|
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 |
|
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. |
| π Get Moonshot Alerts Instantly | Markets move in minutes. Your inbox can't keep up. | Join the Moonshot VIP Text Service and get real-time alerts, portfolio updates, and first-move intelligence directly to your phone. | Be first to know. Anywhere. | π [Join the Text List →] | |
|
|
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 |
The Facts At A Glance AI Power Crunch: What the Numbers Say United States ✔ Georgia Power is requesting ~10 GW of new generation, with regulators confirming ~80% of that demand is coming from data centers and AI-driven load. ✔ U.S. Government Accountability Office (GAO) projects data centers could consume 4.6% to 9.1% of all U.S. electricity by 2030. ✔ Florida approved a multi-year utility rate increase, with part of the pressure tied to capacity upgrades needed to serve large industrial loads, including data centers ✔ Several states have reported delays or pauses on new data-center projects due to grid strain and substation saturation. Europe ✔ Ireland paused new data-center approvals in Dublin after they consumed 30%+ of the region's electricity. ✔ The U.K. government warned that London's grid cannot support the current pace of data-center expansion, causing multiple projects to relocate. Asia ✔ Singapore halted new data-center approvals in 2022 due to grid pressure; reopened only with strict power-efficiency requirements. ✔ Japan is facing severe transformer shortages, delaying AI/cloud buildouts due to electrical infrastructure constraints. Middle East ✔ The UAE is building major new transmission corridors specifically to service AI and cloud-computing hubs due to "unprecedented demand growth." Global Scale ✔ A typical hyperscale facility requires 150–300 MW, equivalent to the electricity usage of a small city. ✔ Industry trackers estimate the world builds 500–700 new data centers per year, with ~200 classified as hyperscale. |
|
|
| π Get Moonshot Alerts Instantly | Markets move in minutes. Your inbox can't keep up. | Join the Moonshot VIP Text Service and get real-time alerts, portfolio updates, and first-move intelligence directly to your phone. | Be first to know. Anywhere. | π [Join the Text List →] | |
|
|
The Physical World Is Now the Limiting Factor ✔ China controls 98% of heavy rare-earth refining, the critical component for motors, turbines, magnets, drones, EVs, and data-center cooling systems. ✔ Global copper demand from data-centers and AI workloads is projected to double by 2027, according to multiple industrial-mineral forecasts (ICSG, IEA). ✔ Transformer shortages in the U.S. and Japan now stretch 18–60 months, delaying AI, cloud, and utility upgrades. ✔ AI data centers use 10× more electricity per square foot than traditional enterprise facilities. ✔ Transmission lines typically take 7–10 years from permitting to completion, meaning supply-side solutions lag demand by a full decade. The Hidden Cost Behind NVIDIA's $57B Quarter ✔ For every $1 spent on GPUs, hyperscale operators often invest $8–$12 in supporting infrastructure (power, cooling, buildings, transformers, substations). ✔ Cooling systems can consume 30–40% of a data center's total power draw. ✔ Several new hyperscale projects exceed $2–$4 billion in buildout cost before GPUs are installed. ✔ Electricity demand for AI workloads is doubling every 18–24 months, based on a combination of IEA projections and hyperscaler CapEx guidance. ✔ U.S. utilities now classify AI-driven data-centers as "industrial-grade loads" requiring specialized planning. ✔ U.S. will require an estimated 200–300 GW of new generation capacity to support projected AI/data-center demand by 2030–2035. ✔ The world currently adds 70–90 GW of new clean-energy capacity per year, far below what AI-driven demand curves require. ✔ Multiple U.S. regions report multi-year delays for new data-center hookups due to grid constraints. ✔Analysts warn AI could become one of the largest industrial electricity consumers globally by the early 2030s. |
|
|
π Premium Content Begins Here π |
|
In today's Premium Section, you'll find two new updates on plays we're putting our money in during this next, and explosive, stage of the commodity and rare-earth supercycle. | I hope you've been paying attention because many of our picks are currently beating the S&P by up to 3-to-1 this year. | Most financial newsletters charge $500, $1,000, even $5,000 per year. Why? Because they know they can. | I don't. | I built my wealth the old-fashioned way, not by selling subscriptions. | That's why I priced this at $25/month, or $250/year. | Not because it's low quality, but because I don't need to charge the typical prices other newsletters charge. | One good trade, idea, or concept could pay for your next decade of subscriptions. | The question isn't 'Why is this so cheap?' The question is, 'Why would I charge more?' | π Upgrade to Premium Now | P.S. If this newsletter were $1,000 per year, you'd have to think about it. | You'd weigh your options. You'd analyze the risk. | But it's $25 a month. | That's the price of a bad lunch decision. | And remember, just one good idea could pay for your subscription for a decade. | π Upgrade to Premium Now | |
|
|
No comments:
Post a Comment