Aug 22, 2025 · 23m · a16z
The State of AI: Growth, Fragmentation, and the Next Wave
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
At the a16z LP Summit 2025, General Partners Martin Casado and Sarah Wang present data-driven insights on the state of AI, detailing exponential revenue growth across foundation models, the rapid expansion and defensibility of AI-native applications, and key capital strategies for navigating market fragmentation and competitive risks.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The host holds 8.7% of the talking time here. How this is scored →
The host as informed peer 2.5 Guest teaching 3.8 Guest disagreement 1.0 The host pushing back 0.1
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 7:10 Dismissing GPT Wrapper Concept
Martin forcefully rejects the prevailing derogatory narrative around 'GPT wrappers', arguing that calling apps wrappers makes as little sense as calling cloud software a 'cloud wrapper'.
Hardest push from the host ▶ 5:07 Challenging Foundation Model DominanceThe host explicitly challenges the premise that foundation models will capture all market value, asking why they haven't won everything.
Biggest teaching moment ▶ 3:58 OpenAI Domain Loss BreakdownMartin educates the host and audience on AI market fragmentation by citing how OpenAI repeatedly lost early category leads in code, images, and video to specialized players.
The host holds their own ▶ 10:22 Framing AI Defensibility MechanicsThe host demonstrates strong familiarity with venture and AI strategy by prompting the guest with specific technical moat dimensions including state, context, and brand.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Event Title Sequence and Legal Disclosures | 1 | 2 | 0 | 0 | The host opens with a simple high-level prompt asking for the state of play in AI. Sarah and Martin lay out the overview of their internal firm findings, framing AI as distinct subspaces rather than a monolith. | |
| Exponential Revenue Ramps of Foundation Models | 2 | 4 | 1 | 0 | The host asks clear follow-up prompts regarding market scale and cross-stack opportunities. Martin educates on how OpenAI lost early leads in specific domains like code, image generation, and video. | |
| Application Explosion and Re-evaluating GPT Wrappers | 3 | 5 | 3 | 1 | The host raises the common industry debate around GPT wrappers and adds a witty meta-joke about VC funds. Martin forcefully rejects the derogatory premise of the term 'GPT wrapper', comparing it to calling software a cloud wrapper. | |
| AI-Native Growth and the Innovator's Dilemma | 2 | 4 | 0 | 0 | The host prompts a comparison between AI-native companies and legacy SaaS. Sarah breaks down Stripe ARR metrics and explains how innovator's dilemma impacts incumbent SaaS companies. | |
| Evaluating Defensibility and Moats in AI | 4 | 5 | 1 | 0 | The host frames the defensibility discussion with informed categories such as state, context, and brand. Martin details how AI solves the initial bootstrap problem but fails to solve retention without traditional software moats. | |
| Enterprise ROI Case Studies: Cursor and Decagon | 2 | 3 | 0 | 0 | The host prompts a double-click on Cursor as a specific case study. Sarah highlights empirical data from portfolio CTOs showing dramatic productivity gains year-over-year. | |
| Prosumer Dynamics and Enterprise Pipeline Maturation | 3 | 4 | 1 | 0 | The host guides the conversation across prosumer dynamics, company wipeouts, and China's market impact. Guests reframe prosumer adoption as a natural tech cycle and caution against founder 'researcher-itis'. | |
| Foundation Model Thesis and Portfolio Strategy | 3 | 3 | 2 | 0 | The host prompts the guests for their investment thesis and 'spicy takes' regarding competitor positioning. Martin offers mild criticism of rival VC firms becoming conflicted out or sitting on the sidelines. |
Statements from this episode (26)
Wang: SaaS incumbents face classic innovator's dilemma in AI race
“Every SaaS company under the sun has launched an AI product. They're not just sitting on their hands, and you'd think that they'd have a huge advantage given distribution, but we're just seeing classic innovators dilemma.”
Casado: Dismissing AI apps as 'GPT wrappers' is a flawed critique
“GPT wrapper was this, like, derogatory term. Like, we've kind of come to the conclusion, like, that's not even a thing. Like, when someone writes software on, like, whatever the cloud, you don't call it a cloud wrapper.”
Casado: Aggressive early AI investments create conflicts blocking potential winners
“Conflicts really matter in this space, and so if you're too aggressive early and you don't really think through things, it can really keep you from investing in the one that's winning.”
Wang: AI companies are growing faster than a16z expected
“AI companies are growing faster and are larger than even we expected.”
Casado: There is no single AI market, only distinct subspaces
“We've kind of come to the opinion that there is no AI. There's like a bunch of subspaces that are totally different that all require their own strategy.”
Casado: AI is as big as software and requires varied strategies
“This is as big as software and the strategies need to vary as much.”
Wang: Top AI labs outpace early SaaS and hyperscaler revenue ramps
“And if you look at the revenue of just two of the top tier frontier labs You know, if you look at the chart on the left, not only have they surpassed the early revenue ramps of some of the best SaaS companies in history I'll point you over to the right they're…”
Wang: AI market is fragmenting across multiple players, not centralizing
“It's not the case that just two companies are growing very quickly in AI, and in fact that's probably a good segue to the next slide that shows markets are not only growing faster and much larger than expected, they're also fragmenting.”
Casado: OpenAI lost early leads in code, image, and video to rivals
“It was co-pilot, right? But they lost that. And then they were actually the first to image, really, with Dali. They lost that, right? Mid-Journey came out. They were kind of the first to, like, real video with Sora, and they lost that. And yet, they've gotten …”
Casado: Doubters of AI application defensibility and market fragmentation were wrong
“Like anybody that likes decried, oh, defensibility isn't going to work has been wrong. Anybody that's decried, like, it's all going to aggregate has been wrong. So zero sum thinking has been wrong.”
Wang: AI model inference costs dropped 10x year-over-year
“And in fact, I think model inference costs have gone down 10 X year over year.”
Wang: Specialized AI apps win in complex, data-integrated workflows
“You have complex workflows and a ton of customer data where deep integrations actually are necessary to get that last last mile value for the customer. This is where the specialized AI apps are sort of crushing, crushing any either foundation model layer or ot…”
Wang: AI-native companies reach $100M ARR faster than SaaS predecessors
“The first thing I'd call out that seems, that sort of just jumps off the page is that the AI native companies are far outpacing their SaaS counterparts and you can see it in terms of new companies blowing past this golden metric of time to a hundred million of…”
Wang: Most a16z AI founders work in office 6-7 days weekly
“Most of the founders that we work with are in the office six to seven days a week.”
Casado: AI solves startup customer acquisition but not user retention
“AI actually solves that problem. It just solves the bootstrap problem. It's like, these models are so magical. So, like, You know, you wrap one of these models you know, you make it available, and people think it's amazing they show up. But what's also clear i…”
Casado: The AI technology stack offers no inherent competitive moats
“There is no, as far as I can tell, as far as we can tell, there is no inherent endemic mode In the technology stack to AI, other than just overcoming the bootstrap problem.”
Casado: Early AI coding tools reached $400M to $600M in ARR
“The first kind of monetized AI app was code, right? It was Copilot, and so Microsoft had invested a ton of money and matured the market with Copilot, with VS Code, and so everybody knew it. It's just, like, the models weren't quite ready then, and so you had, …”
Wang: a16z portfolio CTOs reported 10% to 15% AI productivity gains
“Last year, it was notable that when we asked, hey, CTOs across 24 portfolio companies how much is AI actually impacting your productivity? And the answer across the board was pretty much 10 to 15%. We're all using GitHub Copilot.”
Wang: All 24 surveyed a16z startups use Cursor, driving up to 10x productivity
“This year I was pretty blown away by the answers that we got. They spanned from, call it, 30 to 50% on the low end in terms of productivity gains to, I kid you not, one CTO told us that he had seen a 10 X productivity lift from himself and his team. They were …”
Wang: Decagon cuts support costs by up to 80% and doubles CSAT scores
“If you talk to a Decagon customer, they're actually slashing their customer support costs by up to 80%, and not only that, are they, they're seeing deflection rates go up from 30% to anywhere from 60 to 80%, and their CSATs, their customer satisfaction scores,…”
Casado: Tech supercycles originate with prosumer adoption before enterprise penetration
“Every time we have a super cycle, it tends to start, you know, in these prosumer ways, right? The internet did this, right? Like, remember when Sun outlawed the browser, right? This is like Sun Microsystems, right? But they didn't really know how to consume it…”
Wang: AI applications lack 95% gross dollar retention of 2010s SaaS
“A lot of these high growth apps are not your typical system of record, 95% gross dollar retention companies that we sort of saw in the twenty-tens but importantly, that doesn't mean you should throw the baby out with the bathwater.”
Casado: China excels at AI models but struggles in enterprise software
“Historically, China's just not been able to build software, at least for, like, the prosumer enterprise market, which is kind of my world. They've just never really been able to do that, and so I think that, you know, their ability to compete at, like, a softw…”
Casado: a16z avoids lesser-known SOTA AI model startups due to heavy subsidies
“So the state-of-the-art model market, this is like the, you know, the Anthropics and the Open AIs it's incredibly competitive, and it's very heavily subsidized, right? I mean, like with Meta and like Google, et cetera, and so kind of our view is, is you have t…”