AnthonieMeijer
@anthoniemeijer.bsky.social
Writing on AI & investing | VC investor | Ex-strategy consultant
https://anthoniemeijer.substack.com/
https://anthoniemeijer.substack.com/
Institutions aren't buying narratives.
- They're buying fee revenue.
- They're buying value accrual.
- They're buying cash flow equivalents.
2026 is the "Dawn of the Institutional Era."
Fundamentals finally matter.
- They're buying fee revenue.
- They're buying value accrual.
- They're buying cash flow equivalents.
2026 is the "Dawn of the Institutional Era."
Fundamentals finally matter.
January 20, 2026 at 5:18 PM
Institutions aren't buying narratives.
- They're buying fee revenue.
- They're buying value accrual.
- They're buying cash flow equivalents.
2026 is the "Dawn of the Institutional Era."
Fundamentals finally matter.
- They're buying fee revenue.
- They're buying value accrual.
- They're buying cash flow equivalents.
2026 is the "Dawn of the Institutional Era."
Fundamentals finally matter.
The crypto valuation playbook, 2026 edition:
- Can you DCF it? (Does it generate revenue?)
- What's the fee quality? (Organic or subsidized?)
- Does value accrue to holders? (Buybacks, burns, distributions?)
Three questions. Most tokens fail all three.
- Can you DCF it? (Does it generate revenue?)
- What's the fee quality? (Organic or subsidized?)
- Does value accrue to holders? (Buybacks, burns, distributions?)
Three questions. Most tokens fail all three.
January 19, 2026 at 5:02 PM
The crypto valuation playbook, 2026 edition:
- Can you DCF it? (Does it generate revenue?)
- What's the fee quality? (Organic or subsidized?)
- Does value accrue to holders? (Buybacks, burns, distributions?)
Three questions. Most tokens fail all three.
- Can you DCF it? (Does it generate revenue?)
- What's the fee quality? (Organic or subsidized?)
- Does value accrue to holders? (Buybacks, burns, distributions?)
Three questions. Most tokens fail all three.
AI's real bottleneck isn't chips—it's power.
- Data center power demand: +165% by 2030
- Grid interconnection wait: 4-7 years
- Transformer lead times: 4+ years
You can buy all the GPUs you want. Good luck plugging them in.
🧵
- Data center power demand: +165% by 2030
- Grid interconnection wait: 4-7 years
- Transformer lead times: 4+ years
You can buy all the GPUs you want. Good luck plugging them in.
🧵
January 18, 2026 at 11:15 AM
AI's real bottleneck isn't chips—it's power.
- Data center power demand: +165% by 2030
- Grid interconnection wait: 4-7 years
- Transformer lead times: 4+ years
You can buy all the GPUs you want. Good luck plugging them in.
🧵
- Data center power demand: +165% by 2030
- Grid interconnection wait: 4-7 years
- Transformer lead times: 4+ years
You can buy all the GPUs you want. Good luck plugging them in.
🧵
New newsletter: Apophenia, the pattern-seeking trap that costs investors money.
Also: why Gen Z spending $2,600/yr on pets matters, a World Cup crypto play, and the forgotten Magnificent Seven stock.
substack.com/@anthoniemei...
Also: why Gen Z spending $2,600/yr on pets matters, a World Cup crypto play, and the forgotten Magnificent Seven stock.
substack.com/@anthoniemei...
Apophenia
The Pattern-Seeking Trap
substack.com
January 17, 2026 at 10:42 AM
New newsletter: Apophenia, the pattern-seeking trap that costs investors money.
Also: why Gen Z spending $2,600/yr on pets matters, a World Cup crypto play, and the forgotten Magnificent Seven stock.
substack.com/@anthoniemei...
Also: why Gen Z spending $2,600/yr on pets matters, a World Cup crypto play, and the forgotten Magnificent Seven stock.
substack.com/@anthoniemei...
Aim to balance being conservative and reactive.
All good investors have a Bayesian model in their head and update their forecasts as new information arrives but you also shouldn't update too aggressively.
All good investors have a Bayesian model in their head and update their forecasts as new information arrives but you also shouldn't update too aggressively.
January 17, 2026 at 8:25 AM
Aim to balance being conservative and reactive.
All good investors have a Bayesian model in their head and update their forecasts as new information arrives but you also shouldn't update too aggressively.
All good investors have a Bayesian model in their head and update their forecasts as new information arrives but you also shouldn't update too aggressively.
Quality investing works in crypto too (to a certain degree).
Most tokens fail basic fundamentals. But a few pass the same filter I use for stocks:
↓
Most tokens fail basic fundamentals. But a few pass the same filter I use for stocks:
↓
January 16, 2026 at 5:55 PM
Quality investing works in crypto too (to a certain degree).
Most tokens fail basic fundamentals. But a few pass the same filter I use for stocks:
↓
Most tokens fail basic fundamentals. But a few pass the same filter I use for stocks:
↓
On physical AI:
- Waymo: 96% fewer crashes than human drivers.
- Amazon robots: 78% faster order-to-shipping.
- China: more industrial robots than the US + rest of world combined.
The headlines focus on white collar disruption. Blue collar is next, just on a longer timeline.
- Waymo: 96% fewer crashes than human drivers.
- Amazon robots: 78% faster order-to-shipping.
- China: more industrial robots than the US + rest of world combined.
The headlines focus on white collar disruption. Blue collar is next, just on a longer timeline.
January 15, 2026 at 7:32 PM
On physical AI:
- Waymo: 96% fewer crashes than human drivers.
- Amazon robots: 78% faster order-to-shipping.
- China: more industrial robots than the US + rest of world combined.
The headlines focus on white collar disruption. Blue collar is next, just on a longer timeline.
- Waymo: 96% fewer crashes than human drivers.
- Amazon robots: 78% faster order-to-shipping.
- China: more industrial robots than the US + rest of world combined.
The headlines focus on white collar disruption. Blue collar is next, just on a longer timeline.
Fluence Energy is up 15% today (mentioned in the last newsletter)
January 15, 2026 at 7:22 PM
Fluence Energy is up 15% today (mentioned in the last newsletter)
On the AI agent hype
Andrej Karpathy: "Calling everything an agent creates unrealistic expectations."
Only 10% of orgs have scaled true agents. Meanwhile, redesigning workflows with AI-in-the-loop is unlocking real value today.
This is the decade of agents, not the year.
Andrej Karpathy: "Calling everything an agent creates unrealistic expectations."
Only 10% of orgs have scaled true agents. Meanwhile, redesigning workflows with AI-in-the-loop is unlocking real value today.
This is the decade of agents, not the year.
January 14, 2026 at 7:03 PM
On the AI agent hype
Andrej Karpathy: "Calling everything an agent creates unrealistic expectations."
Only 10% of orgs have scaled true agents. Meanwhile, redesigning workflows with AI-in-the-loop is unlocking real value today.
This is the decade of agents, not the year.
Andrej Karpathy: "Calling everything an agent creates unrealistic expectations."
Only 10% of orgs have scaled true agents. Meanwhile, redesigning workflows with AI-in-the-loop is unlocking real value today.
This is the decade of agents, not the year.
AI prompting is solved. The new bottleneck is context.
Models know the public internet but nothing about your Q1 goals or yesterday's email.
The platform wars are heating up: Google & Microsoft racing to embed AI everywhere.
Whoever holds your context holds your attention.
Models know the public internet but nothing about your Q1 goals or yesterday's email.
The platform wars are heating up: Google & Microsoft racing to embed AI everywhere.
Whoever holds your context holds your attention.
January 13, 2026 at 5:18 PM
AI prompting is solved. The new bottleneck is context.
Models know the public internet but nothing about your Q1 goals or yesterday's email.
The platform wars are heating up: Google & Microsoft racing to embed AI everywhere.
Whoever holds your context holds your attention.
Models know the public internet but nothing about your Q1 goals or yesterday's email.
The platform wars are heating up: Google & Microsoft racing to embed AI everywhere.
Whoever holds your context holds your attention.
AI agents still can't carry legal accountability, build trust, or own ethical responsibility.
These aren't technical gaps—they're governance constraints.
For regulated industries, that's a durable investment thesis.
These aren't technical gaps—they're governance constraints.
For regulated industries, that's a durable investment thesis.
January 12, 2026 at 5:04 PM
AI agents still can't carry legal accountability, build trust, or own ethical responsibility.
These aren't technical gaps—they're governance constraints.
For regulated industries, that's a durable investment thesis.
These aren't technical gaps—they're governance constraints.
For regulated industries, that's a durable investment thesis.
Moats are obvious in hindsight.
The skill is spotting them before everyone agrees.
Ask: What would it cost a competitor to replicate this? If the answer is "billions and a decade," you might have something.
The skill is spotting them before everyone agrees.
Ask: What would it cost a competitor to replicate this? If the answer is "billions and a decade," you might have something.
January 11, 2026 at 11:15 AM
Moats are obvious in hindsight.
The skill is spotting them before everyone agrees.
Ask: What would it cost a competitor to replicate this? If the answer is "billions and a decade," you might have something.
The skill is spotting them before everyone agrees.
Ask: What would it cost a competitor to replicate this? If the answer is "billions and a decade," you might have something.
Reposted by AnthonieMeijer
75 years of stock market data, one lesson:
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
January 10, 2026 at 1:25 PM
75 years of stock market data, one lesson:
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
75 years of stock market data, one lesson:
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
January 10, 2026 at 1:25 PM
75 years of stock market data, one lesson:
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
- Any single year: -37% to +52%
- Over 20-year periods: +6% to +18%
- Not a single negative 20-year outcome. Ever.
- $100k → $908k at 11.7% avg annual returns.
Time in the market > timing the market.
Source: JP Morgan
Stock up 50%? "Visionary CEO, massive TAM." Down 50%? "Execution issues, overvalued."
The company didn't change. The price did.
New post on why narrative follows price and what to do about it. Plus some weekly stock/crypto theses.
anthoniemeijer.substack.com/p/price-make...
The company didn't change. The price did.
New post on why narrative follows price and what to do about it. Plus some weekly stock/crypto theses.
anthoniemeijer.substack.com/p/price-make...
Price Makes Narrative
Not the other way around
anthoniemeijer.substack.com
January 10, 2026 at 10:41 AM
Stock up 50%? "Visionary CEO, massive TAM." Down 50%? "Execution issues, overvalued."
The company didn't change. The price did.
New post on why narrative follows price and what to do about it. Plus some weekly stock/crypto theses.
anthoniemeijer.substack.com/p/price-make...
The company didn't change. The price did.
New post on why narrative follows price and what to do about it. Plus some weekly stock/crypto theses.
anthoniemeijer.substack.com/p/price-make...
Visa doesn't sell anything. It collects a toll.
Every swipe. Every transaction. Forever.
The best businesses are invisible infrastructure you can't avoid.
Every swipe. Every transaction. Forever.
The best businesses are invisible infrastructure you can't avoid.
January 10, 2026 at 8:25 AM
Visa doesn't sell anything. It collects a toll.
Every swipe. Every transaction. Forever.
The best businesses are invisible infrastructure you can't avoid.
Every swipe. Every transaction. Forever.
The best businesses are invisible infrastructure you can't avoid.
"When will AI agents replace knowledge workers?"
Wrong question for investors.
Better: Which workflows can be restructured today, and who's actually doing it?
Execution > capability roadmaps.
Wrong question for investors.
Better: Which workflows can be restructured today, and who's actually doing it?
Execution > capability roadmaps.
January 9, 2026 at 7:55 PM
"When will AI agents replace knowledge workers?"
Wrong question for investors.
Better: Which workflows can be restructured today, and who's actually doing it?
Execution > capability roadmaps.
Wrong question for investors.
Better: Which workflows can be restructured today, and who's actually doing it?
Execution > capability roadmaps.
Quality companies at high valuations are still risky.
They don't need a crisis to fall. They just need 18% growth instead of 22%.
At elevated multiples, the bar for "disappointing" is low.
They don't need a crisis to fall. They just need 18% growth instead of 22%.
At elevated multiples, the bar for "disappointing" is low.
January 9, 2026 at 5:55 PM
Quality companies at high valuations are still risky.
They don't need a crisis to fall. They just need 18% growth instead of 22%.
At elevated multiples, the bar for "disappointing" is low.
They don't need a crisis to fall. They just need 18% growth instead of 22%.
At elevated multiples, the bar for "disappointing" is low.
Evaluating AI-native SaaS? Look at where they're deploying it.
Support triage, document validation → real efficiency gains.
Strategy, pricing, hiring → probably overpromising.
Support triage, document validation → real efficiency gains.
Strategy, pricing, hiring → probably overpromising.
January 8, 2026 at 8:32 PM
Evaluating AI-native SaaS? Look at where they're deploying it.
Support triage, document validation → real efficiency gains.
Strategy, pricing, hiring → probably overpromising.
Support triage, document validation → real efficiency gains.
Strategy, pricing, hiring → probably overpromising.
Reading your 15th article on whether NVIDIA is overvalued isn't research.
It's procrastination disguised as diligence.
Real research: What's your thesis? What would disprove it? What's your position size?
It's procrastination disguised as diligence.
Real research: What's your thesis? What would disprove it? What's your position size?
January 8, 2026 at 7:32 PM
Reading your 15th article on whether NVIDIA is overvalued isn't research.
It's procrastination disguised as diligence.
Real research: What's your thesis? What would disprove it? What's your position size?
It's procrastination disguised as diligence.
Real research: What's your thesis? What would disprove it? What's your position size?
The best portfolios, like the best notes, are written for your future self.
Atomic positions. Clear thesis. Compounding through time.
Maintenance, not reinvention.
Atomic positions. Clear thesis. Compounding through time.
Maintenance, not reinvention.
January 7, 2026 at 8:01 PM
The best portfolios, like the best notes, are written for your future self.
Atomic positions. Clear thesis. Compounding through time.
Maintenance, not reinvention.
Atomic positions. Clear thesis. Compounding through time.
Maintenance, not reinvention.
Everyone asks "when should I buy?"
Nobody asks "when will I sell?"
Write your exit rules before you need them. In the moment, you'll rationalize anything.
Nobody asks "when will I sell?"
Write your exit rules before you need them. In the moment, you'll rationalize anything.
January 7, 2026 at 7:02 PM
Everyone asks "when should I buy?"
Nobody asks "when will I sell?"
Write your exit rules before you need them. In the moment, you'll rationalize anything.
Nobody asks "when will I sell?"
Write your exit rules before you need them. In the moment, you'll rationalize anything.
Nike. Novo Nordisk. Charter. Fiserv.
All multi-baggers. All now 40-70% off highs.
The hard part isn't finding compounders — it's knowing when they've stopped.
All multi-baggers. All now 40-70% off highs.
The hard part isn't finding compounders — it's knowing when they've stopped.
January 6, 2026 at 5:18 PM
Nike. Novo Nordisk. Charter. Fiserv.
All multi-baggers. All now 40-70% off highs.
The hard part isn't finding compounders — it's knowing when they've stopped.
All multi-baggers. All now 40-70% off highs.
The hard part isn't finding compounders — it's knowing when they've stopped.
In many ways, humans are designed for social conformity
Conforming is easier, and it works 95% of the time
Hence the phrase, “Don’t reinvent the wheel".
However, original thinkers knows that while social and expert authority is often correct, it can also be incredibly wrong too
Conforming is easier, and it works 95% of the time
Hence the phrase, “Don’t reinvent the wheel".
However, original thinkers knows that while social and expert authority is often correct, it can also be incredibly wrong too
January 5, 2026 at 7:36 PM
In many ways, humans are designed for social conformity
Conforming is easier, and it works 95% of the time
Hence the phrase, “Don’t reinvent the wheel".
However, original thinkers knows that while social and expert authority is often correct, it can also be incredibly wrong too
Conforming is easier, and it works 95% of the time
Hence the phrase, “Don’t reinvent the wheel".
However, original thinkers knows that while social and expert authority is often correct, it can also be incredibly wrong too