Sep 24, 2025 · 59m · a16z
Software Finally Eats Services - Aaron Levie
Aaron Levie · 23m spoken Steven Sinofsky · 17m spoken Martin Casado · 12m spoken Erik Torenberg · 1m spoken
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
In this episode of The a16z Podcast, host Erik Torenberg and panelists Aaron Levie, Steve Sinofsky, and Martin Casado explore how AI transforms developer productivity, accelerates startup creation, and expands software markets into professional services.
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 2.8% of the talking time here. How this is scored →
The host as informed peer 0.7 Guest teaching 2.2 Guest disagreement 2.0 The host pushing back 0.4
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 5:27 Aaron pushes back on A16Z startup framing
Aaron explicitly rejects Martin's premise about startup hiring by pointing out that Andreessen Horowitz's portfolio is not representative of the broader startup ecosystem.
Hardest push from the host ▶ 12:53 Host introduces METR study counter-evidenceThe host challenges the narrative of unquestioned AI productivity gains by bringing up the METR study findings showing developers were actually less productive with AI tools.
Biggest teaching moment ▶ 16:36 Steven explains domain expertise and early adopter forgivenessSteven reframes self-reported productivity metrics by educating the panel on how early adopters forgive major product flaws while domain experts amplify output.
The host holds their own ▶ 55:54 Host constructs mobile platform shift comparative framingThe host demonstrates deep ecosystem knowledge by pointing out that despite major new products in mobile like Instagram and Uber, incumbents like Facebook and Google captured the bulk of long-term value.
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 |
|---|---|---|---|---|---|---|
| Group Chat Memes and Episode Opening | 1 | 1 | 4 | 1 | The host lightly introduces the episode with banter about memes and kicks off the H-1B immigration debate. Aaron Levie and Martin Casado politely clash over how salary thresholds impact early-stage startups versus Andreessen Horowitz-backed companies. | |
| Reevaluating the H-1B Lottery System vs. Salary Thresholds | 0 | 2 | 3 | 0 | The host remains entirely silent as the panel debates body shops, IT salary bands in Florida, and Keith Rabois's proposed $20k salary floor. The guests reframe tech hiring dynamics amongst themselves. | |
| AI Developer Productivity vs. the METR Study Findings | 2 | 3 | 3 | 2 | The host introduces the METR study showing reduced developer productivity, prompting Aaron to detail internal 30% AI code generation metrics and background agents. Steven Sinofsky reframes productivity measurements by highlighting early adopter forgiveness. | |
| Historical Evolution of Startup Velocity Across Tech Eras | 0 | 2 | 2 | 0 | The host does not intervene while the guests compare startup velocity across 1995, 2005, and 2025. Sinofsky notes that AI is a fundamental refactoring of product velocity. | |
| Shadow Productivity, Enterprise AI Failures, and Code Quality | 0 | 2 | 2 | 0 | The host is absent from the discussion as Martin and Steven break down enterprise AI failures, shadow productivity, and why non-deterministic software conflicts with large enterprise operations. | |
| The VisiCalc Analogy and Domain Expertise as an AI Multiplier | 0 | 3 | 1 | 0 | Without host participation, Steven asks Aaron how domain expertise prevents AI misuse, leading to a comparison between pre-spreadsheet financial modeling in 1985 and modern AI workflows. | |
| Prosumer AI Monetization, Jevons Paradox, and Personal Utility | 0 | 2 | 1 | 0 | The host is silent while Martin explains prosumer monetization, Jevons paradox in creative AI tools, and why output richness increases even if production costs remain unchanged. | |
| AI as a Landscape Reset for Younger Founders | 3 | 2 | 2 | 1 | The host cites Patrick Collison and brings up Gen Z founder success rates. Aaron explains how AI neutralizes traditional incumbent distribution advantages for younger founders. | |
| Overestimating Incumbent Advantages in Platform Shifts | 0 | 3 | 1 | 0 | The host does not speak as Steven and Aaron analyze past platform shifts, pointing out how market watchers repeatedly overestimate incumbent advantages during major technological transitions. | |
| Organizational Inertia and Changing Buying Behaviors | 0 | 3 | 2 | 0 | In a guest-only discussion, Martin and Steven critique the popular misapplication of Clay Christensen's Innovator's Dilemma to modern software businesses. | |
| Expanding TAM and the Shift from Services to AI Software | 0 | 2 | 2 | 0 | The host is inactive while Martin describes his yearly AWS re:Invent reassurance routine for open-source founders and Aaron explores AI capturing non-software services TAM. | |
| Rapid Consumer Adoption and Brand Dynamics in AI | 0 | 2 | 1 | 0 | The host remains silent as Steven compares 1999 Pew study data on internet adoption rates to contemporary consumer AI usage patterns. | |
| Predicting the Long-Term Winner Landscape in AI | 3 | 2 | 2 | 1 | The host frames a final question comparing AI market cap accumulation to the mobile era where incumbents captured the largest gains. Aaron, Steven, and Martin project long-term value allocation between incumbents and insurgents. |
Statements from this episode (31)
Levie: AI adoption speed exceeds prior tech shifts and changes daily routines
“The universal adoption of this as a consumer technology and then bleeding into prosumer is, it exceeds anything I've ever experienced, and I think it is, it will just fundamentally change people's sort of daily pattern.”
Casado: Senior engineering teams using AI are superhumanly productive
“The more senior small teams that use AI are superhuman. It's like they woke up, and they were all f***ing Tony Stark, and it's unbelievable, and like, their productivity is insane.”
Casado: Replacing H-1B lottery with pricing helps startups beat big tech
“This system has been gamed for a very long time. It's very hard for startups. To hire because of the lottery system. It's locked up by the large companies, the consultants, Amazon and Google and like that has to change, and I think a very reasonable way to do …”
Levie: Pricing H-1B visas allows big tech to capture most talent
“I think that you'd have a situation where the Amazons and Googles would probably actually capture the vast portion of the talent in this situation. So it's not clear to me that, like, startups sort of are come out ahead or better off from this particular imple…”
Casado: Price-based H-1B visas will squeeze IT consultancies
“There are a number of organizations that are consultancies. That actually are price sensitive. That would be squeezed by this. I would think that You know, given that they're in the top 15, they make up like four or five of them. That would be a significant fr…”
Sinofsky: Big tech narrowed domestic university recruiting to 25-30 schools
“For the past 25 years or so, they really went on this sort of bifurcated curve, which is hiring on, you know, for the people in the office, focusing on, say, 25 or 30 university departments, and then basically Everybody else was like, well, it's so much easier…”
Casado: H-1B consulting firms squeeze out $80k to $120k US IT jobs
“It's very, very tough to get a job between like 80 and a 120 K in, in much of the United States because of this, right? And so this isn't about like a new grad being a software engineer, because the reality is, is the expected value of a software engineer over…”
Levie: AI currently generates 30 percent of Box's codebase
“About 30% of our code right now Is, is coming from AI.”
Levie: Small startup founders report 3x to 10x AI productivity gains
“I will regularly talk to three, five, ten-person startup founders that, that self-report they might be getting somewhere on the order of, like, three to five to 10 X productivity improvements.”
Levie: AI adoption by young founders is the biggest startup shift ever
“If I look at how these companies run versus today, it's the biggest change in, in how you start and run a company that I've ever seen.”
Levie: AI transforms company creation in ways cloud computing never did
“By 2005, other than the fact that our resources were in the cloud versus, you know, you had to go to a data center. In our case, we actually still went to the data center. Like, not that much about the company building process was different. Today, in 25, ever…”
Sinofsky: AI accelerates product creation while SaaS primarily accelerated monetization
“This is, the thing that AI is, is an accelerant in building the product. Whereas what sass and the cloud did was accelerate getting paid.”
Sinofsky: Non-determinism will be the main challenge for enterprise AI adoption
“And I really feel like that is going to be The hugest challenge in, in large organizations figuring out how to adopt things is, like, they're gonna just get all bunched up over, over non-determinism.”
Casado: AI coding assistants improve code quality rather than shipping velocity
“So like, what do people, in my experience, the more senior folks actually use AI for code? It's like documentation, writing, testing. I mean, it's a lot of the other stuff that may not actually like increase like the shipping schedule, but like you get a lot m…”
Levie: Domain experts, not generalists, benefit most from AI tools
“This is this really counterintuitive thing where, and you've talked about specialization on the last one is like it, the biggest gains of AI go to people who have some degree of expertise in an area to know what is actually true. What is not going to work? Wha…”
Casado: Generative AI monetization is driven by professionals, not casual users
“If you draw a dollar at random that's monetized, it's from a professional. And for obvious reasons, because, you know, like, you know, they can produce, if you draw a user at random, it's casual and it's in the tail. And so it's fairly clear that, you know, th…”
Casado: Contracting AI creatives costs the same as using traditional techniques
“I have seen the shift for contracting someone that uses traditional techniques to AI. Dollar amounts are the same. And so again, like, you know, this is Jevon's paradox all over again. Like, they spend just as much time. You know, it's just the output tends to…”
Levie: Distribution is incumbents' only real advantage over AI startups
“There, there's like all, there is incumbent advantage in distribution, but that is it. There's no other, other real advantage.”
Levie: Startups gain enterprise scale because incumbents refuse to retool workflows
“And they're not going to retool their entire internal engineering workflows to move at 10 X the pace. And so a brand new startup can go and do that and then instantly get the scale of a larger company.”
Steve Sinofsky: Intel missed both the GPU and data center market opportunities
“This, you know, this is like Intel with the GPU. Like they missed the GPU. In 2005. Right. And they missed the opportunity to buy the company, to do the work or whatever, and they kind of missed the data center too. It just took a longer time to figure out tha…”
Casado: AI creates new buying behaviors that disfavor incumbents over startups
“I will say incumbents are very bad when new user behaviors and buy behaviors show up. Like, in particular, they don't know how to cater to it. AI is definitely a new user behavior and a new buying behavior, and so this is very much an advantage of startups, ju…”
Casado: Microsoft's initial Copilot product was actually created by OpenAI
“I learned recently it was created by OpenAI, so that kind of explains it.”
Sinofsky: Apple launched the iPod as a desperate Hail Mary move
“People always talk about like the brave, it's like their computer business was Dead, dead, dead. Like there was, it was like three percent share and going nowhere. So like it was like a Hail Mary. The iPod was a Hail Mary.”
Levie: Tech incumbents rarely disrupt themselves, they expand into new markets
“And to your point, it's actually very rare whenever we say these companies disrupted themselves, it's almost never the case that they disrupted themselves. It's the case that they went after a market that they had no, No actual, you know, take the, you know, m…”
Casado: AWS has never put a startup out of business
“I can't think of one company AWS has ever put out of business by launching a service, and you know what, they've all been fine.”
Sinofsky: Professional Services Firms Will Transition Into Software Companies
“Huge numbers of these verticals are just going to be existing pro serve that turn into software providers.”
Levie: AI-native systems integrators will decisively outcompete legacy service firms
“If you wanted to be a new systems integrator, and your whole point is, is that we are a systems integrator, but we use cloud code, or we use cognition, or we use, you know, cursor to get the output, you will have such an advantage over any incumbent because th…”
Levie: Consumer adoption will drive a massive enterprise AI upgrade cycle
“So this is why, this is just lays the foundation for why we're going to see just a massive upgrade cycle in the enterprise.”
Sinofsky: Distribution is no longer a moat because smartphones are ubiquitous
“Also, to build on your point earlier, like, about distribution, and the reason that distribution is not the advantage that it used to be is because It's already exists on seven billion phones.”
Casado: Early AI breakouts gain strong brand effects over model performance differences
“Honestly, last quick point on this is, like, for the first time in a very long time, we're seeing brand effects with an early technology, and what I mean by that is, if you look at, like, whatever, the major model providers, you know, how much better are they …”
Levie: AI market will follow SaaS pattern with $100B new disruptors
“I think it's going to look like what we saw in something like SAS or cloud, which is the incumbents get bigger, but then there's all of these new categories that we would not have been able to predict. And then there's lots of 10 and 20 and fifty and one hundr…”