The MAD Podcast with Matt Turck

Episode: AI Engineering Revolution: Winners, Chaos & What’s Next | FirstMark
Date: July 3, 2025
Host: Matt Turck
Guest: David Walter (FirstMark Capital)

Episode Overview

In this episode, Matt Turck and his FirstMark colleague David Walter take a venture capital lens to one of the biggest stories of the year: the meteoric rise of AI-powered software engineering. They explore how generative AI has supercharged the productivity and capacity of engineering teams, creating runaway growth for startups in the space — but also a range of novel challenges, opportunities, and industry ripple effects. Drawing from historical analogies, in-the-trenches anecdotes, and recent survey data, they break down both the excitement and the chaos unleashed by AI coding tools, with lessons for founders, CTOs, and investors navigating the ever-evolving landscape.

Key Discussion Points & Insights

The Unprecedented Growth of AI Coding Startups

• Notable companies have experienced explosive ARR growth:
  • Cursor: $500M ARR in ~2 years, possibly the fastest B2B growth ever ([00:00], [07:31])
  • Lovable: $0 → $60M ARR in two quarters
  • GitHub Copilot: $400M ARR, 15M developers ([00:00], [08:01])
  • Windsurf and Replit: huge ARR leaps in short timeframes ([09:25])
• “There really is no plateau in sight for innovation. You might miss something spectacular and industry shifting.”
  — David Walter [00:11]

Why Has Generative AI Flourished in Coding?

• Four main reasons AI is especially well-suited for code ([04:12]):
  1. Massive, high-quality open source training data, especially from GitHub
  2. Code is structured and precise — ideal for machine understanding
  3. Coding relies on patterns AI can easily learn and reproduce
  4. Clear ROI: saves huge amounts of “grunt work” and time
• Adoption has been rapid because coding already had behaviors like “copy/paste from Stack Overflow” and code completion (IntelliSense, 1996) which set the stage ([06:28]).

Industry Impact — Productivity & Process Shifts

• Tangible improvements across engineering teams:
  • 30-50% faster throughput
  • 12% increase in PR merges
  • 17% more time on roadmap features vs. maintenance
  • 82% of engineers already use AI to write code ([11:03])
• “That is just, again, an adoption curve that is pretty much unprecedented.”
  — David Walter [11:41]
• The UI and “distribution model” for AI coding fits seamlessly into dev workflows (IDEs), driving easy uptake ([11:46]).

Challenges: Production Surges Lead to “Cleanup Crews”

• Historical analogy: like the Gutenberg press or the Ford assembly line, big tech leaps create downstream chaos, requiring new markets and industries (“cleanup crews”) for quality, safety, management, and governance ([13:31], [14:50]).
• In software, AI codegen has vastly accelerated code output (commits), but downstream DevOps (CICD, testing, QA, review) is straining and “kind of breaking” ([17:00], [19:10]).

Historical Parallels:

• Gutenberg Press → Book editors, publishers, libraries, regulation to handle misinformation & overload
• Ford Assembly Line → Inspectors, mechanics, highways—and new risks, regulations
  — “As production surges, you see all of these problems come in the wake of that production and then, and then new industries come around.”
  — David Walter [17:19]

New Roles and Organizational Changes

• Engineers in some orgs are shifting to “professional code reviewers” as AI generates more code ([20:09])
  • “People sitting in a toll booth letting cars pass or not pass, very much doing the same thing with code.” — David Walter [20:16]
• Hiring shifts: The highest unemployment rates among college majors now include CS grads, despite strong demand at a macro level ([29:15])
  • Rise of the “editor,” reviewer, and prompt engineer — less demand for traditional code monkeys
• The structure of teams is evolving:
  • Smaller teams, more reliant on reviewing and managing AI-written code ([31:12])
  • Product, engineering, and design (EPD) boundaries are blurring; if anyone can code, how do teams work? ([33:23])

What’s Breaking? Where Are the Bottlenecks?

• Security vulnerabilities are spiking ([24:02]):
  • Models trained on open source potentially propagate old vulnerabilities
  • CI/CD pipelines, testing, and QA are overwhelmed
  • New types of “flake”—i.e., tests that unpredictably pass/fail
• “A new focus on code review that is pretty unrelenting, that I'd say has gone from a step in a process to very much a job in and of itself.”
  — David Walter [25:54]

AI Reviewing AI & The Next Wave of Tools

• Can AI review AI-generated code? Agentic tools starting to emerge: semantic QA, code reviewers, auto-patching tools ([25:54], [27:12])
  • Companies are racing to solve newly emerging pain points as fast as they appear
• “All of the problems that I'm talking about are very much our opportunities as investors and founders.”
  — David Walter [26:00]

How CTOs & Leaders Must Adapt

• Major decisions loom on:
  • Talent: Editors/prompt engineers over just coders ([29:15])
  • Architecture: From design docs to machine-readable guardrails, “IaC-esque” constraints for AI ([31:12])
  • Team structure: Fewer, smaller, multidisciplinary teams
  • Governance: New need for code lineage (“provenance”), fine-grained reviews, and tight constraints ([35:01])
  • Security: Emergence of tools that catch bugs as code is written, not days later ([35:01])

Is the Market Tapped Out? Or Is There Still Opportunity?

• On competitive barriers:
  “It's always incredibly easy to say the alpha's left the room... At once there's a million things to go build. And on the other hand, there's also probably a million people trying to do it.”
  — David Walter [40:30]
• Developers are “picky” buyers, so there’s still plenty of room for new entrants with unique approaches
• Giant platforms don’t close the market, they open new waves of downstream need (e.g., AWS & the cloud)

The Power of Social Proof & Distribution

• Startups can ride the “halo effect” from being included in the hot new engineering stack (e.g. Supabase and Neon benefit from Cursor/Lovable’s growth) ([42:34])
• In this platform shift, brand new companies can be more credible or desirable than older, bigger incumbents if perceived as “AI native” ([44:23])
  • “You can be a one year old or two year old company and actually be a lot more credible than a 5, 7, 10 year old company...”
    — Matt Turk [44:23]
• Word-of-mouth in the B2B developer universe is more intense than in consumer tech

The Rise, Fall, and Rise of Developer Tools (from an investment perspective)

• “Sweet revenge” for developer tools: once seen as tough, “cheap” buyers, but now the hottest segment in generative AI ([46:15], [47:07])
  • “Their relative importance across organizations has seemed to just go up and to the right over the last decade or so.”
    — David Walter [47:25]
  • The “definition of a developer tool” is broadening—Stripe as a dev tool example

Notable Quotes & Memorable Moments

• On speed of innovation:
  “If you go offline and you aren't keeping up with the updates for a week at a time, you really might miss something spectacular and industry shifting.”
  — David Walter [11:05]

• On historical cycles:
  “With every surge in production, there's just a cleanup crew that naturally comes and a new market and industry that follows in its wake.”
  — David Walter [13:31]

• On engineering roles changing:
  “Many of [Malte Ubl’s] great engineers are actually becoming predominantly professional code reviewers... as AI code gen has just totally increased the rate at which they're producing and outputting code commits.”
  — David Walter [20:09]

• On hiring market shifts:
  “Computer science grads are actually among the top five or six majors graduating from college right now with the highest unemployment rate... There's a huge demand for software engineering as a concept, [but] the people who are trained in that practice are actually not in high demand.”
  — David Walter [29:15]

• On new skillsets:
  “We're seeing CTOs very much focused on hiring great editors and reviewers and prompt engineers who can almost shape and validate and curate that AI generated output.”
  — David Walter [30:07]

• On constraints and governance:
  “Do whatever you want with AI, but whatever you pull over the fence needs to be a few hundred lines.”
  — Matt Turk [34:40]

• On market opportunity:
  “It would be like saying, well, AWS, Azure and GCP came around and they ate up the whole cloud opportunity. So I guess there's no more money in cloud anymore. Of course that wasn’t the case.”
  — David Walter [41:42]

• On platform shift credibility:
  “You can be a one year old or two year old company and actually be a lot more credible than a 5, 7, 10 year old company... because you’re part of that platform shift and you’re ‘AI native’...”
  — Matt Turk [44:23]

Timestamps for Important Segments

• [00:00] – Opening statistics on explosive ARR growth (Cursor, Lovable, Copilot)
• [04:12] – Why coding is a uniquely good fit for generative AI
• [07:31] – How AI coding behaviors build on existing developer habits
• [11:03] – Measurable impacts of AI tools on engineering orgs
• [13:31] – Historical analogies: Gutenberg Press, Ford assembly line, and their “cleanup crew” industries
• [19:10] – Downstream strain: debugging, security, testing, quality review breakpoints
• [20:09] – The rise of professional code reviewers
• [22:18] – Survey data: positive and negative effects on engineering orgs
• [24:02] – Deep dive: what’s breaking (security, pipelines, QA)
• [27:12] – Emerging investment opportunities from new bottlenecks (auto-patching, agentic review tools)
• [29:15] – CTO decisions: hiring, architecture, team structure, and governance in the AI era
• [33:23] – The democratization of coding and its impact on enterprise team dynamics
• [35:01] – Governance: enforcing review constraints, code provenance, managing velocity/entropy
• [40:30] – Is the era of opportunity over? How to find “alpha” in a crowded field
• [44:23] – Social proof and AI-native credibility for startups
• [46:15] – The changing fortune of developer tools as a VC category

Conclusion

The engineering revolution brought on by generative AI is exhilarating, chaotic, and wide open — but also creating new, urgent bottlenecks throughout the software development pipeline. Waves of disruption echo historical tech leaps, spawning entirely new industries as the “cleanup crew” behind every productivity surge. As engineering roles, hiring, team structures, and best practices shift under our feet, the enduring opportunities for founders lie in innovative solutions to the chaos, tactically riding new distribution channels and platform effects. For VCs and founders alike, the only constant is rapid change and the pressing need to stay plugged into the fast-moving ground truth of engineering in the AI-native era.