The four phases of the AI revolution

27 March 2024
the four phases of the AI revolution

While Nvidia’s 73% rise this year makes it the poster child for the AI investing boom, there’s a lot more to AI than a chipmaker.
Research from Goldman Sachs has broken down the AI revolution into four phases, looking to identify the stocks and sectors set to benefit most after Nvidia.

(To have this post read to you, click here)

Phase 1: Nvidia

Phase one began less than a year and a half ago, with the release of OpenAI’s ChatGPT. The generative AI bot took the world by storm, sparking a surge in demand for shares of Nvidia, whose AI-powering chips made it the market’s clearest near-term AI beneficiary.

What’s the opportunity?

Despite Nvidia’s massive share price surge (dark blue line), its current valuation – judging by its forward-looking price-to-earnings ratio (or P/E, in gray) – is almost unchanged. And that’s because its estimated earnings per share (or EPS, in light blue) have been shooting rapidly higher too.

Nvidia stock price

This chart looks great right now, but as companies get bigger, they usually have a harder time achieving rapid growth and maintaining high margins. For now, it seems right to just enjoy the investing ride, knowing that, eventually, even Nvidia will see its growth slow.

Phase 2: AI infrastructure

AI is somewhat like its own little world, with a ton of firms involved in its infrastructure. These include semiconductor companies whose chips are the critical components needed to train and use AI. And there are subsectors within semiconductors, including design companies such as ARM, and fabless designers (who design but don’t manufacture) like Broadcom and Advanced Micro Devices. There are also memory companies like Micron, and semiconductor manufacturing equipment companies like Applied Materials and ASML.

Plus, there are data center companies, such as Equinix, which own and operate physical places that contain the servers needed to train and run AI models. And there are hardware and equipment companies, like Vertiv, supplying the things needed to build and operate the data centers.

Utility companies, like Duke Energy, meanwhile, deliver vast amounts of electricity to power the data centers. Cloud providers, like Microsoft, Amazon, and Alphabet help train, run, and maintain AI models through their computing and data storage solutions. And security companies like CrowdStrike work to keep data safe.

What’s the opportunity?

Goldman looked closely at these companies’ EPS estimates and P/E ratios three years into the future – and then accounted for various unforeseen factors. After all, lots of things can impact valuations and stock prices. And this is how the investment bank sees things panning out.


The companies that fall below the “best fit” gray diagonal line are potentially more attractive to investors, because they have higher expected profit growth relative to their P/E valuations.

For example, in the bottom right corner of the chart, “foundry and integrated device manufacturers” look to have a relatively good-looking setup of strong expected EPS growth with modest valuations. The companies in that subsector are TSMC, Intel, and Global Foundries.

Meanwhile, security stocks, which appear in the top right of the chart, have strong earnings expectations but could struggle to perform, since they already trade at very high valuations, with their prices above 50x their earnings (that’s more than double the average for companies in the S&P 500).

Utilities, at the bottom left of the chart, could represent an interesting opportunity because they look cheap in valuation terms. But since they operate in a highly regulated industry, they could continue to struggle to make meaningful gains.

Overall, it’s worth noting that semiconductors are still a big component of Phase 2 companies, and so the iShares Semiconductor ETF could be a good place to look for some broad-based AI exposure.

Phase 3: AI-enabled sales

The companies that will benefit at this stage are the ones that can add AI into their product offerings to boost what they sell. Many of them are software and IT services companies, like Meta, MongoDB, Intuit, Nutanix, ServiceNow, and Uber.

What’s the opportunity?

Loads of software and IT services firms have already described how they’ll take advantage of AI. So Goldman had a look across the industry, scouting for companies whose share prices have high beta (or volatility), and have been rising with Nvidia’s. Then they zeroed in on the ones whose executives specifically mentioned AI in their latest quarterly earnings calls with investors. That left the bank with a long list of companies. Here are the top 15.

AI-enabled sales.

One can invest in the firms individually or gain access to a broad batch of software and IT services companies through an ETF, like the iShares Expanded Tech-Software Sector ETF.

Phase 4: AI productivity

Eventually, the AI trade will focus on companies that are using AI to improve productivity – across a wide range of industries – with the biggest potential likely to be in more labor-intensive industries.

Goldman estimated that profits at software and IT services, and commercial and professional service companies could benefit the most. After all, a hefty segment of their well-paid workers are in jobs that could be replaced by AI automation, which could dramatically reduce those firms’ labor costs. Goldman’s report highlighted Match Group and News Corp. as a couple of prime examples.

What’s the opportunity?

Phase four is about the grand promise of the technology: the way things shift when AI is adopted by a wide range of companies. Goldman sees software and IT services firms, and commercial and professional services firms affected the most, as AI adoption improves their labor productivity and dramatically shrinks their wage costs. This next chart plots the potential productivity gains that could be felt across various sectors and industries.

exposure to automation by AI

To bet on AI’s fourth phase, you could dip back into the 15 companies from phase three’s handy chart or beef up your allocation to the iShares Expanded Tech-Software Sector ETF.

The big picture

It’s important to remember that we’re still in the early stages with AI. Phase one star Nvidia has been the standout performer, by far, but there have been gains in stocks from across each of the tech’s phases.

Taking on more risk can potentially bring bigger returns. And picking stocks and sectors is generally riskier than choosing the broader market index. If AI is set to improve productivity across a huge number of sectors and the wider economy, as suggested in phase four, owning the S&P 500 through the SPDR S&P 500 ETF Trust (SPY; 0.09%) or the Invesco S&P 500 Equal Weight ETF (RSP; 0.2%) could well prove to be a good long-term bet.

Related posts