Is rapidly scaled AI revenue as durable and high-quality as traditional software revenue?

Sarah Tavel · May 6, 2024

Whether AI revenue is durable can be judged by two things: whether the value proposition is enduring from first principles, and what the early customer cohorts show

A product like DeepL that replaces hiring human translators with instant human-quality translation is obviously enduring from first principles, and cohort behaviour supplies the empirical evidence

20:27 20VC: Benchmark's Sarah Tavel on Are Foundation Models Commoditising | Why Frontier Models Will Be Closed Source | Why the Value is in the Application Layer | The Future of AI is "Selling the Work" Not the Tools

Saam Motamedi · Jul 15, 2024

A fundamental lens — market dynamics, real product-market fit, retention, defensibility — must override growth data, and a company going 0 to $20M in six months is not worth investing in if those fundamentals fail.

The venture industry has largely forgotten this fundamental lens over the last year, and Greylock is comfortable passing on explosive growth when the answers to those questions are bad.

9:32 20VC: Why We Are in a Bubble & Now is Frothier Than 2021 | Why $1M ARR is a BS Milestone for Series A | Why Seed Pricing is Rational & Large Seed Rounds Have Less Risk | Why Many AI Apps Have BS Revenue & Are Not Sustainable with Saam Motamedi @ Greylock

Martin Mignot · Aug 11, 2025

With AI companies, auditing the quality and durability of revenue is critical rather than taking the growth number at face value

You need to know whether the revenue is long-lasting and sticky; with enough cohorts the numbers show it, otherwise you must talk to customers and understand the use case

Scope: specific to AI companies

23:24 20VC: Figma, Scale, Wiz: Inside Index's Decacorn Factory | Decision-Making, Investment Process, Biggest Lessons, Biggest Misses | Why Gross Margin is a Fallacy at Seed | Never Turn Down a Deal on Price with Martin Mignot, Partner @ Index Ventures

Brendan Foody · Sep 15, 2025

Retention numbers plus customer anecdotes are the most important test of whether AI application-layer revenue is healthy and durable

Initial pilots and contracts are low-friction to access, so retention and customer love are the real signs of market fit; a company where 95% of pilots fail is probably a bad investment

Scope: applies to application-layer AI companies

36:01 20VC: Mercor: From $1M to $500M in 17 Months: The Fastest Growing Company in the World | How to Think About Margins and Revenue Sustainability in AI | Why Evaluation Benchmarks in AI are BS Today with Brendan Foody

David George · Dec 15, 2025

Fast-scaled AI revenue means as much as revenue always did, provided it comes with high retention and high engagement — fast growth does not imply transience or lower quality, but the bar for assessing it is now much higher

You can't observe years of renewal behavior in companies that scaled this fast, so you must lean on short retention cycles and especially engagement as leading indicators of value delivered

Scope: conditional on high retention and engagement

30:10 20VC: a16z's David George on How $BN Funds Can 5×, Do Margins & Revenue Matter in AI & the Most Controversial Bet at a16z

Jason Lemkin · May 27, 2024

It is fine not to know whether AI revenue growth comes from experimental or sustainable budgets — investing in things that are exploding without over-analyzing is what venture is

You make ~20 investments per fund, so over-analyzing individual AI explosions means missing the opportunity entirely

52:53 Dont overanalyze budget source just invest in explosive growth

Elias Torres · Mar 21, 2025

The hypergrowth revenue of AI tools like Cursor could disappear because customers can switch trivially between them, and this is a brand new phenomenon nobody understands yet.

Users can move from Cursor to Windsurf to Copilot with no friction, so the revenue isn't anchored to lifetime customers.

44:03 Low switching costs make hypergrowth ai revenue fragile and unpredictable

David Schneider · Sep 11, 2024

The source and quality of a company's customer base predicts whether it plateaus: revenue from other young technology companies deserves a discount, while enterprise and vertical customers like pharma and banking signal durable expansion

Small companies are more susceptible to market waves, especially the hot-and-cold cycle of venture funding, whereas proof of selling to large enterprises makes him believe they can sell to other similarly sized customers in future

38:19 Customer base composition young tech vs enterprise predicts revenue durability

Eric Vishria · Sep 25, 2024

Explosive early revenue growth in AI companies deserves almost no credit as evidence of company quality; it only proves that demand exists

Customers perceive these products as magic with tremendous ROI and want to experiment, so the pull is on the demand side; the real question is whether the company has sustainable advantage over time

25:32 Explosive early revenue growth proves demand not sustainable advantage