the internet, but companies are betting big that it will.
Hyperscalers, the biggest data center providers on the planet, are spending hundreds of billions on the proverbial picks and shovels necessary to develop and run AI models.
For instance, money manager Fidelity Investments wrote recently in a research report that Amazon’s AWS, Microsoft’s Azure, Google Cloud, and Meta Platforms have increased spending “from roughly $100 billion in 2023 to more than $300 billion in 2025 — a figure that could exceed half a trillion dollars within the next few years.”
The AI gold rush has similarly won over the hearts and minds of investors, who have bid up top AI stocks over the past couple of years in anticipation that all these investments will eventually translate into bigger profits.
And it’s not just mom-and-pop investors who are making the bet. The largest funds in the country have invested heavily in AI stocks, including Fidelity, a giant with $5.9 trillion in discretionary assets under management.
Fidelity’s portfolio managers have recently shared their thoughts on AI stocks, highlighting top picks in their funds that are benefiting from the AI boom.
One of the biggest criticisms of the big-cap tech hyperscalers has been that the size and pace of spending are too optimistic, and that it will be years, if ever, before sales and profits tied to AI justify the cost.
Nvidia CEO Jensen Huang is riding an AI spending wave.Shutterstock
While it remains to be seen how that argument shakes out, the big tech stocks are starting to reap some benefits. Fidelity portfolio manager Priyanshu Bakshi, who manages the Fidelity Select Communication Services Portfolio (FBMPX), believes they aren’t overvalued.
Despite big-cap tech stocks growing to represent over one-third of the S&P 500, Bakshi points out that members of the Mag 7 (Nvidia, Microsoft, Apple, Alphabet, Amazon, Meta, and Tesla) are delivering “mid-20% range” earnings growth, outpacing “mid-single-digit growth for the rest of the S&P 500.”
Bakshi’s two biggest holdings — Alphabet and Meta — generate significant revenue from advertising, and they’re already profiting from AI improvements. The two, which account for nearly 50% of the Select Communication Services Portfolio, collectively generate $500 billion in digital ad sales.
AI tools that can deliver more relevant ads, likely to translate into sales and higher ad rates, could drive future growth.
“I believe there’s still a lot of runway for improvement,” said Bakshi.
Alphabet (GOOGL) targets capital expenditures of $91 to $93 billion this year, while Meta Platforms (META) targets spending of $70 billion to $72 billion.
Nvidia’s sales have surged from $27 billion in 2022, when ChatGPT launched, to $187 billion over the past 12 months. The company’s success is directly tied to its graphic processing units, or GPUs, which are far better at handling AI training and inference than legacy CPUs deployed within data centers.
Nvidia’s (NVDA) lineup, including its H100 and H200, built on the Hopper architecture, and more recently, the B100, B200, and GB200 Superchip, built on the Blackwell architecture, are the de facto gold standard for AI speed and efficiency.
The company’s next-generation AI chips, built on its Vera Rubin platform, are expected to become available in 2026.
Those chips are manufactured by Taiwan Semiconductor Manufacturing Company (TSM), the world’s largest contract chipmaker. Most of its manufacturing is done in Taiwan, but TSMC operates fabs in Washington and has a growing next-gen fab in Arizona.
Fidelity fund manager Chris Lin, who manages Fidelity OTC, thinks that despite the hype, people are still underestimating AI’s potential.
“Nobody knows how long it will take to play out, but I believe most investors are underestimating how impactful AI will ultimately be,” said Lin.
As of the end of October, Nvidia is Fidelity OTC‘s biggest holding, comprising 15.7% of its $35 billion portfolio, and Taiwan Semi is the 9th largest position, accounting for 2.8%.
“AI requires computation, and these 2 companies are the main providers of it,” said Lin.
Nvidia is the Goliath in the space, but the AI buildout is also boosting demand at other chipmakers, including those building custom XPUs designed explicitly for use in hyperscaler networks and larger, super-fast memory.
Fidelity’s Adam Benjamin, who manages the Fidelity Select Technology Portfolio (FSPTX), believes that achieving human-like intelligence will require much more than simply GPUs.
“Nvidia isn’t just a chip company anymore,” said Benjamin. “They’re selling full rack-scale systems — essentially complete supercomputers designed to train and run AI models… The next wave of gains is happening at the system level, not the chip level. This is a rack-scale problem now.”
Solving the rack scale problem creates sales and profit opportunities for Broadcom (AVGO) and Marvell Technology (MRVL), which manufacture XPUs and interconnect equipment that enables server networks to work together. Memory manufacturer Micron (MU) is also a beneficiary as racks become increasingly packed with more memory, including next-generation high-bandwidth memory, or HBM.
Marvell Technology is the fourth-largest holding (5%) in the Select Technology Portfolio as of October 31, while Micron ranks 10th (2.6%).
AI training and running AI apps require massive amounts of power, and that’s already taxing existing power generation grids.
Generating sufficient power to support increasingly larger and more powerful data centers presents growing opportunities for related companies, according to Clayton Pfannenstiel, co-manager of the Fidelity Select Industrials Portfolio (FIDRX) at Fidelity.
Natural gas turbines are among Pfannenstiel’s favorite solutions to easing the power bottleneck. While small nuclear reactor technology could benefit players like Rolls-Royce in the long term, those are unlikely to arrive until the 2030s.
“If we need power now, the main source is gas turbines,” said Pfannenstiel.
GE Vernova (GEV) is his fourth-largest holding, accounting for 5.4% of the portfolio. Pfannenstiel says it’s already seeing higher natural gas turbine orders, driving management to boost its earnings guidance.
He’s also a fan of Eaton (ETN), which sells and manages electrical systems necessary to power data centers, and Trane Technologies (TT), an HVAC giant that’s seeing demand growth as data center cooling demand rises. Trane is the fund’s second-largest holding, and Eaton is the eighth-largest holding.
For similar reasons, Fidelity’s Shilpa Mehra, manager of the Fidelity Growth Strategies Fund (FDEGX) and Fidelity Trend Fund (FTRNX), owns shares in Comfort Systems USA (FIX) and EMCOR Group (EME).
“AI is still in the build phase,” said Pfannenstiel. “There are a lot of ‘picks and shovels’ companies that could potentially benefit.”
The use of natural gas turbines understandably supports natural gas demand, providing a tailwind for Energy Transfer (ET), a midstream player that builds and operates the pipelines and processing facilities, according to Kristen Dougherty, manager of Fidelity Select Energy Portfolio (FSENX).
Overall, many are debating whether the run-up in AI stocks has created a bubble, as we witnessed with the internet. Those comparisons may be off the market, though, suggests Fidelity Director of Global Macro Jurrien Timmer.
Unlike during the internet boom, when spending was overwhelmingly funded by debt, it’s being mainly financed with cash by highly profitable companies like Google this time around.
Also, during the internet boom — and trust me, I know, given I worked as a Wall Street analyst at the time — people bid no-revenue, no-earnings companies to extraordinary highs.
“Valuations today are not even close to what’s been experienced during bubble extremes of the past,” said Timmer.
For example, Nvidia’s forward P/E ratio, a standard valuation measure, is approximately 24 — hardly high relative to what we saw from leading internet companies at their peak.
