During Nvidia's (NVDA -0.77%) Q2 conference call, it made a jaw-dropping market projection: Management believes global data center capital expenditures will reach $3 trillion to $4 trillion by 2030. That's a ton of money being spent on AI infrastructure, and if it's right, investors should be racing into AI stocks to capitalize on this spending.

However, I'm not referring to companies that are spending the money on AI infrastructure; I'm investing in the companies that are providing the infrastructure. I think Nvidia, Taiwan Semiconductor (TSM 1.42%), and Broadcom (AVGO -0.00%) are among the best buys in the market right now, and if this projection is correct, they will make investors a ton of money over the next five years.

Person looking at a computer screen shocked in excitement.

Image source: Getty Images.

Nvidia's projection isn't as aggressive as one may think

At first, this projection may seem outlandish. However, it isn't as far-fetched as it seems when you break it down. During its Q2 conference call, Nvidia stated it projects 2025's data center capital expenditure figure to total $600 billion (originally, CEO Jensen Huang stated that this was the big four AI hyperscalers, but was later corrected to be all AI data centers). So, global data center expenditures are expected to rise $600 billion to $4 trillion by 2030, which would require a compound annual growth rate (CAGR) of 46%. That's huge growth, but is it realistic?

While AI hyperscalers are currently spending a lot on this infrastructure, all of them have stated that they expect substantial increases in capital expenditures in 2026. I'd expect this trend to continue, as many of the data center projects that were announced in 2025 will begin construction in 2026 and be outfitted with computing equipment in 2027 or 2028. This stretches out the record-breaking capex spend over multiple years.

Additionally, the U.S. isn't the only area where the AI revolution is ongoing. Chinese-based AI companies will also be spending a ton of money on AI, and it's unlikely that their American counterparts will vastly outspend them. Europe is just now getting into the AI game, and it has massive resources that could rival the U.S. or China when combined.

While the $4 trillion high end of this estimation is still incredibly optimistic, it underscores that there is still a ton of spending left to realize in the AI arms race. This makes a handful of stock genius investments, and I think investors need to ensure their portfolio has exposure to Nvidia, Taiwan Semiconductor, and Broadcom.

This trio is set to cash in on massive AI spending

Nvidia has been the king of AI investing since the start of the AI arms race. Its graphics processing units (GPUs) have become the computing unit of choice for training AI models due to their incredible flexibility and best-in-class performance. Nvidia is one of the companies most likely to benefit from increased AI spending, and if you don't own shares of Nvidia, now is not too late.

Broadcom is starting to challenge Nvidia in the AI computing unit field. While Nvidia's GPUs can be adapted to nearly any computing situation, Broadcom's computing units couldn't care less about that. Instead, they're partnering with an AI hyperscaler to design custom AI accelerator chips that excel in only one type of workload. Because the hyperscalers know what their AI workloads will look like, they can purchase chips through Broadcom that have better performance and cost less than an Nvidia GPU, but can only be used for one purpose. I think this option will become far more popular in the next few years, making Broadcom a worthy competitor to Nvidia.

Lastly, there is Taiwan Semiconductor. Neither Nvidia nor Broadcom has the capabilities to produce their own chips. So, they farm that work out to the world's leading semiconductor manufacturer. Because Taiwan Semi is acting as a fabrication company that's not trying to compete directly in the AI arms race, it is positioned nicely to benefit from any increased AI spending. Odds are high that any chip being used in the AI arms race originated from a TSMC facility, making it a winner as long as data center capital expenditures are increasing.

All three of these companies are promising and can continue their dominant run over the next five years. If you don't own shares now, it's not too late to buy, as there could be incredible upside if Nvidia's management is correct on the general market trend.