The headline numbers from Big Tech’s AI infrastructure buildout are staggering. Microsoft, Google, Amazon, and Meta have collectively committed over $725 billion in capital expenditure toward AI infrastructure – data centers, networking equipment, custom chips, and the land and power systems to run them. Nvidia captures most of the investor attention as the primary beneficiary of this spending wave, but two other categories of companies are quietly reaping enormous rewards from the buildout, largely unnoticed by mainstream financial coverage.

The Power Infrastructure Play

Every AI data center is, at its core, a machine that converts electricity into intelligence. The computational demands of training and running large AI models are extraordinary, and the physical infrastructure required to deliver, condition, and distribute electrical power at the scale these facilities need has created a massive opportunity for companies that most tech investors have never considered.

Power management companies, electrical equipment manufacturers, and the utilities themselves are experiencing demand growth they have not seen in decades. Companies that make transformers, switchgear, uninterruptible power supplies, and cooling systems for data centers are sold out months or years in advance. The demand is so far exceeding supply in some categories that it has become a genuine constraint on how quickly AI data centers can actually be built and switched on.

• Vertiv Holdings, which makes thermal management and power distribution equipment for data centers, has seen its stock price and order backlog surge as AI data center construction accelerates.
• Eaton Corporation’s electrical components division is booking data center orders at unprecedented rates.
• Nuclear power is being reconsidered as the only energy source capable of delivering the reliable, carbon-free electricity that AI data centers need at the scale required – leading to investment interest in nuclear energy companies that had been largely ignored by growth investors for years.

The Fiber Optic and Networking Infrastructure Play

The second underappreciated beneficiary is the networking layer that connects AI data centers to each other and to the internet. AI models are not just trained and stored in a single data center – they are distributed across multiple facilities, require massive data transfers during training runs, and must serve inference requests to users globally with low latency. This creates enormous demand for optical networking components, fiber optic cables, and the specialized switches and routers that connect GPU clusters internally.

Companies like Coherent, II-VI, and Ciena that manufacture the optical components enabling high-bandwidth data center interconnects have seen demand surge. The buildout of undersea fiber cables to support AI traffic between continents has also accelerated, benefiting specialized marine cable laying companies that most technology investors have never encountered.

Why These Companies Get Less Attention

Power infrastructure and networking components are not glamorous. They do not generate the same narrative excitement as AI chips or cloud software. But the investment case is compelling: without adequate power and networking, even the most advanced AI chips are useless, and the companies supplying these essential inputs have pricing power that comes from being bottlenecks in a supply chain that the world’s largest technology companies are desperate to unclog.

Frequently Asked Questions

Are power infrastructure stocks still a good investment after their recent run-up?

The AI infrastructure buildout is a multi-year trend with committed capital already allocated. Whether current prices adequately reflect this tailwind requires individual due diligence – consult a licensed financial advisor before making investment decisions.