How Edwin Chen Built a $1B+ ARR AI Company in 5 years without any Investors
The invisible genius who quietly humiliated Silicon Valley's golden boy
I met this brilliant AI founder who bootstrapped to $1B+ ARR in 5 years with 110 employees and $0 VC funding.
He is now a contender for #1 on the Lean AI Leaderboard.
Meet Edwin Chen, the invisible genius who quietly humiliated Silicon Valley's golden boy, Alexandr Wang.
While Scale AI's CEO was doing victory laps on every podcast, bragging about Meta's $14.3 B investment, Edwin was silently building something bigger (with 10x fewer people).
The numbers tell the story:
Scale AI: $870 M revenue, 1,000+ employees, $1.5 B raised, not profitable (losing $150M/year)
Edwin’s Surge AI: $1B+ revenue, 110 employees, $0 raised, highly profitable
(Privately, he mentioned that he hit much more than $33M revenue/employee and would easily be #1 on the leaderboard, but refused to share more details.)
This billion-dollar company began with just Edwin and his laptop. He built the first version in just one month in his apartment.
While Scale was burning through hundreds of millions, Edwin was profitable from day one.
But this wasn't beginner's luck or a fluke. His advantage came from years of frustration watching tech giants fail at something basic.
Every company he worked for - Facebook, Google, Twitter - suffered from the exact same issue.
These tech giants were terrible at getting quality data for their AI models.
The breaking point came during his time at Twitter. His team had to label 10,000 tweets for a project they were building.
It took their internal platform 3 months.
When the data finally arrived, Edwin was horrified. It was complete junk. Obvious slang, memes, and hashtags, were completely mislabeled. The quality was so poor that it was essentially unusable.
Bigger projects — like changing the Timeline’s optimization objectives from clicks to human-aligned notions of quality were impossible.
This painful experience became the foundation of his billion-dollar insight. Edwin realized that while these companies had massive resources, they fundamentally lacked understanding of how to build quality data systems.
In 2020, he quit his comfortable job at Twitter and decided to tackle this issue himself.
While every competitor was fighting over brain-dead tasks, Edwin went the opposite way.
He targeted the complex stuff and the hard problems. The work that actually requires intelligence and sophisticated thinking.
His insight was brilliant, and he explained it perfectly with an analogy: You could ask a 10-year-old and you could ask Hemingway to draw a bounding box around a car. And like Hemingway's not going to outperform the 10-year-old by that much.
But if you were saying like, write me a poem of a moon that makes me cry, like I would expect Hemingway to be better than a 10-year-old.
This went beyond intellectual interest, though Edwin admitted he found simple labeling tasks boring. He wanted to build a defensible business with higher margins.
His thought process was clear: Human intelligence is just like a more complex problem that hasn’t been commoditized. And so, as with any industry where you have commoditized inputs and non-commoditized inputs, the latter is just a better business to be in.
Instead of 30-second tasks that anyone could do, Edwin focused on work that took 30 minutes to hours to complete. Work that required genuine expertise and sophisticated thinking.
Customers immediately noticed the difference between Surge and everything else in the market.
Within 12 months of launch, his team of just 2-3 people was generating 8 figures in revenue. Today, he generates more revenue than the supposed industry leader with 10x fewer people.
His growth strategy is dead simple: Be so good that customers can't function without you.
That's it. No complex playbook or growth hacks.
And it worked like magic.
The results speak for themselves.
Researchers move between labs, and the first thing they say is: We need to get Surge here or we're not doing anything.
This has been a major growth engine for them, as people from one lab or team move to a new lab or team within the company.
This word-of-mouth expansion itself was so powerful that Edwin never needed traditional sales, marketing, or ads.
The quality difference was stark. Customers would tell Edwin: "Your data is just so amazing, it's so much better than everybody else's."
One customer perfectly captured Surge's appeal: "Surge is boring in the best possible way, they just do a good job. I don't have to worry about seeing Edwin in the headlines."
The contrast with Scale could not be clearer.
Scale's data quality is some of the lowest, and it's an industry open secret.
Edwin bluntly said, "Scale.ai was a failed company”. They focused on the hype. They focused on the short-term PR instead of actually building a good long-term product.
Edwin believed Scale's problems stemmed from their origins: "You can't outrun your origin as a company. Scale's origin is in this commodity vision problem... if you're not built from the ground up for that problem, you're not going to be successful."
While Scale raised billions and hired thousands, Edwin saw them as essentially "a marketing company" rather than a technology company. "They were just pure human body operations," he said.
The celebrity CEO approach also rubbed Edwin the wrong way. While Alex Wang became a Silicon Valley celebrity posting constantly on social media and doing podcasts, Edwin deliberately stayed invisible.
Instead of actually building a good product, Edwin thinks they focused on the hype.
In his mind, Scale had admitted defeat and lost when Alex took the deal, because the company was instantly gutted, and it would never be able to help push us towards AGI.
This broader pattern of style over substance is exactly why Edwin thinks the industry is moving in the wrong direction.
Bad data is everywhere, distracting from real AGI progress. Then there's LMSYS, encouraging researchers to optimize for flashy responses and benchmarkmaxx, instead of genuine quality and usefulness.
Edwin is particularly passionate about this issue, calling LMSYS a scourge on the industry.
He thinks it's a very click-baity report, because people aren't spending any time even reading the responses. They're just spending 10 seconds looking at:
- Which response is more impressive to them?
- Which response has more emojis?
- Which response has the most votes?
- Which response is long?
- Which response just feels like vibey in like an intelligent sense?
This creates a negative spiral where frontier labs explicitly ask companies like Surge: "What can you do to make it climb LMSYS?"
Edwin refuses these requests because he has the luxury of only working on streams that he believes will advance AGI. He hates being involved in something that appeals only to the most superficial and uncreative creators in the world.
He is deeply passionate about AGI and quality data, and I could sense that rare and genuine commitment throughout our conversation.
He is a rare breed of OG founders who are technologists at heart and genuinely committed to solving extremely difficult problems.
Unlike today's new-wave founders who hop between problems and chase PMF just to pitch VCs for quick exits, Edwin focused on what matters.
When asked about his inspirations, Edwin mentioned he's more inspired by Einstein than Mark Zuckerberg. His goal is to push the frontier of human knowledge forward and build a highly successful business as a byproduct of that.
This philosophical approach extends to his philosophy of advice-giving. When asked if he had advisors or mentors, Edwin said: "I don't do advice. Just first principles."
He believes in disregarding conventional wisdom and thinking critically about problems from first principles, rather than following the crowd.
This is why his philosophy about talent and growth is completely different.
Rather than trying to motivate people with equity or the promise of future riches, Edwin focuses on rewarding excellence with current profits.
"I absolutely believe in 10x engineers and yeah, 10x engineers... We don't want them to join because of money. We want to reward them because of money," he said.
The company's profitability allows them to be creative with compensation. They can do buybacks or dividends for top performers because they're not beholden to investor preferences.
This strategy comes from his firsthand experience with how large companies mismanage talent.
He applies insights from them to his own hiring. For example, at big tech, 20% of people are doing 80% of the work. The problem is that big tech systemically does not know who that 20% are.
Edwin's advantage is that he can identify and specifically target those high performers. He hires those 20% and that allows Surge to stay lean while attracting and retaining top talent through superior compensation and meaningful work.
At the end of our conversation, I asked Edwin if he would take $30B from Meta tomorrow. His response revealed everything about his priorities.
"There's almost no reason we would want to get acquired. We get to work with all the labs. We get to do all the research we want. We already make a lot of money. But why would we do this?"
Edwin genuinely believes that Surge's current setup gives him everything he wants: the freedom to work on meaningful problems, collaborate with the best researchers, and make substantial money, all while maintaining complete control over the company's direction.
He sees acquisition as potentially counterproductive to his mission of advancing AGI through quality data.
He also shared some honest advice for founders, though he prefaces it with characteristic directness: