Pinned Tweet
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
23
31
422
252,520
Patrick OShaughnessy retweeted
Cliff Sosin on the AI build-out:
"America's capital markets, and the capital markets in general, are just incredibly underappreciated.
We just redeployed 3% of GDP into this highly speculative venture, basically on the turn of a dime.
There's no one else in the world that’s been able to do it."
"America's capital markets, and the capital markets in general, are just incredibly underappreciated.
We just redeployed 3% of GDP into this highly speculative venture, basically on the turn of a dime.
There's no one else in the world that’s been able to do it."
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
0
2
18
5,326
Excited for this. Grew up loving LIFE.
LIFE returns to print 11.23.26.
Read our Editors’ Letter and learn more about the new LIFE here: new.life.com
3
0
28
7,834
Cliff Sosin on why Jevons paradox taps out and the market for intelligence could shrink:
"Jevons paradox is just a narrow example of price elasticity of demand. In general, all markets are U-shaped.
When the product is terrible, there's no demand. Then the product gets better ot, but they won't buy very much.
As the price goes down, the size of the market will rise. But then you get to a point where further price decreases don't drive enough demand and the market shrinks.
If you take the price to zero, the size of the market goes to zero.
One example that I've thought about is the cost of fiber optic communication, or just communication.
The cost of moving information used to be super high and people moved very little of it. Then the cost of moving information went down a lot and the amount of information moved went up a ton.
Then the market stopped growing and the cost kept falling and the market kept shrinking.
Cogent Communications, I think, routes 30 or 40% of all IP traffic across the internet. And Cogent Communications is a few billion dollar market cap company. It's just not a very big industry.
I'm not saying that we're going to get quite that small in terms of intelligence, but Jevons paradox is a statement about a particular spot on a curve."
"Jevons paradox is just a narrow example of price elasticity of demand. In general, all markets are U-shaped.
When the product is terrible, there's no demand. Then the product gets better ot, but they won't buy very much.
As the price goes down, the size of the market will rise. But then you get to a point where further price decreases don't drive enough demand and the market shrinks.
If you take the price to zero, the size of the market goes to zero.
One example that I've thought about is the cost of fiber optic communication, or just communication.
The cost of moving information used to be super high and people moved very little of it. Then the cost of moving information went down a lot and the amount of information moved went up a ton.
Then the market stopped growing and the cost kept falling and the market kept shrinking.
Cogent Communications, I think, routes 30 or 40% of all IP traffic across the internet. And Cogent Communications is a few billion dollar market cap company. It's just not a very big industry.
I'm not saying that we're going to get quite that small in terms of intelligence, but Jevons paradox is a statement about a particular spot on a curve."
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
27
10
129
23,883
Patrick OShaughnessy retweeted
The people of Silicon Valley are unwell.
If you work at a frontier lab or decacorn, you’ve got it bad. You haven’t seen your kids, your girlfriend’s annoyed, invites from friends to hang out are waning, even weddings are on hold. This is the time to make your mark and make your money, after all.
If you’re building a startup that’s trying to compete with or be one of these companies, you’ve got it bad. Maybe you passed up a job offer at one a couple years back, which means you left $100M on the table, you fucking idiot. Being an underdog isn't as fun as you thought.
If you’re working on a pre-AI startup, you’ve got it bad. Not too long ago you were the belle of the ball, but now your 30% year-on-year growth doesn’t feel like growth at all. It feels like death.
"Too many of my friends are spiraling," Brie Wolfson writes. "I don’t know how we’re supposed to live and work like this."
It’s all too much, too fast. Too much money, too much growth, too much power, too much change. Every month, a new frontier model is released. Every day, a new unicorn is minted. Seriously. 300 days into the year 2026, and 311 unicorns have been minted so far.
So what’s left for the sane ones? Brie went looking for teams playing a different game. She found one in Ottawa: a search infrastructure company that many of the AI darlings rely on.
It's growing fast, and its people are well.
Read @zebriez’s latest company dispatch from inside the sane asylum of turbopuffer.
colossus.com/article/sane-a…
If you work at a frontier lab or decacorn, you’ve got it bad. You haven’t seen your kids, your girlfriend’s annoyed, invites from friends to hang out are waning, even weddings are on hold. This is the time to make your mark and make your money, after all.
If you’re building a startup that’s trying to compete with or be one of these companies, you’ve got it bad. Maybe you passed up a job offer at one a couple years back, which means you left $100M on the table, you fucking idiot. Being an underdog isn't as fun as you thought.
If you’re working on a pre-AI startup, you’ve got it bad. Not too long ago you were the belle of the ball, but now your 30% year-on-year growth doesn’t feel like growth at all. It feels like death.
"Too many of my friends are spiraling," Brie Wolfson writes. "I don’t know how we’re supposed to live and work like this."
It’s all too much, too fast. Too much money, too much growth, too much power, too much change. Every month, a new frontier model is released. Every day, a new unicorn is minted. Seriously. 300 days into the year 2026, and 311 unicorns have been minted so far.
So what’s left for the sane ones? Brie went looking for teams playing a different game. She found one in Ottawa: a search infrastructure company that many of the AI darlings rely on.
It's growing fast, and its people are well.
Read @zebriez’s latest company dispatch from inside the sane asylum of turbopuffer.
colossus.com/article/sane-a…
24
33
551
116,466
Cliff Sosin on the limits of intelligence:
"Superintelligence arrived. You probably didn't notice, because it turned out to be kind of incremental.
I think we overrated intelligence, or misread what it is.
We hold a thin scatter of facts about the world, and intelligence or ills the space between them.
Where the space between the facts behaves well, this is close to godlike. Most of what matters doesn't behave like that.
The universe is mostly the emergent behavior of complex systems. Stir cream into coffee. Watch a storm build out of nothing but atmospheric temperature differences and water vapor.
You have to run the thing.
The takeoff story assumed the limit was thinking. It isn't. The limit is contact with reality.
A smarter reasoner fills the gaps between known facts in simple areas faster and better, but it doesn’t produce new facts.
An AI will hand you a genuinely clever design for a jet turbine blade. It might be far more likely to work than anything your engineers came up with.
It'll still probably fail, because that’s the base rate at the edge of what anyone knows. The only way to find out is to build the blade and try to break it.
That's the real limit on learning, and it doesn't care how smart you are. Coming up with ideas was never the hard part. The hard part is how fast reality answers them.
Most problems can't be worked out at a desk by thinking. Those stay bound by how fast we can build the thing and see if it holds, and that clock barely moves."
From his essay "Maybe Intelligence Ain't All That." He walks through it here.
"Superintelligence arrived. You probably didn't notice, because it turned out to be kind of incremental.
I think we overrated intelligence, or misread what it is.
We hold a thin scatter of facts about the world, and intelligence or ills the space between them.
Where the space between the facts behaves well, this is close to godlike. Most of what matters doesn't behave like that.
The universe is mostly the emergent behavior of complex systems. Stir cream into coffee. Watch a storm build out of nothing but atmospheric temperature differences and water vapor.
You have to run the thing.
The takeoff story assumed the limit was thinking. It isn't. The limit is contact with reality.
A smarter reasoner fills the gaps between known facts in simple areas faster and better, but it doesn’t produce new facts.
An AI will hand you a genuinely clever design for a jet turbine blade. It might be far more likely to work than anything your engineers came up with.
It'll still probably fail, because that’s the base rate at the edge of what anyone knows. The only way to find out is to build the blade and try to break it.
That's the real limit on learning, and it doesn't care how smart you are. Coming up with ideas was never the hard part. The hard part is how fast reality answers them.
Most problems can't be worked out at a desk by thinking. Those stay bound by how fast we can build the thing and see if it holds, and that clock barely moves."
From his essay "Maybe Intelligence Ain't All That." He walks through it here.
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
14
12
114
23,389
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
9
8
328
44,587
Patrick OShaughnessy retweeted
This was very good
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
4
8
92
19,592
Patrick OShaughnessy retweeted
I'm a simple man.
I see a new Cliff Sosin podcast.
I stop what I'm doing.
And I listen.
I see a new Cliff Sosin podcast.
I stop what I'm doing.
And I listen.
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
2
7
90
19,609
Patrick OShaughnessy retweeted
Cliff Sosin runs one of the most interesting portfolios in public markets.
Nearly $2B in six companies, with almost 80% in Carvana alone.
"If you read history, of course, all the people who did really great tended to be very concentrated."
Nearly $2B in six companies, with almost 80% in Carvana alone.
"If you read history, of course, all the people who did really great tended to be very concentrated."
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
10
10
94
38,226
Patrick OShaughnessy retweeted
golden age
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
4
8
139
20,340
Cliff Sosin on why you want to own the complements to cheap intelligence, like Meta does:
"I've found it helpful to assume everyone has really great intelligence and ask what's left.
If intelligence becomes cheap and ubiquitous, what you want to own are the complements to plements are data, attention, love, physical proximity, scale, regulatory capture.
Let's talk about Meta. They have $250 billion of revenue, and they spend $250 billion a year. The question you might ask is, what do they spend it on? A really reductionist view is that they spend it all on intelligence.
They're buying data centers. They're literally buying inference. They're paying really smart people to write algorithms. They're writing code. That's just intelligence.
If you think forward with the fact that the price of intelligence is falling by a factor of a hundred or something over the next five years, their costs kind of go away.
Maybe instead of taking their cost to zero with much better intelligence at much lower prices, maybe they can sell a few more ads.
All of this is downstream of what they possess, which is not destroyed by ubiquitous intelligence, which is data and attention.
I don't think that the availability of cheap and ubiquitous intelligence is gonna cause people not to be interested in watching funny 20-second videos."
"I've found it helpful to assume everyone has really great intelligence and ask what's left.
If intelligence becomes cheap and ubiquitous, what you want to own are the complements to plements are data, attention, love, physical proximity, scale, regulatory capture.
Let's talk about Meta. They have $250 billion of revenue, and they spend $250 billion a year. The question you might ask is, what do they spend it on? A really reductionist view is that they spend it all on intelligence.
They're buying data centers. They're literally buying inference. They're paying really smart people to write algorithms. They're writing code. That's just intelligence.
If you think forward with the fact that the price of intelligence is falling by a factor of a hundred or something over the next five years, their costs kind of go away.
Maybe instead of taking their cost to zero with much better intelligence at much lower prices, maybe they can sell a few more ads.
All of this is downstream of what they possess, which is not destroyed by ubiquitous intelligence, which is data and attention.
I don't think that the availability of cheap and ubiquitous intelligence is gonna cause people not to be interested in watching funny 20-second videos."
Cliff Sosin (@CliffordSosin) is one of my favorite people to call, ask some big questions, and then shut up and listen.
This conversation is a version of one of those calls.
We discuss:
- The limits of intelligence
- Owning the complements to it
- Meta, the frontier labs, and tal markets
- Why he's comfortable with 80% of his portfolio in Carvana.
TIMESTAMPS
0:00 Intro
10:41 The Limits of Intelligence
18:03 Investing Beyond Cheap Intelligence
24:03 Will AI Become Commoditized?
35:10 Who Profits From AI?
43:00 AI Agents Reshape Competition
52:06 America's Capital Markets
58:04 Transforming Companies With AI
68:19 AI and Portfolio Concentration
73:16 Education in the AI Era
22
7
120
58,646
@CliffordSosin Links
Youtube: youtube.com/watch?v=EH6DsN…
Spotify: open.spotify.com/episode/6wnOxt…
Apple: podcasts.apple.com/us/podcast/inv…
Youtube: youtube.com/watch?v=EH6DsN…
Spotify: open.spotify.com/episode/6wnOxt…
Apple: podcasts.apple.com/us/podcast/inv…
0
2
11
9,933
Patrick OShaughnessy retweeted
The best, most comprehensive take on AI I’ve read. After a year of confusion and fear about the future, I finished it feeling like a weight had physically lifted.
It’s long. Make the time.
@vgr is undefeated.
contraptions.venkateshrao.com/p/cognitive-me…
It’s long. Make the time.
@vgr is undefeated.
contraptions.venkateshrao.com/p/cognitive-me…
61
275
2,429
435,658
Good day to rewatch
Krishna Rao is the CFO of Anthropic, and this is his first podcast appearance.
He joined the company two years ago when run-rate revenue was about $250M. Today it is $30B. He has helped raise ~$75B and is responsible for the procurement and allocation of compute.
I feel lucky is like to sit inside a company this consequential at a moment this pivotal.
We discuss:
- The cone of uncertainty
- How he allocates compute across Trainium, TPUs, and GPUs
- What investors misunderstand about model companies
- Why the returns to frontier intelligence keep rising
- Platform vs application and where Anthropic builds its own products
- How Anthropic uses Claude internally
I have asked my closing question about the kindest thing more than 500 times. Krishna's answer is one I have never heard before.
Enjoy!
Timestamps:
0:00 Intro
2:38 The Compute Canvas
6:51 The "Cone of Uncertainty"
11:58 Why the Returns to Frontier Intelligence Are So High
16:45 Recursive Self-Improvement
20:20 Scaling Laws
23:30 Sourcing $100 Billion in Compute
28:05 Platform vs. Application Strategy
32:52 Pricing Dynamics
38:48 How Anthropic’s Finance Team Uses Claude
43:24 Raising Capital & Overcoming Investor Skepticism
52:32 Public Perception, Risks, and Government Regulation
57:25 Mythos Release
1:12:33 What Could Derail the AI Revolution?
1:13:47 Biotech and Healthcare
1:15:31 The Kindest Thing
4
9
57
15,216
A list of stats from the Anthropic S-1:
- NDR is ~900%
- Revenue per employee is over $6.6M, up from about $1.4M a year earlier
- Revenue went from $41M three years ago to $16.3B in the first six months of 2026
- Quarterly revenue went from $4.7B in Q1 to $11.6B in Q2
- Q2 2026 was the first profitable quarter: $974M in operating income ($1.7B operating loss the quarter before)
- Adjusted operating income was $1.6B in the first half of 2026, up from a $1.8B adjusted loss a year earlier
- Revenue grew 1,368% year over year in Q2, while compute and infrastructure costs grew 300%
- Compute went from 251% of revenue to 48% in five quarters
- The value of Anthropic’s own internal Claude usage grew 250x from January to September 2026
- Anthropic spends more on tokens (at list price) than on payroll
- Over 1,900 customers pay $1M+ a year, and about 250 pay $10M+
- Customers paying $100K+ went from 950 to over 11,900 in a year
- Over half of the Fortune 500 pays $100K+ a year
- Those Fortune 500 customers still spend less than 0.1% of their total expenses with Anthropic
- The top 5% of Claude Code users spend roughly 20x what non-users on Claude Enterprise do
- About 40 million smaller customers
- $72.4B of the $80.7B first-half net loss is an accounting charge on convertible notes (not operating losses)
- Nearly $800B in compute commitments over 10 years
$141.1B raised in total, with over $130B in cash and short-term investments
- Nearly half of revenue comes through Amazon, Google and Microsoft’s cloud marketplaces, all of which also compete with Claude
- A single share held by the seven co-founders carries 50.1% of the vote
- That share is redeemed for nothing if fewer than three founders remain
- NDR is ~900%
- Revenue per employee is over $6.6M, up from about $1.4M a year earlier
- Revenue went from $41M three years ago to $16.3B in the first six months of 2026
- Quarterly revenue went from $4.7B in Q1 to $11.6B in Q2
- Q2 2026 was the first profitable quarter: $974M in operating income ($1.7B operating loss the quarter before)
- Adjusted operating income was $1.6B in the first half of 2026, up from a $1.8B adjusted loss a year earlier
- Revenue grew 1,368% year over year in Q2, while compute and infrastructure costs grew 300%
- Compute went from 251% of revenue to 48% in five quarters
- The value of Anthropic’s own internal Claude usage grew 250x from January to September 2026
- Anthropic spends more on tokens (at list price) than on payroll
- Over 1,900 customers pay $1M+ a year, and about 250 pay $10M+
- Customers paying $100K+ went from 950 to over 11,900 in a year
- Over half of the Fortune 500 pays $100K+ a year
- Those Fortune 500 customers still spend less than 0.1% of their total expenses with Anthropic
- The top 5% of Claude Code users spend roughly 20x what non-users on Claude Enterprise do
- About 40 million smaller customers
- $72.4B of the $80.7B first-half net loss is an accounting charge on convertible notes (not operating losses)
- Nearly $800B in compute commitments over 10 years
$141.1B raised in total, with over $130B in cash and short-term investments
- Nearly half of revenue comes through Amazon, Google and Microsoft’s cloud marketplaces, all of which also compete with Claude
- A single share held by the seven co-founders carries 50.1% of the vote
- That share is redeemed for nothing if fewer than three founders remain
Krishna Rao is the CFO of Anthropic, and this is his first podcast appearance.
He joined the company two years ago when run-rate revenue was about $250M. Today it is $30B. He has helped raise ~$75B and is responsible for the procurement and allocation of compute.
I feel lucky is like to sit inside a company this consequential at a moment this pivotal.
We discuss:
- The cone of uncertainty
- How he allocates compute across Trainium, TPUs, and GPUs
- What investors misunderstand about model companies
- Why the returns to frontier intelligence keep rising
- Platform vs application and where Anthropic builds its own products
- How Anthropic uses Claude internally
I have asked my closing question about the kindest thing more than 500 times. Krishna's answer is one I have never heard before.
Enjoy!
Timestamps:
0:00 Intro
2:38 The Compute Canvas
6:51 The "Cone of Uncertainty"
11:58 Why the Returns to Frontier Intelligence Are So High
16:45 Recursive Self-Improvement
20:20 Scaling Laws
23:30 Sourcing $100 Billion in Compute
28:05 Platform vs. Application Strategy
32:52 Pricing Dynamics
38:48 How Anthropic’s Finance Team Uses Claude
43:24 Raising Capital & Overcoming Investor Skepticism
52:32 Public Perception, Risks, and Government Regulation
57:25 Mythos Release
1:12:33 What Could Derail the AI Revolution?
1:13:47 Biotech and Healthcare
1:15:31 The Kindest Thing
60
71
624
138,900
Patrick OShaughnessy retweeted
Krishna Rao is the CFO of Anthropic, and this is his first podcast appearance.
He joined the company two years ago when run-rate revenue was about $250M. Today it is $30B. He has helped raise ~$75B and is responsible for the procurement and allocation of compute.
I feel lucky is like to sit inside a company this consequential at a moment this pivotal.
We discuss:
- The cone of uncertainty
- How he allocates compute across Trainium, TPUs, and GPUs
- What investors misunderstand about model companies
- Why the returns to frontier intelligence keep rising
- Platform vs application and where Anthropic builds its own products
- How Anthropic uses Claude internally
I have asked my closing question about the kindest thing more than 500 times. Krishna's answer is one I have never heard before.
Enjoy!
Timestamps:
0:00 Intro
2:38 The Compute Canvas
6:51 The "Cone of Uncertainty"
11:58 Why the Returns to Frontier Intelligence Are So High
16:45 Recursive Self-Improvement
20:20 Scaling Laws
23:30 Sourcing $100 Billion in Compute
28:05 Platform vs. Application Strategy
32:52 Pricing Dynamics
38:48 How Anthropic’s Finance Team Uses Claude
43:24 Raising Capital & Overcoming Investor Skepticism
52:32 Public Perception, Risks, and Government Regulation
57:25 Mythos Release
1:12:33 What Could Derail the AI Revolution?
1:13:47 Biotech and Healthcare
1:15:31 The Kindest Thing
He joined the company two years ago when run-rate revenue was about $250M. Today it is $30B. He has helped raise ~$75B and is responsible for the procurement and allocation of compute.
I feel lucky is like to sit inside a company this consequential at a moment this pivotal.
We discuss:
- The cone of uncertainty
- How he allocates compute across Trainium, TPUs, and GPUs
- What investors misunderstand about model companies
- Why the returns to frontier intelligence keep rising
- Platform vs application and where Anthropic builds its own products
- How Anthropic uses Claude internally
I have asked my closing question about the kindest thing more than 500 times. Krishna's answer is one I have never heard before.
Enjoy!
Timestamps:
0:00 Intro
2:38 The Compute Canvas
6:51 The "Cone of Uncertainty"
11:58 Why the Returns to Frontier Intelligence Are So High
16:45 Recursive Self-Improvement
20:20 Scaling Laws
23:30 Sourcing $100 Billion in Compute
28:05 Platform vs. Application Strategy
32:52 Pricing Dynamics
38:48 How Anthropic’s Finance Team Uses Claude
43:24 Raising Capital & Overcoming Investor Skepticism
52:32 Public Perception, Risks, and Government Regulation
57:25 Mythos Release
1:12:33 What Could Derail the AI Revolution?
1:13:47 Biotech and Healthcare
1:15:31 The Kindest Thing
99
821
6,886
3,446,745
Patrick OShaughnessy retweeted
Today, we are breaking down @Affirm.
Affirm's stock fell more than 95% after its IPO in 2021. Today, it processes over $50 billion in GMV and is growing more than 30%, faster than when it was a fraction of its current size.
Its founder, Max Levchin (@mlevchin), grew up in Ukraine, moved to the US as a teenager, and co-founded PayPal in his early 20s.
While in college, he funded his first startups with credit cards and unknowingly destroyed his credit score. Years later, even after PayPal's IPO, he was denied a car loan because of his credit history.
That experience helped inspire Affirm's founding mission of building more honest financial products.
Fraser Christie, founder and CIO of Sandstone Global Investment, joined us to break down the business, its remarkable recovery, and why he believes Affirm can become the payment network of the internet.
Some interesting facts from our conversation:
1/ Affirm has roughly 50% market share in US buy now, pay later
2/ Max Levchin earns less than $40,000 in annual salary and owns billions in Affirm stock. He now spends much of his time working with a small engineering team on new products.
3/ Affirm can have negative customer acquisition costs.
4/ 70% of purchase volume comes from interest-bearing loans
5/ Affirm earns roughly 4% gross profit on purchase volume after funding costs and credit losses.
6/ Affirm underwrites every transaction individually
7/ Affirm Card already accounts for roughly 20% of purchase volume and is growing over 100% annually.
8/ Only 75 of the 250 largest US merchants have integrated Affirm directly into checkout.
9/ Headcount, G&A expenses, and stock-based compensation have remained essentially flat for several years despite significant growth.
10/ 95% of transactions come from existing customers. Average usage has grown from 2 transactions per year to 7 since its IPO.
11/ Roughly half of Affirm's loans are sold to third parties, with the other half remaining on its balance sheet.
12/ Affirm has no late fees, no compounding balances, and no deferred interest. Every loan is designed to be paid down to zero.
Enjoy!our conversation:
1/ Affirm has roughly 50% market share in US buy now, pay later and is gaining share against Klarna and Afterpay.
2/ 95% of transactions come from existing customers. Average usage has grown from 2 transactions per year to 7 since its IPO.
3/ Affirm can have negative customer acquisition costs. It makes money when a new customer signs up and completes their first purchase.
4/ 70% of purchase volume comes from interest-bearing loans, rather than the shorter-term pay-in-four products most people associate with BNPL.
5/ Affirm earns roughly 4% gross profit on purchase volume after funding costs and credit losses.
6/ Affirm underwrites every transaction individually, using AI to determine how much to lend, to whom, and at what rate.
7/ Affirm Card already accounts for roughly 20% of purchase volume and is growing over 100% annually.
8/ Only 75 of the 250 largest US merchants have integrated Affirm directly into checkout.
9/ Headcount, G&A expenses, and stock-based compensation have remained essentially flat for several years despite significant growth.
10/ Max Levchin earns less than $40,000 in annual salary and owns billions in Affirm stock. He now spends much of his time working with a small engineering team on new products.
11/ Roughly half of Affirm's loans are sold to third parties, with the other half remaining on its balance sheet.
12/ Affirm has no late fees, no compounding balances, and no deferred interest. Every loan is designed to be paid down to zero.
Enjoy!
$AFRM
open.spotify.com/episode/4Oozus…
Affirm's stock fell more than 95% after its IPO in 2021. Today, it processes over $50 billion in GMV and is growing more than 30%, faster than when it was a fraction of its current size.
Its founder, Max Levchin (@mlevchin), grew up in Ukraine, moved to the US as a teenager, and co-founded PayPal in his early 20s.
While in college, he funded his first startups with credit cards and unknowingly destroyed his credit score. Years later, even after PayPal's IPO, he was denied a car loan because of his credit history.
That experience helped inspire Affirm's founding mission of building more honest financial products.
Fraser Christie, founder and CIO of Sandstone Global Investment, joined us to break down the business, its remarkable recovery, and why he believes Affirm can become the payment network of the internet.
Some interesting facts from our conversation:
1/ Affirm has roughly 50% market share in US buy now, pay later
2/ Max Levchin earns less than $40,000 in annual salary and owns billions in Affirm stock. He now spends much of his time working with a small engineering team on new products.
3/ Affirm can have negative customer acquisition costs.
4/ 70% of purchase volume comes from interest-bearing loans
5/ Affirm earns roughly 4% gross profit on purchase volume after funding costs and credit losses.
6/ Affirm underwrites every transaction individually
7/ Affirm Card already accounts for roughly 20% of purchase volume and is growing over 100% annually.
8/ Only 75 of the 250 largest US merchants have integrated Affirm directly into checkout.
9/ Headcount, G&A expenses, and stock-based compensation have remained essentially flat for several years despite significant growth.
10/ 95% of transactions come from existing customers. Average usage has grown from 2 transactions per year to 7 since its IPO.
11/ Roughly half of Affirm's loans are sold to third parties, with the other half remaining on its balance sheet.
12/ Affirm has no late fees, no compounding balances, and no deferred interest. Every loan is designed to be paid down to zero.
Enjoy!our conversation:
1/ Affirm has roughly 50% market share in US buy now, pay later and is gaining share against Klarna and Afterpay.
2/ 95% of transactions come from existing customers. Average usage has grown from 2 transactions per year to 7 since its IPO.
3/ Affirm can have negative customer acquisition costs. It makes money when a new customer signs up and completes their first purchase.
4/ 70% of purchase volume comes from interest-bearing loans, rather than the shorter-term pay-in-four products most people associate with BNPL.
5/ Affirm earns roughly 4% gross profit on purchase volume after funding costs and credit losses.
6/ Affirm underwrites every transaction individually, using AI to determine how much to lend, to whom, and at what rate.
7/ Affirm Card already accounts for roughly 20% of purchase volume and is growing over 100% annually.
8/ Only 75 of the 250 largest US merchants have integrated Affirm directly into checkout.
9/ Headcount, G&A expenses, and stock-based compensation have remained essentially flat for several years despite significant growth.
10/ Max Levchin earns less than $40,000 in annual salary and owns billions in Affirm stock. He now spends much of his time working with a small engineering team on new products.
11/ Roughly half of Affirm's loans are sold to third parties, with the other half remaining on its balance sheet.
12/ Affirm has no late fees, no compounding balances, and no deferred interest. Every loan is designed to be paid down to zero.
Enjoy!
$AFRM
open.spotify.com/episode/4Oozus…
0
4
28
10,854
Christina Cacioppo on high growth v. high margins:
"I think in the fullness of time, an 80% growth business with 80% margins will actually be better than a 160% growth business with 30% margins.
One of these will return more cash in the fullness of time. But today the latter ltiple of the former.
And so there's a part where everyone's being irrational here. And then there's a very clear game-on-the-field piece.
If you need to be financed, you're looking at investors today, many of whom are AI maximalists, that are like, 'We'll invest in the labs and we'll invest in tractors in Iowa. Nothing in between.'"
"I think in the fullness of time, an 80% growth business with 80% margins will actually be better than a 160% growth business with 30% margins.
One of these will return more cash in the fullness of time. But today the latter ltiple of the former.
And so there's a part where everyone's being irrational here. And then there's a very clear game-on-the-field piece.
If you need to be financed, you're looking at investors today, many of whom are AI maximalists, that are like, 'We'll invest in the labs and we'll invest in tractors in Iowa. Nothing in between.'"
My second conversation with Christina Cacioppo (@christinacaci), Vanta's cofounder and CEO.
Charlie Songhurst says the best returns come from companies that are highly boring and highly complex, because fewer founders want to build them.
Vanta is one of those companies.
When ars ago, Christina told me about building the first version of Vanta in Excel.
Since then she has turned it into the category leader in security compliance. Now she is rebuilding it around AI.
We discuss:
- Why great software businesses got cheap
- Why Vanta eats the cost of AI instead of marking it up
- Why comp for top people is up 10X in five years
- What is over/under hyped in AI
- How AI attackers are changing security
- Vanta marketing + billboards
- Building trust in an era of low trust
Enjoy!
TIMESTAMPS
0:00 Intro
4:17 Surviving the SaaSpocalypse
9:40 The AI Pricing Trap
16:17 Rebuilding Software for AI
22:43 Silicon Valley's Talent Wars
31:57 What's Overhyped in AI
38:39 Security at Machine Speed
47:16 Reinventing Software
55:37 The Billboard War
60:38 Rebuilding Trust in Technology
20
10
62
28,506
