Blueprint

Where the moat is when anyone can build software

Sam Lessin's argument that software and models are commodities, what he says replaces them (data, trust, distribution), and where the other hosts and a few guests push back or speak for their own companies.

Type
Blueprint
Confidence
Low
Sources
21 videos
First published
Updated

Voice and stakes

Voice: one voice (Sam, ~62 % of attributed items). The thesis is Sam’s, repeated across about a dozen episodes between January and October 2026. Jess, Dave and Brit support or qualify parts of it, and a few guests add their own answers. (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 18:24)

Stakes. Slow Ventures, Sam’s firm, put a 3 million cheque into Airtable at a 90 million post around 2012 (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:56), and Polsia is a Slow portfolio company (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 15:03). Sam invested 100k in AssemblyAI’s first round (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 28:29) and promotes his own app Vox Americanus, a network of verified real people (Sam · Every CTO Is Quietly Switching (Jul 2026) ▶ 10:05). Dave’s own stakes are listed on Host stakes ledger. The guests who give moat answers (Ben Broca of Polsia and Henrik Werdelin of Audos) speak partly for their own companies.

The argument in brief

  1. Sam says software is not a business: you use it to build businesses, but it is not defensible in itself. (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 18:24)
  2. He says code is just content and tends towards zero, so nobody will build a company on software this easy to copy. (Sam · Fire Sam Altman, The End of So (Mar 2026) ▶ 16:24)
  3. The same goes for models: it is very hard to build a lasting business on a model with perfect replacements. (Sam · Silicon Valley Is Great at Tec (Oct 2026) ▶ 11:37)
  4. What to build instead is a compounding asset (data, trust, a marketplace, a “dark pool” of access), with software as a cheap or free on-ramp. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 14:32)
  5. With no software lock-in, he says distribution is what matters. (Sam · Stripe Buys OpenRouter, OpenAI (Aug 2026) ▶ 40:52)
  6. Guests agree there is no moat in the code, but offer speed, relationships or infrastructure and data instead. (Ben Broca · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 44:30)

Code is content

Sam’s January 2026 line is blunt.

(Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 18:24)

He adds that AI is wonderful to build software with and also a bad business, and that both can be true (May 2026). (Sam · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 27:51) Writing a lot of software used to be a defensible position and that is gone (Feb 2026). (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 30:26) He extends the idea to agent “harnesses”, which he calls copyable markdown files that nobody will make money on (Jun 2026). (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 30:11)

Jess backs the mood, if not the detail. She says software companies are scared, and that a durable business is something unique that people need plus a business model; copycats have always existed. (Jess · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 18:55)

Sam’s own example is Granola. He built his own replacement, which he uses to argue that with no implementation cost switching is instant and nobody earns real money. (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 28:29) He applies the same logic to Polsia, saying someone could clone what it figures out for fewer tokens, so margin tends to zero. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 17:12)

What the hosts say replaces it

Dark pools and trust. Sam says that when a market is open, competition collapses margins to zero; a good business needs proprietary access that nobody else has and that compounds. (Sam · Anthropic’s Bet on Coding Is W (Mar 2026) ▶ 36:16) He expects the world to become more aristocratic, with openness collapsing into private channels, and calls that “very medieval”. (Sam · Anthropic’s Bet on Coding Is W (Mar 2026) ▶ 37:52) He also says a company’s value is the customer relationship, brand and trust, and that once commoditised it is worthless. (Sam · Silicon Valley Is Great at Tec (Oct 2026) ▶ 32:23) Jess adds that relationships and in-person skills are how founders find those pools. (Jess · Anthropic’s Bet on Coding Is W (Mar 2026) ▶ 38:38) See Dark pools and the trust moat.

Data. Sam’s strongest version: if your data is not unique, you will not be valuable long term (Aug 2026). (Sam · Peter Deng & Rachel Holt (Aug 2026) ▶ 20:24) At the same panel the venture investor Josh Wolfe says enterprises will keep a proprietary data silo and run open models on it. (Josh Wolfe · Peter Deng & Rachel Holt (Aug 2026) ▶ 23:01) The hosts split on how much a dataset is worth. Dave says Cursor’s traces are a data set nobody else has, while Sam says selling a company for its data is selling it for scrap (Apr 2026). (Dave · What Tim Cook’s “Executive Cha (Apr 2026) ▶ 31:33) See Proprietary data moats.

Distribution. When software is cheap to build there is no lock-in, so distribution is what matters (Sam, Aug 2026). (Sam · Stripe Buys OpenRouter, OpenAI (Aug 2026) ▶ 40:52) The evidence here is largely assertion; see Distribution as the moat.

Speed and the guests’ answers

The cleanest guest statement comes from Ben Broca, CEO of Polsia, in March 2026.

(Ben Broca · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 44:30)

He says model companies have a head start of about three months by knowing the next model first. (Ben Broca · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 44:38) His co-guest Henrik Werdelin adds relationship capital as a second moat. (Henrik Werdelin · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 45:52) Both also say a human still sells to a human, and that Audos had to build a just-in-time interface because users need visual screens. (Henrik Werdelin · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 28:27)

In Oct 2026 Adrian, an investor and friend of the hosts, says AI is a commodity: once an idea appears everyone copies it and the value collapses. (Adrian · Adrian (Oct 2026) ▶ 51:09) He gives three places left to build real AI businesses: infrastructure, proprietary data, or secret algorithms. (Adrian · Adrian (Oct 2026) ▶ 51:28) He expects “winner take some” among agents, because the cost of setting a product up is high for users. (Adrian · Adrian (Oct 2026) ▶ 31:07) Sam’s own speed point is on the cost side: following is much cheaper than blazing the trail, so clonable businesses see margins collapse. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 17:59)

Models as a commodity

The same argument runs through the model labs (see Model commoditisation and switching). Sam says all frontier models are a commodity and he would switch from Claude Code tomorrow. (Sam · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 14:41) Dave says he switches models all the time and the labs look like infrastructure providers. (Dave · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 20:43) Dave also reports that DeepSeek has a model at roughly a hundredth of the cost of a leading rival, a figure he relays from OpenRouter data (Sep 2026). (Dave · Anthropic Wants to Slow Down. (Sep 2026) ▶ 06:19)

Is SaaS dying?

Evidence is thinner than the rhetoric (Where is the moat when anyone can build software? folds in the SaaS question). Sam says the most exposed companies are those selling a monthly set of capabilities behind easy-to-figure-out APIs. (Sam · The Seed Market Is Broken: Meg (Jan 2026) ▶ 35:32) Dave relays, secondhand and without naming the vendor, that a buyer’s CTO rebuilt a roughly 150,000 contract with Claude Code in about a week. (Dave · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 17:35) Guest Guillermo Rauch, whose company sells infrastructure for generated apps, says the seat-based model “definitely gets disrupted”. (Guillermo Rauch · Guillermo Rauch (Apr 2026) ▶ 10:45)

Sam adds that the SaaS era ran on a narrative of infinite lifetime value that collapsed. (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 31:17) Counter-views: Jess says Rippling can build strong AI on its own context, which is why SaaS is still alive. (Jess · Anthropic Wants to Slow Down. (Sep 2026) ▶ 26:25) Brit lists large raises by application-layer startups, and Sam answers that raising money means nothing. (Brit · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 31:40)

How the view moved

  • Dec 2025. Dave’s “AI is the new UI” prediction: people generate the interface for any task with AI coding tools. (Dave · Hype vs Reality: Our 2026 Pred (Dec 2025) ▶ 46:25)
  • Jan 2026. Sam’s headline: software is not a business. (Sam · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 18:24)
  • Feb 2026. The on-ramp framing: write the software, give it away, build the compounding asset. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 14:32)
  • Mar 2026. Dark pools and the “medieval” turn; the Audos and Polsia founders give the speed answer. (Sam · Anthropic’s Bet on Coding Is W (Mar 2026) ▶ 37:52) (Ben Broca · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 44:30)
  • Apr 2026. Sam says there is zero lock-in and he cannot see how anyone makes money (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 39:18); Brit predicts stickiness persists (Brit · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 30:22).
  • Aug 2026. Data becomes Sam’s lead answer at the investor panel. (Sam · Peter Deng & Rachel Holt (Aug 2026) ▶ 20:24)
  • Oct 2026. Adrian’s “winner take some” and three places to build; Sam repeats that models are a commodity with no lock-in. (Adrian · Adrian (Oct 2026) ▶ 51:28)

Where the hosts disagree

  • Does stickiness survive? Sam says nobody makes money when switching is free; Brit predicts many services keep their stickiness because normal users want good enough. (Brit · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 30:22)
  • Will models commoditise? Brit expects models to flip-flop for months rather than commoditise, “contrary to what Sam would say”. (Brit · Why the Met Gala Hates Tech, E (May 2026) ▶ 40:13) Jess reports that frontier models have been gaining share against open source. (Jess · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 25:45) Sam concedes premium positioning may be an exception. (Sam · Anthropic’s Bet on Coding Is W (Mar 2026) ▶ 22:53)
  • Is data worth much? Dave and Sam split on the Cursor traces, as above.
  • Does personal software help? Dave sees agency for millions; Sam doubts it makes people dramatically better off. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 32:00)
  • Do interfaces disappear? Hosts predict agents replace websites; the Audos guest says users still need visual interfaces. (Henrik Werdelin · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 28:27)

What could make this wrong

  • It is one host’s thesis. Most software-not-business items are Sam alone, so repetition is not independent confirmation.
  • Sam himself says we are not there yet on commodity software, only heading that way. (Sam · Apple, OpenAI & Why Nobody Mak (Feb 2026) ▶ 14:16)
  • The guests’ moats are partly marketing for their own businesses (Audos, Polsia). (Ben Broca · Henrik Werdelin & Ben Broca (Mar 2026) ▶ 44:30)
  • Key cases are secondhand or anecdotal: the 150,000 contract story is relayed and the vendor is unnamed. (Dave · SaaS Companies Beware: AI Is T (Jan 2026) ▶ 17:35)

How strong is the evidence

Weak. The topic runs through 40 or so episodes, but Sam supplies most attributed items, and the independent outside voices are few: Josh Wolfe, Guillermo Rauch, Adrian, Ben Broca and Henrik Werdelin, mostly agreeing and mostly speaking about their own business. The speed moat rests on one guest pair in one episode. The three places to build, and “winner take some”, rest on Adrian in one episode. The Cursor data-value split is one episode. Brit and Jess are the only hosts who push back on commoditisation, each in a few remarks.

Go deeper

Sources

  1. (01)

    Why the Met Gala Hates Tech, Elon vs OpenAI Drama, and the Rise of Chinese AI Models

    YouTube

    40:13 into “Why the Met Gala Hates Tech, Elon vs OpenAI Drama, and the Rise of Chinese AI Models”

  2. (02)

    SaaS Companies Beware: AI Is The New UI (Anthropic's Claude Code and Cowork)

    YouTube

    17:35 into “SaaS Companies Beware: AI Is The New UI (Anthropic's Claude Code and Cowork)”

  3. (03)

    The Seed Market Is Broken: Mega-Rounds, Narrative Games, and Walkouts

    YouTube

    35:32 into “The Seed Market Is Broken: Mega-Rounds, Narrative Games, and Walkouts”

  4. (04)

    Elon Musk vs OpenAI Trial, Google Cloud Surge, Meta’s Blocked Acquisition, Anthropic Winning

    YouTube

    14:41 into “Elon Musk vs OpenAI Trial, Google Cloud Surge, Meta’s Blocked Acquisition, Anthropic Winning”

  5. (05)

    Apple, OpenAI & Why Nobody Makes Money in AI | The SaaS Apocalypse

    YouTube

    14:16 into “Apple, OpenAI & Why Nobody Makes Money in AI | The SaaS Apocalypse”

  6. (06)

    OpenAI's $852 Billion Valuation Makes Zero Sense (ClawCon Tokyo, Anthropic's Leak, Goodbye Internet)

    YouTube

    28:29 into “OpenAI's $852 Billion Valuation Makes Zero Sense (ClawCon Tokyo, Anthropic's Leak, Goodbye Internet)”

  7. (07)

    Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative

    YouTube

    16:24 into “Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative”

  8. (08)

    Stripe Buys OpenRouter, OpenAI vs. Anthropic & the AI Backlash

    YouTube

    40:52 into “Stripe Buys OpenRouter, OpenAI vs. Anthropic & the AI Backlash”

  9. (09)

    The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition

    YouTube

    20:24 into “The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition”

  10. (10)

    SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.

    YouTube

    25:45 into “SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.”

  11. (11)

    Jeff Dean Leaves Google, Airtable Sells, and the Bag-Securing Era

    YouTube

    19:56 into “Jeff Dean Leaves Google, Airtable Sells, and the Bag-Securing Era”

  12. (12)

    What Tim Cook’s “Executive Chairman” Move Really Means | Musk vs OpenAI, xAI and Cursor Rumors

    YouTube

    31:33 into “What Tim Cook’s “Executive Chairman” Move Really Means | Musk vs OpenAI, xAI and Cursor Rumors”

  13. (13)

    Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)

    YouTube

    20:43 into “Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)”

  14. (14)

    Hype vs Reality: Our 2026 Predictions for Tech, Media, and Culture

    YouTube

    46:25 into “Hype vs Reality: Our 2026 Predictions for Tech, Media, and Culture”

  15. (15)

    Vercel CEO: 70% of Our Traffic Is Now AI Agents "Nobody Was Prepared" | Anthropic, OpenClaw, OpenAI

    YouTube

    10:45 into “Vercel CEO: 70% of Our Traffic Is Now AI Agents "Nobody Was Prepared" | Anthropic, OpenClaw, OpenAI”

  16. (16)

    Meta Muse vs Instinct vs OpenAI Dots: Do AI Agents Have Any Moat?

    YouTube

    31:07 into “Meta Muse vs Instinct vs OpenAI Dots: Do AI Agents Have Any Moat?”

  17. (17)

    Anthropic Wants to Slow Down. Is It Because They're Losing?

    YouTube

    6:19 into “Anthropic Wants to Slow Down. Is It Because They're Losing?”

  18. (18)

    Anthropic’s Bet on Coding Is Working (OpenAI Shopping Pivot, A16Z’s Top 50 List, $1B Tennis Channel)

    YouTube

    22:53 into “Anthropic’s Bet on Coding Is Working (OpenAI Shopping Pivot, A16Z’s Top 50 List, $1B Tennis Channel)”

  19. (19)

    Silicon Valley Is Great at Tech and Terrible at Business

    YouTube

    11:37 into “Silicon Valley Is Great at Tech and Terrible at Business”

  20. (20)

    The New Startup Stack: One Founder + Agents | Henrik Werdelin (Audos) and Ben Broca (Polsia)

    YouTube

    28:27 into “The New Startup Stack: One Founder + Agents | Henrik Werdelin (Audos) and Ben Broca (Polsia)”

  21. (21)

    Every CTO Is Quietly Switching Off Anthropic and OpenAI | GLM 5.2, Vox Americanus, Claude Fable 5

    YouTube

    10:05 into “Every CTO Is Quietly Switching Off Anthropic and OpenAI | GLM 5.2, Vox Americanus, Claude Fable 5”

(By email)

Compounding Insights