Andrew Ng: Product Team Ratios Evolving to Just One Software Developer for Every Two Product Manager

Silicon Valley is abuzz with the latest provocation from AI luminary Andrew Ng – and it’s not about some fancy new model or algorithm. It’s about product managers. Yes, you read that right. In a recent talk at Y Combinator’s Startup School, Ng suggested that the traditional makeup of software teams is on the cusp of a radical shift. How radical? Try two software developers for every one product manager. In Ng’s own words: “Just yesterday, one of my teams came to me, and for the first time... this team proposed to me not to have 1:4 PM/engineers, but to have 1:0.5 PM/engineers.”reddit.com That works out to twice as many PMs as engineers on a team. product managers two software developers for every one product manager “Just yesterday, one of my teams came to me, and for the first time... this team proposed to me not to have 1:4 PM/engineers, but to have 1:0.5 PM/engineers.” reddit.com twice as many PMs as engineers https://www.youtube.com/watch?v=RNJCfif1dPY&embeds_referring_euri=https%3A%2F%2Fapp.hackernoon.com%2F&source_ve_path=MjM4NTE&embedable=true https://www.youtube.com/watch?v=RNJCfif1dPY&embeds_referring_euri=https%3A%2F%2Fapp.hackernoon.com%2F&source_ve_path=MjM4NTE&embedable=true This bold claim is entwined with the rise of “vibe coding”, a term coined by Andrej Karpathy to describe using AI to generate code from high-level prompts. With AI copilots writing boilerplate code at lightning speed, Ng argues the bottleneck in software development is no longer engineering capacity – it’s deciding what to build. “I don't see product management work becoming faster at the same speed as engineering. I'm seeing this ratio shift.” Ng said, observing that while AI drastically accelerates coding, product decision-making isn’t speeding up in tandem. If code is cranked out like cheap candy, then whoever figures out the right candy recipes (product managers) suddenly becomes indispensable. In Ng’s view, “I still don't know if this is a good idea, but... for the first time in my life, managers are proposing having twice as many PMs as engineers. I think it's a sign of where the world is going.” “vibe coding” what to build “I don't see product management work becoming faster at the same speed as engineering. I'm seeing this ratio shift.” “I still don't know if this is a good idea, but... for the first time in my life, managers are proposing having twice as many PMs as engineers. I think it's a sign of where the world is going.” Is this hype, heresy, or a bit of both? Ng’s statement is certainly a head-turner in an industry that has long glorified engineers as the primary creators and PMs as the support crew. But with generative AI writing code on vibe, maybe the real power shifts to those who set the vibe – the PMs setting direction. Let’s unpack the arguments on both sides of this PM vs Engineer ratio debate, because the Twittersphere (or X-sphere) and HackerNews commentariat are anything but silent on this. writing code on vibe set Ng’s Bold Prediction: Why PMs Could Become Even More Important Andrew Ng isn’t casually throwing out the 2:1 (Eng:PM) ratio claim without reasoning. The core logic stems from simple economics of complements. He notes that “writing software, especially prototypes, is becoming cheaper” thanks to AI, which inevitably “will lead to increased demand for people who can decide what to build.” In other words, when building things becomes easy, figuring out the right thing to build becomes the critical path. AI-assisted coding (the essence of vibe coding) might let one engineer do the work of five or ten, but no amount of AI can yet replace understanding user needs, crafting a product vision, and defining features – the bread and butter of product managers. As Ng tweeted, “This will significantly increase demand for people who can come up with clear specs for valuable things to build.” (Translation: hello, PMs via linkedin.com) “writing software, especially prototypes, is becoming cheaper” “will lead to increased demand for people who can decide what to build.” figuring out the right thing to build becomes the critical path right “This will significantly increase demand for people who can come up with clear specs for valuable things to build.” linkedin.com What does this future look like in practice? Ng gives the anecdote above of an AI Fund team proposing an unprecedented 1:0.5 PM-to-engineer ratio – a scenario where PMs outnumber engineers. Even Ng admits it “sounds very extreme”, but he’s seeing the early signs of this trend. And he’s not alone. Tech investor SC Moatti chimed in to back Ng up: “Excellent point Andrew Ng – our projections are that [the] current 10:1 engineer:PM ratio is going to become 2:1 because PMs will be able to do much more with less engineers.” When you have venture capital types essentially agreeing that the PM is the new rockstar, it’s a sign the idea isn’t coming out of left field. 1:0.5 PM-to-engineer ratio “sounds very extreme” “Excellent point Andrew Ng – our projections are that [the] current 10:1 engineer:PM ratio is going to become 2:1 because PMs will be able to do much more with less engineers.” the PM is the new rockstar Other product leaders share similar experiences, especially in the era of AI tooling. Hoda Mehr, a product founder, described how using AI made her feel like “the navigator. I point in the direction, and the AI implements. So, what Andrew says resonates with me.” In this view, the PM becomes the coach and strategist, while the AI-augmented dev team executes faster than ever. If “everyone codes” (Ng even mentioned his receptionist codes now, thanks to AI tools businessinsider.com), coding itself gets commoditized. The differentiation shifts to vision, prioritization, and product intuition – classic PM territory. “the navigator. I point in the direction, and the AI implements. So, what Andrew says resonates with me.” coach and strategist “everyone codes” receptionist codes businessinsider.com Ng’s broader point is that AI is reshaping roles: “Many companies have an Engineer:PM ratio of, say, 6:1... As coding becomes more efficient, teams will need more product management work (as well as design work) as a fraction of the total workforce.” He even founded a course on AI Product Management and often evangelizes it as a hot career path. In the AI era, product managers who understand AI (and maybe can sling a bit of code themselves) are in high demand. The fundamental reasoning: if one engineer with AI can now build an app prototype in a day, the limiting factor becomes how quickly you can decide on the next feature or validate the next idea. Execution is cheap; ideation and prioritization are the new gold. Ng’s memorable one-liner? “Writing software...is becoming cheaper. This will lead to increased demand for people who can decide what to build. AI Product Management has a bright future!” “Many companies have an Engineer:PM ratio of, say, 6:1... As coding becomes more efficient, teams will need more product management work (as well as design work) as a fraction of the total workforce.” AI Product Management “Writing software...is becoming cheaper. This will lead to increased demand for people who can decide what to build. AI Product Management has a bright future!” The Pushback: Why Many Technologists Are Skeptical (or Terrified) Before we start crowning PMs as the new kings and queens of tech, let’s address the chorus of eye-rolls and “oh please” retorts coming from many engineers and tech observers. The idea of two PMs per engineer is, as one Redditor bluntly put it, “on the list of stupid shit I have seen in this new AI wave, this ranks pretty highly.” Engineers are notoriously allergic to meeting overload and micromanagement, and the thought of an army of PMs swarming around a couple of devs is nightmare fuel for many. One commenter quipped, “Two PMs for one engineer? This is my actual nightmare.” – no translation needed. chorus of eye-rolls and “oh please” retorts one Redditor “on the list of stupid shit I have seen in this new AI wave, this ranks pretty highly.” meeting overload and micromanagement “Two PMs for one engineer? This is my actual nightmare.” A major argument against Ng’s rosy PM-heavy future is that AI might automate parts of the PM role too. Gaurav Khanna provides a counterpoint that Ng’s assumption “overlooks the fact that the same AI advancements could reduce the need for PMs by automating core aspects of their work.” He notes that “AI tools are increasingly capable of generating product specifications, prioritizing features based on user data, and analyzing feedback to identify opportunities for improvement. These traditionally labor-intensive tasks could be streamlined or even replaced by AI, leading to leaner product teams.” Ouch. In this scenario, instead of more PMs to manage the now-hyperproductive devs, maybe the PMs just get smarter tools and a single PM can handle what used to take three. After all, if ChatGPT can write user stories or analyze customer feedback, why hire extra humans to do it? AI might automate parts of the PM role too “overlooks the fact that the same AI advancements could reduce the need for PMs by automating core aspects of their work.” “AI tools are increasingly capable of generating product specifications, prioritizing features based on user data, and analyzing feedback to identify opportunities for improvement. These traditionally labor-intensive tasks could be streamlined or even replaced by AI, leading to leaner product teams.” the PMs just get smarter tools There’s also a camp that believes engineering won’t be as cheap and easy as Ng implies – at least not to the extreme of replacing devs with prompts. Sure, an AI can scaffold an app quickly, but seasoned engineers know the devil is in the details: debugging, integrating systems, dealing with edge cases, scaling the software, etc. One cynical Hacker News commenter derided Ng’s idea, saying they’d love “whatever this guy is smoking, cause 1 PM to 0.5 engineers is some grade-A mind rotted insanity right there.” They doubted PMs who “weren’t a net negative on projects” could magically “build jackshit by throwing a bunch of LLM-hallucinated crap at the wall.” The punch line: “Sure, the devs are the ones that are going to be replaced by the clueless middle managers who only exist to waste everyone's time.” – dripping with sarcasm, highlighting the skepticism that PMs (pejoratively cast as “middle managers”) could ever displace good engineers. engineering won’t be as cheap and easy as Ng implies One cynical Hacker News commenter “whatever this guy is smoking, cause 1 PM to 0.5 engineers is some grade-A mind rotted insanity right there.” “weren’t a net negative on projects” “build jackshit by throwing a bunch of LLM-hallucinated crap at the wall.” “Sure, the devs are the ones that are going to be replaced by the clueless middle managers who only exist to waste everyone's time.” Another perspective is the survival of the fittest: perhaps AI will expose weak PMs and weak engineers. As one observer mused, “Or is it the other way around? ... What if technically sharp designers and well-rounded developers actually end up pushing out incompetent managers?” In other words, if coding becomes easier, maybe engineers start encroaching on product turf themselves (since they can handle more of the product thinking when freed from grunt coding), or simply that bad PMs will have nowhere to hide. After all, if an engineer can ask an AI to generate ten variations of a feature, maybe they’ll rely less on a PM to specify everything up front. The power dynamic could actually tilt in favor of engineers who are versatile (“full-stack” in both code and product thinking), making excess PMs redundant. survival of the fittest and “Or is it the other way around? ... What if technically sharp designers and well-rounded developers actually end up pushing out incompetent managers?” bad PMs will have nowhere to hide excess And let’s not forget the pure cultural resistance: Many devs simply don’t want more PM oversight. A satirical scenario on Reddit imagined two PMs each pulling an engineer in different directions: “they both have different product visions but won’t talk to each other – they just both tell you to ignore the other one and do what they want.” Anyone who’s worked in a meeting-heavy, manager-heavy environment can relate to the dread of too many cooks in the kitchen. Doubling up on PMs could introduce communication overhead, conflicting directives, and bureaucratic bloat – the very opposite of the agile, lean teams that have been championed in software development for years. “they both have different product visions but won’t talk to each other – they just both tell you to ignore the other one and do what they want.” too many cooks in the kitchen Balanced (But Bullish) Take: Who’s Right About the Future Team Ratio? So, where does this leave us? As with most tech debates, the truth likely lies somewhere in the middle – but leaning toward Ng’s vision more than the skeptics might admit. Here’s the Kara Swisher-esque editorial verdict: Ng is onto something real, but the 2:1 ratio is an exaggerated scenario to get people thinking. It’s a provocative moonshot of an idea – and sometimes you need to state the extreme to spark the discussion. Ng is onto something real, but the 2:1 ratio is an exaggerated scenario On the bullish side, it’s hard to deny Ng’s core insight. AI has made building software faster and cheaper in many cases. Prototypes that once took a team weeks can now be whipped up by a savvy PM or designer with an AI assistant in a day. Ng’s own org, AI Fund, is apparently seeing teams experiment with the PM-heavy model in real life. And the market is rewarding those who can bridge business needs with AI capabilities. As Ng tweeted, “AI Product Management has a bright future!” – and we’ve seen a proliferation of courses, workshops, and roles specifically for AI-centric PMs. has “AI Product Management has a bright future!” However, being bullish doesn’t mean wearing blinders. It’s entirely possible that we won’t literally see teams with 2 PMs per engineer across the board. Instead, we might see the definition of roles blur. Perhaps tomorrow’s “engineers” are also doing a lot of product work (writing specs, choosing features) and tomorrow’s “PMs” are more technical (able to do light coding or at least AI-assisted prototyping). The ratio might not need to shift to 2:1 if each side starts encroaching into the other’s territory with AI’s help. A single engineer could output what 5 used to, and a single PM could manage what 3 used to – meaning you still have a 5:1 or 4:1 ratio in headcount, but vastly more productivity on both sides. The bottleneck might shift back and forth depending on the project phase: sometimes it’s coming up with the right idea (PM-heavy), other times it’s executing a tricky technical detail (engineer-heavy). we won’t literally see teams with 2 PMs per engineer across the board. still have a 5:1 or 4:1 ratio bottleneck might shift It’s also worth noting that not all product management scales linearly with engineering output. There’s a certain amount of user research, design insight, strategic thinking and yes, human judgment that goes into great product management. AI can assist but not fully replace those human elements – at least not yet. So demand for strong product thinkers will indeed grow. As one LinkedIn commenter noted in the debate: even using AI daily, it “still needs a lot of input throughout the process to provide anything that has true and specific value... It's not quite there yet for it to truly understand users.”l In other words, AI can help summarize and speed up tasks, but it can’t (currently) truly empathize with a user’s pain point or have the eureka moment of a killer feature. That’s human PM territory. human judgment “still needs a lot of input throughout the process to provide anything that has true and specific value... It's not quite there yet for it to truly understand users.” l Meanwhile, engineers aren’t going away either – but their roles are evolving. They might become more like technical directors or editors, guiding AI-generated code, integrating systems, and tackling the hard problems that AI can’t. One might envision a future where an “engineer” oversees multiple AI coding agents – effectively managing higher output – while a PM oversees multiple product directions or experiments simultaneously. In that scenario, the team structure becomes more fluid. technical directors or editors team structure becomes more fluid Ng’s extreme ratio example might not become the norm everywhere, but it signals a direction: the relative importance of pure coding vs. product thinking is shifting. His provocation is probably meant to wake up companies to invest in product strategy and not just throw more coders at a problem. And on that, he’s likely correct. As he quipped, those who still ban AI coding tools in their companies need to get over it – the future is coming, whether we’re ready or not. the relative importance of pure coding vs. product thinking is shifting More PMs, More Code, More Questions Andrew Ng has essentially thrown down the gauntlet, predicting a world where product managers are far more numerous – and arguably more critical – than today. It’s a world where code is cheap, “vibes” are the new specs, and the best ideas (or the best prompts) win. Will we truly see two PMs for every engineer in the average startup? Color me skeptical, but not dismissive. We might see new hybrid roles and AI-wrangling “prompt engineers” bridging the gap. We might see teams oscillate between needing heavy product input and heavy engineering input at different times. We’ll almost certainly see a lot of awkward experiments (and failed ones) as companies figure out the right mix. two PMs for every engineer new hybrid roles But as someone who’s watched tech trends come and go (and come back again) – from the browser wars to the social media boom – I’ll say this: never bet against a paradigm shift when AI is involved. A year ago, few imagined that non-coders could build functioning apps via natural language. Now it’s happening. So if Andrew Ng envisions a future with PMs swarming, it’s at least a future we should prepare for, even if just to ensure we don’t end up in that “PM hell” engineers fear. never bet against a paradigm shift when AI is involved In the end, whether you’re Team PM or Team Eng, the smart move is to embrace the tools and level up your game. If you’re a developer, get comfy with AI coding assistants and maybe polish those product sense skills. If you’re a product manager, better sharpen your technical chops and learn to harness AI in your workflow. As Ng says, “everyone should learn to code” (yes, even the PMs). And everyone – coders and PMs alike – should learn to work with AI, because it’s changing the rules of the game. embrace the tools and level up your game “everyone should learn to code” So, will the classic “one PM to 7 engineers” rule of thumb go the way of the fax machine? Possibly. At the very least, we’ll talk about product team composition in a very different way in five years. Kara Swisher’s take: It’s about time we re-thought how we build products. If AI is automating the grunt work, let’s refocus human talent where it really counts – creative, strategic, empathetic product leadership and high-level engineering ingenuity. Whether that means more PMs, fewer PMs, or simply smarter PMs and engineers, one thing is clear from this debate: “vibe coding” is changing the vibe of software teams, and none of us can afford to ignore it. “one PM to 7 engineers” It’s about time we re-thought how we build products and smarter “vibe coding” is changing the vibe of software teams, and none of us can afford to ignore it. Watch the talk in full: Watch the talk in full: https://www.youtube.com/watch?v=RNJCfif1dPY&embedable=true https://www.youtube.com/watch?v=RNJCfif1dPY&embedable=true https://www.youtube.com/watch?v=RNJCfif1dPY&embedable=true&transcript=true&video=false https://www.youtube.com/watch?v=RNJCfif1dPY&embedable=true&transcript=true&video=false