McKinsey says companies will spend $7 trillion on AI data centers...
Felix Prehn 🐶@felixprehn
21 views
Oct 28, 2025
~3 min read
Advertisement
1
McKinsey says companies will spend $7 trillion on AI data centers by 2030.
Most traders don’t realize only 20% of this money goes to computer chips everyone talks about.
The other 80% flows to companies you've never heard of.
Here's where the real money is going:
Most traders don’t realize only 20% of this money goes to computer chips everyone talks about.
The other 80% flows to companies you've never heard of.
Here's where the real money is going:
0:54
2
History repeats in every tech boom:
- Gold Rush: miners went broke, shovel sellers got rich
- Dot-com: websites failed, Cisco rose 3,800%
- Cloud: software struggled, data centers profited
- AI: same pattern happening right now
- Gold Rush: miners went broke, shovel sellers got rich
- Dot-com: websites failed, Cisco rose 3,800%
- Cloud: software struggled, data centers profited
- AI: same pattern happening right now
0:36
3
The AI world has 3 main tier where money flows: Manufacturing Companies, System Builders, and Power Providers.
Most people only focus on layer 1, and usually just the chip designers.
The real opportunities lie in the complete infrastructure stack.
Most people only focus on layer 1, and usually just the chip designers.
The real opportunities lie in the complete infrastructure stack.
4
Tier 1: Manufacturing Companies:
- Nvidia designs chips, TSMC makes them (sales up 34%)
- ASML is the only company making advanced chip machines
- Applied Materials & KLA Corp create quality control systems
These trade at 23x earnings vs Nvidia's 80x
- Nvidia designs chips, TSMC makes them (sales up 34%)
- ASML is the only company making advanced chip machines
- Applied Materials & KLA Corp create quality control systems
These trade at 23x earnings vs Nvidia's 80x
0:55
5
Applied Materials & KLA Corp create quality control systems for chip manufacturing.
These companies sell for 23 times earnings compared to Nvidia's 80. ASML's machines cost $400M each and would take 15-20 years to copy.
They face zero competition.
These companies sell for 23 times earnings compared to Nvidia's 80. ASML's machines cost $400M each and would take 15-20 years to copy.
They face zero competition.
6
The opportunity in manufacturing companies is huge.
While everyone chases popular AI stocks, these firms control critical equipment. No chips exist without them.
This is where institutional investors quietly build positions while retail investors chase headlines.
While everyone chases popular AI stocks, these firms control critical equipment. No chips exist without them.
This is where institutional investors quietly build positions while retail investors chase headlines.
7
Interested in understanding these manufacturing companies?
I've recorded a 12-minute video covering the chip supply chain.
RT this and comment "MANUFACUTURING" and I'll DM it to you.
Learn what Wall Street doesn't want retail investors to know about AI infrastructure.
I've recorded a 12-minute video covering the chip supply chain.
RT this and comment "MANUFACUTURING" and I'll DM it to you.
Learn what Wall Street doesn't want retail investors to know about AI infrastructure.
8
These companies profit regardless of which AI firm wins. TSMC makes chips for both Nvidia AND AMD.
ASML sells to every major chip maker.
This protects your investment from betting on the wrong company. It's like selling pickaxes during a gold rush - safer than mining.
ASML sells to every major chip maker.
This protects your investment from betting on the wrong company. It's like selling pickaxes during a gold rush - safer than mining.
9
Tier 2: System Builders:
- Super Micro: AI servers, 23% market share, advanced cooling
- Broadcom: custom AI chips, 220% growth, supplies Google/Meta
- Marvell Technology: custom chips, 78% revenue growth
- Dell & HP: also building server racks for AI infrastructure
- Super Micro: AI servers, 23% market share, advanced cooling
- Broadcom: custom AI chips, 220% growth, supplies Google/Meta
- Marvell Technology: custom chips, 78% revenue growth
- Dell & HP: also building server racks for AI infrastructure
1:13
10
Tier 3: Power Providers - the opportunity most miss.
Data centers already use 4% of US electricity, growing to 12% in 5 years. Eaton Corp's data center sales up 45%. G Venova has $600M in orders.
Siemens and ABB build essential power infrastructure.
Data centers already use 4% of US electricity, growing to 12% in 5 years. Eaton Corp's data center sales up 45%. G Venova has $600M in orders.
Siemens and ABB build essential power infrastructure.
1:04
11
Newer AI chips need even MORE power. Problems create profit opportunities.
The biggest AI challenge isn't software - it's heat, power, and infrastructure.
While everyone focuses on the same few AI stocks, smart money quietly invests in companies solving these problems.
The biggest AI challenge isn't software - it's heat, power, and infrastructure.
While everyone focuses on the same few AI stocks, smart money quietly invests in companies solving these problems.
12
Your AI investment plan:
- Check your current AI exposure
- Decide allocation across all three layers
- Buy gradually, not all at once
- Review quarterly
- Learn when to take profits
- Check your current AI exposure
- Decide allocation across all three layers
- Buy gradually, not all at once
- Review quarterly
- Learn when to take profits
13
The most important investing skill? Knowing WHEN TO SELL.
Never let a good profit disappear.
I'm teaching my exact selling rules in a free training at felixfriends.org/x
Which of these AI infrastructure companies interests you most?
Let me know below.
Never let a good profit disappear.
I'm teaching my exact selling rules in a free training at felixfriends.org/x
Which of these AI infrastructure companies interests you most?
Let me know below.