Skip to content
a16za16z

Why AI Moats Still Matter (And How They've Changed)

a16z General Partners David Haber, Alex Rampell, and Erik Torenberg discuss why 19 out of 20 AI startups building the same thing will die - and why the survivor might charge $20,000 for what used to cost $20. They expose the "janitorial services paradox" (why the most boring software is most defensible), explain why OpenAI won't compete with your orthodontic clinic software despite having 800 million weekly users, and reveal how non-lawyers are building the most successful legal AI companies. Timestamps: 0:00 - Intro 1:12 - Do moats still matter? 2:42 - Data network effects only work at mega scale 5:01 - The ankle biter problem 5:48 - Are incumbents more or less defensible? 7:14 - Will companies vibe code their own Zendesk? 8:48 - Why you won't vibe code Microsoft 10:09 - The Goldilocks zone of pricing 11:21 - Greenfield strategy 13:32 - Which software gets cut first 16:22 - Steel man: Brand and velocity as moats 17:44 - "Context is King" 19:58 - Feature vs. product vs. company 21:47 - Will OpenAI build everything? 24:04 - Steve Jobs told Drew Houston Dropbox was a feature 27:05 - Platform risk: Will they compete or tax you? 30:06 - The "gold bricks" conversation with Dan Rose 33:38 - What OpenAI should prioritize 35:26 - Will AI consolidate to winner-take-most? 39:16 - Why Dropbox survived anyway 43:48 - The messy inbox wedge strategy 44:06 - Why AI is different: It's consensus 48:18 - Jobs won't disappear—$1 tasks will explode 49:30 - The Uber/taxi lesson for AI If you enjoyed this episode, be sure to like, subscribe, and share with your friends! Follow David on X: https://x.com/dhaber Follow Alex on X: https://x.com/arampell Follow Erik on X: https://x.com/eriktorenberg Follow a16z on X: https://x.com/a16z Follow a16z on LinkedIn:https://www.linkedin.com/company/a16z Listen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX Listen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711 Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details, please see http://a16z.com/disclosures.

David HaberhostAlex RampellhostErik Torenberghost
Dec 2, 202550mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

AI moats still matter as software shifts from IT to labor

  1. Moats still matter, but AI is more a differentiation accelerant than a defensibility source, with defensibility coming from workflow ownership, context, embedding, and systems of record.
  2. AI lowers the cost of creating software, increasing the number of competitors (“ankle biters”) and making early-stage differentiation harder until companies reach “mega scale” where data/scale effects become meaningful.
  3. Enterprise incumbents face pressure from seat-based pricing and DIY “vibe coding” narratives, yet real switching remains limited due to edge cases, comparative advantage, and deep operational entrenchment.
  4. Pricing and budget dynamics create a “Goldilocks zone” where products are important enough to buy but not important enough to trigger constant switching, while greenfield strategies depend on founder patience and new-company formation rates.
  5. Platform risk persists in AI (compete vs. tax), but the abundance of “gold bricks” (lucrative opportunities) and multiple model providers means OpenAI-scale players won’t build every vertical, leaving room for application companies to expand from feature to product to company.

IDEAS WORTH REMEMBERING

5 ideas

AI features differentiate; moats come from workflow control.

Capabilities like multilingual compliant voice agents can be impressive, but defensibility usually requires owning end-to-end workflows, becoming a system of record, and embedding into customer operations.

Data network effects are real only at “Earth/Sun/Jupiter” scale.

Seeing slightly more customers than a competitor rarely matters early; advantage becomes obvious only when one player has orders-of-magnitude more data/usage, improving outcomes in a measurable way.

AI increases supply of software, intensifying the “ankle biter” problem.

Lower build costs create many near-identical entrants, making it harder to reach the scale where moats emerge; momentum matters because it’s the path to gravitational scale, not a moat itself.

Seat-based pricing is structurally vulnerable in an AI labor-replacement world.

If AI reduces headcount needs (designers, support agents), per-seat revenue can shrink even if product value rises; vendors may need outcome/usage-linked pricing to capture value credibly.

DIY “vibe coding” will replace some tools, but not complex, edge-case-heavy systems.

Teams may prototype competitors to narrow tools, yet products like Word/Office endure because they encode countless edge cases and operational details that are costly to rediscover and maintain.

WORDS WORTH SAVING

5 quotes

The thing that is fundamentally different about this product cycle is that the software itself can actually do the work, and therefore the market opportunity for software today is no longer just IT spend, it's largely labor.

David Haber

I think moats matter just as much as they did before. The one change is that in the supply-demand equation, there's conceptually more supply of software on the cup because the barrier to creating this stuff has gone down dramatically.

Alex Rampell

The AI-ness of that capability, in my opinion, is not a source of defensibility. It, it's largely differentiation.

David Haber

It's almost like gravity. Gravity actually, like one atom actually has, exerts gravity on you, but you only really see it at like very, very large scale.

Alex Rampell

It's not like all the jobs will go away. I actually think that's not gonna happen at all. There are a lot of things where if I could hire somebody for a dollar to do this task, I would a hundred percent do that. I've never been able to hire somebody for a dollar. Now I can hire software for a dollar.

Alex Rampell

Differentiation vs. defensibility in AI productsData network effects and scale thresholds“Ankle biter” competition and zero-to-one challengesSeat-based pricing disruption and software spend rationalizationGoldilocks zone, switching inertia, and greenfield entryFeature vs. product vs. company; wedge strategiesPlatform risk: compete-or-tax; OpenAI strategy and consolidation

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome