To The Moon

The AI boom is generating extraordinary profits, but investors are beginning to ask a difficult question: how much money will the industry ultimately make from the billions being poured into data centres?

The latest earnings projections highlight both the opportunity and the risk. Nvidia and Micron are expected to contribute a substantial share of S&P 500 earnings growth, with Micron’s quarterly profit reportedly surging 1,078% year over year and analysts forecasting approximately 90% earnings growth for Nvidia.

Yet the market’s reaction to AI spending plans suggests that strong earnings alone may no longer be enough. Investors also want evidence that today’s infrastructure investments will generate sustainable returns.

Here are three developments worth watching.

1. The rest of the market is holding up better than some investors think

Despite the concentration of earnings growth among major AI companies, the broader S&P 500 is not standing still.

According to the figures cited in the report, companies outside the top 10 are still expected to deliver approximately 10% earnings growth year over year.

That matters because a healthy market does not necessarily require every company to match Nvidia’s growth rate. If earnings continue expanding across different industries, the market may have a broader foundation than the AI headlines suggest.

However, investors still need to distinguish between companies delivering sustainable earnings growth and those benefiting from temporary pricing power or unusually strong demand.

2. The real risk is whether AI spending delivers adequate returns

Building data centres requires enormous capital investment. Companies must pay for chips, electricity, cooling systems, buildings and network infrastructure before they can generate returns from their AI services.

If demand grows fast enough, these investments could prove highly profitable. But if computing capacity expands faster than customers’ willingness to pay, the industry could face excess capacity and weaker returns.

Meta’s plans to raise additional capital have already highlighted investor sensitivity to the cost of financing AI expansion.

The concern is not that AI has no commercial value. It is whether the eventual profits will justify the scale of spending and the expectations currently reflected in company valuations.

3. Could electricity providers become the next beneficiaries?

One interesting development is the potential rotation from companies supplying AI hardware towards businesses providing the electricity needed to run data centres.

Chipmakers have benefited from intense demand and strong pricing power. But as memory-chip supply constraints eventually ease, investors may look for other ways to benefit from continued AI infrastructure growth.

Electricity demand could remain elevated as data centres expand, even if the pace of chip-price increases moderates.

Companies such as Constellation Energy (CEG), Vistra (VST) and NRG Energy (NRG) offer different forms of exposure to the US power market.

However, rising electricity demand does not automatically translate into higher shareholder returns. Power prices, regulation, capital expenditure and existing valuations all matter.

What this means for investors

The AI investment story may be entering a more selective phase.

The first stage rewarded companies supplying scarce chips and computing equipment. The next stage may place greater emphasis on which businesses can convert AI spending into durable cash flow and profits.

That does not mean semiconductor stocks have reached their peak, nor does it guarantee that utilities will outperform. It means the investment case is evolving.

My takeaway: AI earnings remain powerful, but investors are increasingly looking beyond revenue growth to the returns generated by the enormous capital being invested.

The companies that benefit most over the next few years may not necessarily be those spending the most on AI, but those that can turn that spending into sustainable profits.

For discussion purposes only.