Last week, machines bought $86 billion worth of stocks in five days. The headlines screamed about hedge fund exposure and bank risks, but they missed the bigger story: this isn’t just a market event—it’s a capital expenditure supercycle that’s rewriting the rules of investing.

The real action isn’t in the trading desks. It’s in the data centers—where Amazon, Microsoft, Alphabet, Meta, and Oracle are dropping a combined $700 billion in 2026 alone. That’s a 69% year-over-year increase, and it’s just the beginning. By 2030, AI infrastructure spending is projected to hit $5 trillion.

This isn’t a bubble. It’s a buildout. And the last time we saw this kind of capex, the railroads were laying track across America.
Watch the full breakdown on Ox Talks

The $700 Billion Bet: Who’s Spending What

Let’s break it down. The five hyperscalers—Amazon, Microsoft, Alphabet, Meta, and Oracle—are leading the charge. Their 2026 capex budgets are eye-watering:

- Microsoft: $140 billion (up 50% YoY) - Amazon: $130 billion (up 45% YoY) - Alphabet: $120 billion (up 55% YoY) - Meta: $110 billion (up 60% YoY) - Oracle: $100 billion (up 70% YoY)

For context, that’s more than the GDP of most countries. And it’s not just about buying GPUs—though Nvidia is capturing ~90% of AI accelerator revenue. It’s about building the physical infrastructure: data centers, cooling systems, power grids, and the software to tie it all together.

🏰
MPLY’s Dominance Scoring System is built for this. Nvidia, Microsoft, and Equinix aren’t just leaders—they’re oligopolies with moats as deep as the Grand Canyon. If you believe in the AI capex supercycle, MPLY is the fund that owns the companies with the strongest competitive advantages.

Why This Time Is Different

Every capex boom gets compared to the railroads or the internet. But this one has a twist: the customers are also the builders.

In the 1800s, railroads laid track to connect cities. In the 1990s, telecoms built fiber to connect people. Today, the hyperscalers are building data centers to connect machines. And those machines? They’re not just consumers—they’re the ones driving demand.

This creates a self-reinforcing cycle. More AI models = more data = more compute needed = more capex. It’s a flywheel, and it’s spinning faster than any capex boom in history.

A side-by-side comparison of the railroad expansion in the 1800s, the internet buildout in the 1990s, and the AI data center buildout in the 2020s, showing the exponential growth in capex and infrastructure scale.
The railroads, the internet, and now AI. Each capex supercycle rewrote the rules of investing.

The Risks: When Capex Becomes a Casino

Of course, it’s not all sunshine and GPUs. The S&P’s warning about bank exposure to hedge funds isn’t just noise—it’s a reminder that leverage loves a capex boom, and leverage loves to blow up.

Hedge funds are piling into AI data center stocks with borrowed money. That’s fine—until it’s not. If the capex cycle slows or valuations correct, the unwind could get messy. And let’s not forget: the S&P 500’s forward P/E ratio is already stretched, sitting at 21.4x—above its 10-year average.