Jul 21, 2025 · 1h 7m · 20vc
Surge CEO & Co-Founder, Edwin Chen: Scaling to $1BN+ in Revenue with NO Funding · 20VC with Harry Stebbings
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 deep-dive interview on the 20VC podcast, Surge CEO and Co-Founder Edwin Chen explains his counter-cultural philosophies on building a highly profitable, bootstrapped business, maintaining extreme talent density, and why high-quality human training data is the ultimate bottleneck to achieving Artificial General Intelligence.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 22.1% of the talking time here. How this is scored →
Harry as informed peer 3.6 Guest teaching 5.0 Guest disagreement 3.4 Harry pushing back 2.4
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 1:00:51 Dismissing 50% AI-generated code claims
Edwin directly dismisses statistics cited by Harry from tech leaders like Benioff and Vlad, arguing that generating 50% of code with AI only applies if 90% of your engineers are writing meaningless features.
Hardest push from Harry ▶ 10:42 Interrogating the 'body shop' competitor labelHarry refuses to accept Edwin's sweeping label of competitors as body shops, directly challenging him on what he means and noting industry claims about labor dynamics.
Biggest teaching moment ▶ 45:20 Exposing LMSYS Arena leaderboard gamingEdwin dismantles popular public LLM benchmarks by explaining how models game leaderboards using emojis and verbosity, highlighting a top model that incorrectly claimed Pope Francis was still alive.
Harry holds his own ▶ 52:07 Citing vertical AI investments and architectureHarry demonstrates deep venture expertise by contrasting generalist monolithic models with specialized code models, citing his portfolio investment in Poolside alongside competitors Cursor and Windsurf.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| The Big Tech Myth: Useless Problems and Internal Machine | 2 | 4 | 2 | 1 | Harry sets up the discussion by citing Edwin's quote about Big Tech inefficiency. Edwin explains how 90% of employees at mega-tech firms work on internal machinery rather than customer value. | |
| Inside Big Tech: Empire Builders & Internal Machinery | 2 | 4 | 2 | 0 | Harry asks how to spot empire-building managers versus true doers. Edwin details the contrast in interview questions asked by product-focused doers versus careerist managers. | |
| Meeting Policy: Ruthless Elimination of Standing Meetings | 3 | 5 | 2 | 1 | Harry references Shopify's no-meeting policy and asks about 100x engineers. Edwin breaks down the multiplicative math behind 100x productivity and how AI compounds top-tier talent. | |
| Why Most Competitors are "Body Shops Masquerading as Tech Companies" | 4 | 6 | 6 | 5 | Edwin attacks competitors as body shops masquerading as tech companies. Harry pushes back on the labor camp characterization, prompting Edwin to explain algorithmic quality control. | |
| The Founding of Surge: Moving Beyond Craigslist & Excel | 2 | 5 | 2 | 0 | Edwin shares his origin story at Twitter where labeling sentiment data via Craigslist and Excel failed due to poor tooling and lack of nuance. | |
| Rejecting the Status Game: Why Surge Didn't Raise Money | 2 | 6 | 7 | 1 | Edwin delivers a sharp critique of Silicon Valley's fundraising status game, arguing founders raise money for TechCrunch headlines rather than conviction. | |
| The Importance of Unique Ideas and Finding Genuine Self-Worth | 3 | 4 | 3 | 3 | Harry challenges whether founder uniqueness matters or if success is purely about execution. Edwin insists generational companies require personal alignment and unique insights. | |
| Scaling Surge: Managing Initial Demand and Visceral Data Understanding | 3 | 4 | 2 | 2 | Edwin contrasts Surge's focus on deep data inspection with traditional companies that treat data annotation as mere administrative labor like drawing bounding boxes. | |
| Staying Bootstrapped and Profitable: Why Surge Avoided Sales Teams | 4 | 4 | 3 | 3 | Harry asks how Edwin avoids building a faster horse for early clients. Edwin responds that staying bootstrapped allowed Surge to decline misaligned customers. | |
| The Negative Incentive of Headcount Growth vs. Revenue Per Head | 5 | 4 | 2 | 3 | Harry highlights Big Tech layoffs and the shift toward revenue-per-head metrics. Edwin criticizes headcount inflation as a vanity metric driven by corporate incentives. | |
| Scaling to $1 Billion: ChatGPT as an Inflection Point and Outperforming Legacy Competitors | 5 | 5 | 4 | 2 | Harry probes the customer migration following Scale AI's market movements. Edwin explains how AI labs switched to Surge after getting frustrated with low quality legacy providers. | |
| Unsalable and Mission-Driven: Rejecting Multi-Billion Dollar Acquisition Offers | 4 | 6 | 5 | 4 | Edwin explicitly rejects hypothetical acquisition offers of $30B or $100B, asserting independence. He also criticizes relying solely on CS PhDs who write poor code. | |
| The Advanced Algorithms of Data Quality and Project Speed | 4 | 7 | 4 | 2 | Edwin ranks data quality as the single biggest bottleneck in AI and educates Harry on how LMSYS Arena leaderboards are gamed using formatting, emojis, and verbose responses. | |
| Benchmarks vs. Real-World Use Cases & the High-Intensity Culture of XAI | 4 | 5 | 3 | 3 | Harry presses on Grok's benchmark success. Edwin explains that benchmark gains often mirror solving SAT problems rather than solving real-world engineering challenges. | |
| The Limits of Synthetic Data and the Threat of Model Collapse | 3 | 6 | 3 | 2 | Edwin details the limits of synthetic data, explaining how models collapse into narrow distributions and output bizarre errors like random Hindi characters. | |
| Generalist Monolithic Models vs. Specialized Domain Models | 5 | 4 | 1 | 2 | Harry showcases knowledge of vertical AI bets like Poolside, Cursor, and Windsurf. In response to personal weakness questions, Edwin admits total ignorance of EBITDA and corporate finance. | |
| The Work Ethic Myth: Working Smart vs. Working Hard | 4 | 5 | 6 | 5 | Harry forcefully asserts that building a $10B+ company requires a 7-day workweek. Edwin disagrees, warning against confusing raw hours worked with value creation. | |
| The Timeline to AGI and Why Universal Chat Interfaces Will Prevail | 6 | 6 | 6 | 4 | Harry cites claims from Robinhood and Salesforce that 50% of their code is AI-generated. Edwin dismisses this, arguing AI only writes code for low-value, trivial features. | |
| The Multi-AGI Future and Advice for Day One Founders | 4 | 5 | 2 | 3 | Harry asks whether new AGI entrants can overcome capital constraints. Edwin predicts new AGI labs will emerge due to the vast remaining headroom in AI capabilities. |
Statements from this episode (46)
Edwin Chen: Most AI data competitors are body shops masquerading as tech
“I think a lot of the other companies in our space, they're just not technology companies at the end of the day. They are either body shops or they are body shops masquerading as technology companies.”
Edwin Chen: Lean teams with 10% headcount move 10x faster
“Yeah, so I think the biggest lesson for me was that you can build a completely different kind of company with 10% of the resources and 10% of the people, but you're still moving 10 times faster and building a 10 times better product.”
Edwin Chen: Big Tech employees prioritize internal promotions over customers
“At these bigger companies, a lot of your priorities, a lot of the things that you're building, they're simply, you're simply building them to impress someone. Like, hey, I need to impress my VP. I need to impress my manager. I need to impress my director so th…”
Chen: Differentiating doers from empire builders comes down to candidate interview questions
“I think a big part of it actually just boils down to the kinds of questions they ask me. Like some people, when I interview them, they will ask really interesting questions about our product. They will brainstorm about ideas to make our product even better. Th…”
Edwin Chen holds zero one-on-one meetings with direct reports at Surge
“So like I, for example, personally, I actually have no one-on-one meetings and it's kind of funny because oftentimes people will ask me, well, How often do you meet with your reports? How often do you set aside for, like, for these meetings? I just don't have …”
Chen: Standing weekly one-on-one meetings are a negative management signal
“Like, it's almost like a negative sign if you're having a one-on-one weekly meeting, because it means that you just don't know what's going on with these people. You're not, you're like almost waiting for your weekly meeting to raise, raise interesting questio…”
Edwin Chen: Single-person $1B companies will exist due to AI
“I mean, I absolutely believe that that company was this one day. Like you think about it, like I've always believed in 10 X engineers, even a hundred X engineers. And already you have a lot of these single person startups that are already doing ten million rev…”
Edwin Chen: Multiple single-person startups currently generate $10M in revenue
“And already you have a lot of these single person startups that are already doing ten million revenue.”
Edwin Chen: 100x engineers exist through multiplicative compounding of small advantages
“Some people are simply two to three times better, like two to three times faster than anybody else, right? They just code faster. There are some people who simply have two to three times more better ideas. There are people who simply work two to three times as…”
Edwin Chen: AI tools disproportionately benefit 10x engineers over average workers
“If you don't have to spend that time on the drudgery, but you just have like these endless ideas that are just bouncing around your head and AI just helps you put them to paper. Then I do think it kind of disproportionately favors people who are already like t…”
Edwin Chen: Half of computer science graduates cannot code
“I think half of the people who graduate with a CS degree, they can't even code.”
Chen: Optimizing ML models for clicks creates negative content feedback loops
“Once you optimize for clicks, the most click-baity content starts rising up to the top. You get lots of racy content, lots of girls in bikinis, lots of listicles about 10 horrifying skin diseases, and so on.”
Chen: Built Surge AI V1 solo rather than fundraising $10M-$30M
“As opposed to feeling like I needed to go out and hire 10 engineers in order to build a product, instead of feeling like I needed to go out and fundraise, you know, 10, 20, thirty million dollars in order to hire, you know, more people. Again, I just wanted to…”
Edwin Chen: Silicon Valley fundraising is mostly a status game
“I think one of the things that's always driven me crazy about Silicon Valley is that it really is just a status game for most people. Like, People are just raising for the sake of raising. Their goal isn't to build some great product that solves an idea that t…”
Edwin Chen: 90% of startups should build an MVP before raising capital
“Yeah, for a 90% of companies, no. Like, sure, there are some companies where you actually do need a lot of capital in order to build hardware or, like, whatever it is for a couple years. Like you really need a lot of investment before, before you before you ca…”
Chen: Competitors treat AI data purely as supply, ignoring technology
“There are some companies in this space who will simply think of it as a pure supply problem, and they don't give any consideration to the technology, like both the technology, the underlying technology, like, how do you identify these people? How do you make s…”
Chen: ML engineers historically neglect inspecting their training data
“Historically, a lot of ML engineers, they kind of just don't take the time to look at the data.”
Chen: Surge was profitable from its very first month
“We were very lucky to be profitable from month one.”
Early customers shape product trajectory, so founders should select vision-aligned buyers
“One of the things that I think is actually really important is that you want customers who Especially early on. You want customers who believe in your product and not people who are simply giving you a little bit of money. Like, because your early customers wi…”
Chen: Truly important tech news reaches founders without daily Twitter monitoring
“So every now and then, if something is important enough, like maybe there's some big new product is actually really cool, or there's some really, really interesting new research paper. It'll be like big enough that even though I'm not monitoring Twitter every …”
Chen: ChatGPT served as a massive inflection point for Surge AI
“Things definitely hit an excellent point with ChatGPT because I think people just saw how Incredibly valuable human data and RHF was. So definitely chat CPT was an inflection point for us, but even before that we were, we had very strong growth.”
Chen: Surge AI was already the largest human data provider for top AI labs
“So it's interesting because I think it was an open secret where a lot of top researchers already knew who we were. And they already knew that we were the biggest and the best in the space, even though we've been pretty under the radar. And so most people were …”
Edwin Chen: AI teams are abandoning body shops for high-quality data
“I mean, so I would say I'm pretty sure that a lot of these other companies, they are Like at the end of the day, people want high quality data and they don't want to be working with body shops. And so. I think we've seen, like, a massive wave interest because,…”
Edwin Chen says he wouldn't sell Surge for $100 billion
“No, I mean, I definitely wouldn't sell for thirty billion or even a hundred billion.”
Edwin Chen considers company acquisition an admission of failure
“Because yeah, getting acquired would be really limiting. It would be this admission of failure and jumping ship because you can't make it on your own anymore.”
Chen: Surge AI has more daily working PhDs than Big Tech combined
“Like if you think of all the PhDs, even at Google or Meta or Microsoft, we have way more than all of them combined doing work for us in a single day.”
Chen: 80% of CS PhDs write poor code
“I think 80% of the computer science PhDs I know, they write shitty code because they're only good at math and algorithms.”
Chen: Throwing PhDs at AI training only teaches models to hack benchmarks
“If all you're kind of doing is throwing PhDs as a problem, all you're doing is teaching models how to hack Silly benchmarks and get good at basically the equivalent of SAT problems.”
Chen: Millions of people work on Surge's AI platform
“We have hundreds of thousands, millions of people working on our platform.”
Chen: Surge runs up to 10,000 projects weekly
“You have a thousand projects, like 10,000 projects that are literally running in any given week.”
Edwin Chen: Data Quality Is the Primary Bottleneck to AI Progress
“So I would definitely rank data quality first, followed by compute, followed by the IRLs.”
Edwin Chen: LMSYS Chatbot Arena is the equivalent of clickbait
“So LM Arena is this popular leaderboard of LM models, and it's basically the equivalent of clickbait.”
Edwin Chen: Easiest Way to Game Chatbot Arena Is Lengthy Responses
“And so one of the, I mean, one of the things that we've learned is that the easiest way to improve in this arena is simply to make your model responses a lot longer.”
Surge AI works closely with xAI
“We work really closely with the XAI team and it's actually just incredibly refreshing to see how they operate.”
Edwin Chen: Synthetic data makes AI models good at benchmarks, not real problems
“Synthetic data, it's made models good at synthetic problems, not, not real ones.”
Surge AI CEO: 2,000 human data points beat 10 million synthetic ones
“A lot of them tell us that even a thousand or a couple of thousand pieces of really high quality human data that we generated for them, it's actually been worth more than ten million pieces of synthetic data.”
Edwin Chen: AI models always require human supervision due to non-human reasoning
“You always need this external value system as a kind of safeguard to make sure that the models are working properly just because the models themselves have such a different set of way of thinking.”
Edwin Chen admits he does not understand standard financial terms like EBITDA
“One area where I'm not great is I'm really bad at understanding financials. So sometimes people around the company, they'll try to tell me, Like, hey, do you, have you been paying attention to our revenue numbers? Have you been paying attention to our costs? H…”
Stebbings: Building a $10B+ company requires working seven days a week
“You must work seven days a week if you want to build a ten billion dollar plus company, and the ability to put your phone on the side and not check an email does not exist anymore if you want to build a ten billion dollar plus company.”
Edwin Chen: Number of hours worked should not be confused with progress
“I think a lot of people do confuse working hard with creating value. Like again, I mean, it's maybe a trope to say, but you have to work smart and not just hard. Like, if I think about a lot of what I'm doing, oftentimes the best ideas come to me when I'm just…”
Edwin Chen: Dismissing AI safety ignores current issues like benchmark hacking
“So I think a lot of people think AI safety is overblown, but I think they ignore the paperclip maximizer problem where you have AI models that are accidentally trained towards the wrong objectives, even though this is a big problem that all the models face tod…”
Edwin Chen: AI Will Automate Average Engineers by 2028, Cure Cancer by 2038
“So I think it would be twenty-twenty-eight if you're talking about automating a job at the average engineer, and then twenty-thirty-eight if we're talking about curing cancer.”
Edwin Chen: AI Cannot Write 50% of Meaningful Enterprise Code
“I don't think models today can write 50% of the code and, you know, come up with 50% of the ideas that, that people are, that are actually going to be meaningful to your company.”
Edwin Chen: ChatGPT Has Replaced 50% of My Google Searches
“I mean, I have felt That maybe 50% of the things I used to Google search, they are replaced by ChatGPT, or they're even better with ChatGPT.”
Chen predicts future will feature multiple competing frontier AGI companies
“How I see a world where there actually will be multiple, multiple, Frontier AI companies, Frontier AGIs, just because every one of them will be able to go in a different direction. Like, you see it already, you see it today playing out already with the differe…”
Shorts cut from this episode
▶ The Problem with Big Company Bureaucracy · 20VC with Harry S (@3:50)
▶ Why Synthetic Data Is Overrated · 20VC with Harry Stebbings (@50:22)
▶ “Having a PHD isn’t enough” · 20VC with Harry Stebbings (@40:40)
▶ The Problem with Silicon Valley · 20VC with Harry Stebbings (@18:11)
▶ Why Grok 4 Is Overrated · 20VC with Harry Stebbings (@47:46)
▶ How I Founded a $1BN+ Revenue Company · 20VC with Harry Steb (@17:01)
▶ Why More Humans ≠ Better Data Quality · 20VC with Harry Steb (@12:45)
▶ “I Wouldn’t Sell For $100BN” · 20VC with Harry Stebbings (@0:00)