When NVIDIA Broke Records... and Still Closed Down

Premium Members, check the bottom of this email. I’m releasing an update on two portfolio companies that have moved sharply in recent weeks.

If you’re not Premium yet, I can’t discuss the tickers here, but this is one of the year's most important updates.

The Part of NVIDIA’s Earnings Nobody Talked About

NVIDIA reported one of the most staggering financial quarters in corporate history:

$57 billion in quarterly revenue, a company record.

$51.2 billion from data centers alone, also a company record.

Guidance of $65 billion for the next quarter.

Growth on this level bends the curve of what companies are expected to do in a single quarter.

But here’s the part no one is talking about:

AI is outgrowing the physical world that has to power it.

Everyone saw the revenue numbers. No one asked the real question:

Where is the electricity for this going to come from?

Because AI isn’t running into software limits. It’s running into infrastructure limits.

On the outside, data centers look like big buildings, but they’re not.

From an energy use perspective, they’re more like cities.

A single hyperscale data center can consume as much electricity as a small city, often comparable to that of 100,000 households.

The world is building 500–700 data centers each year, of which roughly 200 are full-scale hyperscale facilities.

That means we’re effectively adding a new city every 12–18 hours, and plugging it directly into a grid that hasn’t been upgraded meaningfully in decades.

And now, governments and utilities around the world are being forced to admit it.

The Global Power Crunch Emerging From AI

Below are recent global examples showing the stress AI is putting on electricity systems.

United States

• Florida approves one of the largest utility rate hikes in state history. Florida Power & Light received approval for a $6.9 billion rate increase, with bills rising for millions of households starting January 2026. A portion of the pressure is tied to grid upgrades required to handle data-center load.

• Georgia Power requests 10 gigawatts of new capacity, and regulators admit 80% of it is being driven by AI datacenter demand, enough power for 8.3 million homes.

• Virginia and North Carolina report delays in data‑center approvals due to power availability strains.

Europe

• Ireland effectively paused all new data‑center approvals in Dublin because they were consuming 30%+ of regional electricity.

• The UK warned that London’s grid cannot support the current pace of data‑center expansion, forcing several projects to relocate.

Asia

• Singapore temporarily banned new data centers in 2022 because they were overwhelming the island’s power capacity. Only recently did they reopen with strict sustainability quotas.

• Japan is facing a transformer shortage, delaying AI and cloud buildouts due to the electrical infrastructure constraints.

Middle East

• The UAE is building entirely new power corridors to serve AI and cloud computing hubs, citing “unprecedented demand growth.”

And all this ties directly back to what the World Economic Forum said just last week:

“Investors are increasingly pricing in geopolitical risk and long-term supply security.”

This is part of the reason why NVIDIA, despite delivering a historic quarter, closed down.

Investors don’t fear the demand. They fear the constraints beneath it.

AI’s Breakpoint Isn’t Software. It’s the Grid.

We’re entering a new phase of the AI boom:

Phase 1: GPUs

Phase 2: Cloud infrastructure

Phase 3: Power, minerals, metals, and transmission

AI models will keep evolving. GPUs will keep getting faster. But the constraint will not be compute. It will be the physical world required to power, cool, and support that compute.

This is where the opportunity lives. Not in the shiny front‑end. But in the bottlenecks beneath the surface.

The minerals. The metals. The substations. The transmission lines. The rare earth magnets. The copper coils inside every motor. The grid capacity nobody thinks about.

AI runs on electricity. Electricity runs through infrastructure. Infrastructure runs on materials. Materials run on supply chains that the West does not control.

That is the story. And NVIDIA just confirmed it.

Where This Goes Next

Everything we’ve talked about over the past few months, the rare‑earth choke points, the copper supercycle, the grid constraints, is now front and center.

AI demand is real, it’s here, and it’s a force. But forces collide with physical limits. And when that happens, the winners are the companies positioned at the bottlenecks.

In the update below, we’re diving deeper into that in the Premium section.

Because that’s where two of the companies in our portfolio sit: directly in the path of the physical buildout AI can’t avoid.

If you’re Premium, scroll down and read the full update. If you’re not, consider this your invitation.

The market just showed you where the next decade of opportunity is going.

Position yourself accordingly.

Double D

🔓 Premium Content Begins Here 🔒