Jul 3, 2025 · 50m · mad
AI Engineering Revolution: Winners, Chaos & What’s Next | FirstMark
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 MAD Podcast, FirstMark's Matt Turck and David Walsher analyze the AI software engineering revolution, examining hyper-growth coding platforms, downstream DevOps bottlenecks, evolving CTO responsibilities, and emerging startup opportunities across derivative AI infrastructure.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 30.2% of the talking time here. How this is scored →
Matt as informed peer 5.5 Guest teaching 3.7 Guest disagreement 0.2 Matt pushing back 0.7
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 40:30 David rejecting 'alpha has left' premise
David directly counters Matt's question on whether startup opportunity has vanished, explaining how picky developer habits and derivative infrastructure needs create massive ongoing opportunities.
Hardest push from Matt ▶ 37:30 Matt raising gross margin and retention flawsMatt refuses to accept surface-level ARR growth metrics without pointing out severe underlying industry issues, such as churn from casual prototyping and unprofitable unit economics.
Biggest teaching moment ▶ 20:09 David on senior devs becoming professional reviewersDavid re-frames the shift in engineering talent by sharing insights from Vercel's CTO, explaining how experienced engineers are pivoting from writing code to acting as strict gatekeepers and reviewers.
Matt holds his own ▶ 4:12 Matt laying out four core adoption driversMatt demonstrates clear domain authority by systematically listing four structural reasons why code lends itself to AI generation, earning explicit validation from the guest.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Two Decades of Software Engineering Evolution | 3 | 4 | 0 | 0 | Matt sets up the presentation format and offers light banter about pre-cloud history being prehistoric times. David outlines six historical tech shifts that set the stage for generative AI in software engineering. | |
| Core Factors Driving AI Coding Adoption | 7 | 1 | 0 | 0 | Matt takes the lead by laying out four core technical and structural reasons why AI coding succeeded early, including public GitHub training data and rigid code syntax. David explicitly confirms Matt nailed the analysis and builds on developer buyer persona details. | |
| The 24-Month Wave and AI Coding Market Winners | 6 | 3 | 0 | 0 | David shares rapid revenue growth figures for AI tools like Cursor and Lovable. Matt showcases domain familiarity by connecting these milestones to past podcast interviews with GitHub's Thomas Dohmke, Vercel's Guillermo Rauch, and Replit's Amjad Masad. | |
| Tangible Engineering Productivity and Adoption Survey | 5 | 4 | 0 | 2 | David presents productivity metrics from engineering teams using code generation. Matt nudges the conversation toward the emerging distinction between agentic tools like Devin and co-pilots like Cursor. | |
| Historical Analogies: Production Surges and Cleanup Markets | 4 | 5 | 0 | 0 | David walks through historical analogies, comparing AI code generation surges to the printing press and Ford assembly line. Matt chimes in with light humor regarding monks losing their jobs to printing presses. | |
| Downstream DevOps Bottlenecks and Survey Findings | 6 | 4 | 0 | 0 | Matt illustrates downstream DevOps bottlenecks using an anecdote from Canva's CTO regarding a massive 50,000-line pull request. David shares survey data showing efficiency gains paired with reliability drops. | |
| Critical Failure Points in the AI Development Lifecycle | 5 | 5 | 0 | 1 | David outlines specific system failures in security, CI/CD pipelines, and testing flakiness caused by non-deterministic code generation. Matt notes new vulnerability types and asks how effectively AI can review AI. | |
| Emerging Startup Opportunities in AI Engineering | 5 | 4 | 0 | 0 | David highlights emerging startup categories addressing QA and code review issues. Matt notes that extreme market noise makes it harder for new startups to break through despite high VC activity. | |
| Organizational Shift and the Evolving Role of the CTO | 6 | 4 | 0 | 1 | David presents structural shifts facing CTOs in hiring, guardrail architecture, and governance. Matt details Canva's resolution to strictly enforce small PR limits to preserve code quality. | |
| CodeGen Productivity Surge and Rapid AI Evolution | 7 | 3 | 1 | 3 | Matt probes beyond the top-line ARR hype, raising critical operational concerns regarding poor retention and negative gross margins among AI coding startups. David acknowledges simple wrapper delivery models and explains how leaders are re-investing capital into proprietary tech. | |
| Founder Opportunities Amid Big Tech Competition and Rapid Model Releases | 6 | 5 | 2 | 2 | Matt asks whether Big Tech dominance and rapid model releases mean alpha has left the room for early-stage founders. David gently counters the premise, noting developer buyers are highly opinionated and that platform shifts create massive derivative software markets. | |
| The Halo Effect in Developer Stacks and Social Proof Velocity | 6 | 3 | 0 | 0 | Matt notes the strong halo effect where AI coding adoption drives rapid growth for adjacent tools like Supabase and Neon. David details how modern social proof and direct dev marketing create major distribution advantages. | |
| The Venture Capital Revival of Developer Tools | 6 | 3 | 0 | 0 | Matt reflects on the historical VC skepticism towards developer tools, calling the current AI coding surge a form of sweet revenge. David elaborates on how devtools evolved into high-value enterprise categories before concluding the interview. |
Statements from this episode (22)
Walsher: Global software developer population grew 7x since 2006
“Over that time, so from, call it, 2006 all the way to now estimated about seven x growth in the number of global software developers”
Turck: Coding is currently the most successful generative AI application
“When people talk about, okay, what generative AI applications have been successful so far, coding seems to be number one by far.”
Turck: Microsoft created IntelliSense automatic code completion in 1996
“I believe IntelliSense was created in 96 or something like that by Microsoft, which was already a automatic code completion.”
Walsher: Cursor announced reaching $500M in ARR
“Cursor, maybe three weeks ago, announced they're at five hundred million dollars of ARR.”
Walsher: Lovable reached $60M ARR in two quarters
“Lovable, which allows people to create prototypes and web apps through a series of prompts, has gone zero to sixty million dollars of ARR in the last two quarters.”
Walsher: GitHub Copilot reached $400M ARR and 15M developers
“GitHub Copilot, which I know you just had Thomas on the podcast not too long ago. Four hundred million dollars of ARR, fifteen million developers, very much the steward of this category first mover.”
Walsher: Windsurf recently reached $100M in ARR
“Windsurf, which is rumored to have sold to OpenAI Hit a hundred million dollars of ARR quite recently.”
Walsher: Replit grew ARR from $10M to $100M in six months
“And then Replit, which went from 10 to a hundred million dollars of ARR in just the last six months.”
FirstMark survey: 82% of developers use AI to write code
“We've seen a 30 to 50% faster throughput. We've seen a 12% increase in PR merges. This is a really important compounding stat here on the bottom left, which is a 17% increase in the amount of time folks are spending on roadmap versus maintenance and keeping th…”
Walsher: End-to-end AI coding agents are currently limited to low-complexity tasks
“And for now, many of the agentic solutions that are truly end to end, you know, complete something are very much being pointed at lower level tasks that I'd say are quote unquote more mindless that have less dependency complexity you know, that have less of a …”
Walsher: Production surges naturally create downstream cleanup crew markets
“It tends to be the case that actually a lot of problems emerge and in their wake markets follow or maybe a punchier way to say that would be with every surge in production, there's just a cleanup crew that naturally comes and a new market and industry that fal…”
Canva engineer submitted a 50,000-line AI pull request
“We had Brendan Humphreys, CTO of Canva, who was talking about that exactly, which is, okay, this is great that you can create code, but, like, we had one guy who submitted a PR that was 50,000 lines and they have a peer review culture, and Yeah. They basically…”
Vercel CTO says top engineers are becoming primarily code reviewers
“Malte, among many interesting things, I think one of the most fascinating things that he said was that most of his great engineers who have been with the company for a while, as they've dogfed you know, vZero, and they've used things like Cursor and Windsurf i…”
Walsher: Code test flakiness is skyrocketing due to AI code generation
“We're just seeing like flakiness generally across code skyrocketing.”
Walsher: Computer science graduates face top-tier college unemployment rates
“Computer science grads are actually among the top five or six majors graduating from college right now with the highest unemployment rate.”
Turck: Many fast-growing AI startups operate at a unit loss
“And there's reportedly open questions around gross margins as well which means that on a unit basis, a lot of those companies operate at a loss. The more they serve customers, the more they lose money,”
Walsher: Product ubiquity is rare in the developer tools market
“It's rare that you see ubiquity is what I'm trying to say in the world of developers.”
Turck: Supabase and Neon gained massive uptake from Cursor and Lovable
“So famously Supabase and Neon on the database side have had a massive sort of uptake based on the success of Curse and Lovable.”
Walsher: B2B software tools get more intense public discussion than consumer apps
“You see more reviews and love and hate for software tools in the B to B universe than you do sometimes for, like, mass market consumer phenomenons”
Walsher: Direct marketing and distribution are now the primary startup moats
“In many ways it feels like we are at risk of being like a copycat world where you see some success online and then you go copy it the next day, but it feels like marketing and distribution and the ability to communicate directly one-to-one with your audience h…”
Turck: AI-native startups hold more buyer credibility than larger incumbents
“You can be a one-year-old or two-year-old company and actually be a lot more credible than a five, seven, ten-year-old company, which may be 10 X, hundred X your size, but because you're part of that you know, platform shift and you're like AI native, people t…”
Walsher: Tech buyers value Cursor's opinion over $80B legacy software incumbents
“Most people will care, at least in our world, What the smartest buyer at Cursor thought about a given tool than what the smartest person at the, you know, eighty billion software business that IPO'd in, in 2012 might think.”