Surge AI built a $25B company on philosophy while competitors sold features

How one data labeling company escaped the commodity trap by acting differently, then messaging differently.

TL;DR

• There are nearly 1,000 data labeling companies. Most of them describe themselves as "data factories" or "annotation solutions." Surge AI now describes itself as guiding AGI with the “richness of human experience: curious, witty, imaginative.”
• In 2021, Surge launched with messaging that sounded like every competitor: "Data Labeling for the Richness of Language. Stunning AI depends on stunning data." Now their homepage opens with Hemingway, Kahlo, and von Neumann.
• Surge grew to $1.2B in revenue without a single dollar of venture capital [1]. Their messaging evolved alongside that trajectory: from features in 2021, to social proof in 2024, to philosophy in 2025. 

Most data labeling companies sound similar.

• Labelbox: "The data factory for AI teams."
• Snorkel AI: "The AI Data Research Lab."
• Sama: "Data Certainty Starts Here."
• Appen: "Off-the-Shelf AI Training Datasets."
• Scale AI: "Breakthrough AI from Data to Deployment."

There are 56 human data companies on Albert Chun's viral LinkedIn list [2]. Probably closer to 1,000 if you count everyone. When you scroll through their websites, you'll notice nearly identical language for data pipelines, training datasets, annotation services, and quality at scale.

Then there's Surge AI.

Their homepage doesn't mention data labeling at all. It opens with a question: "What made Hemingway, Kahlo, and von Neumann extraordinary?" 

And it ends with a mission statement that sounds like it belongs to a philosophy department: "Our mission is to raise AGI with the richness of human experience: curious, witty, imaginative, and full of breathtaking brilliance. Humanity's children, sculpted by science and art."

That. is. damn. bold.

The sea of sameness.

Data labeling is the unsexy backbone of AI. Before a model like ChatGPT can generate poetry or write code, humans must manually tell it what "good" looks like. It's tedious work. Here's how some of the top players position themselves::

Labelbox: "The data factory for AI teams. Labelbox delivers innovative services and software to operate, build, or staff your modern AI data factory."

‍

Snorkel AI: "Shaping how AI learns through better data. The AI Data Research Lab. Snorkel develops the datasets, benchmarks, and evaluation methods that help AI and agentic systems learn, adapt, and perform in the real world."

‍

Sama: " Data Certainty Starts Here. Annotation, validation, and evaluation data services that convert your raw data into high-quality training data."

‍

Notice the pattern? Every company leads with process. What they do. How they do it. The mechanics of data annotation. It's like an architect describing themselves as a “designer of load-bearing enclosures for human occupancy.” 

Technically accurate. Not very memorable (in a good way, at least).

Surge's first era: in the same sea of sameness.

Edwin Chen founded Surge AI in May 2020. His frustration was legitimate. At Google and Facebook, he spent years waiting months for training data that arrived riddled with errors [3]. He saw 30% of Google's widely-used GoEmotions dataset was mislabeled [4]. So he built something different.

The initial homepage in 2021 reflected the technical origin story.

‍

Surge led with: "Data Labeling for the Richness of Language." [5]

The full pitch continued: "Build powerful NLP datasets using Surge AI's global data labeling workforce and platform. Stunning AI depends on stunning data.” 

The messaging focused on features. 

• "Build powerful NLP datasets." 
• "Global data labeling workforce and platform." 
• "Workforce and tools, together at last." 

In a nascent space with few technically superior products, Surge reached millions in revenue and became cash flow positive by 2022 [6]. But Chen wasn't satisfied.

The positioning shift nobody expected.

With dozens of competitors popping up in the space, Surge started transforming its messaging in 2024.

[IMAGE: Surge AI homepage - November 2024]

‍

The new homepage led with: "Surge AI Powers the World's Leading LLMs." Notice the shift. Not "we help companies with data labeling." Not "we provide annotation services." Instead: "We power the world's leading LLMs." That's a much bolder claim. While others say they operate data factories, Surge says they power frontier models. Similar products, different confidence levels.

But the current messaging goes further. Much further.

‍

Now Surge opens with literary and mathematical giants. The copy asks what made Hemingway, Kahlo, and von Neumann extraordinary. 

It answers: "Their life experiences: war, love, triumph, loss. The people they met, the cities they explored, the thousand choices that made them who they were."

Then comes the leap: "Data does for AI what life does for humans."

No feature lists. 

No process explanations. 

No mention of annotation pipelines or quality metrics. 

Just a philosophical claim about the nature of intelligence itself.

Why this works.

The standard advice in B2B marketing is to lead with benefits. Tell customers what problem you solve. Show them the ROI. Surge ignored all of it.

Here's what they understood that competitors didn't: when you sell the same thing as 999 other companies, benefits become meaningless. 

Every data labeling company promises quality. 

Everyone promises speed. 

Everyone promises scale. 

All of these claims cancel each other out.

Edwin Chen said it directly in his interviews: "Scale's data quality is some of the lowest, and it's an industry open secret" [7]. But he didn't try to out-feature them. He out-positioned them by changing the conversation entirely.

When Surge talks about Hemingway and Kahlo, they're not describing data labeling. They're describing a philosophy of what AI should become. They're asking customers to believe that the data training their models is an act of parenting, not factory work.

That philosophy justifies premium pricing. Surge charges up to 10x more than competitors [8]. Customers pay because they're buying into a vision, not a commodity service.

The results speak.

The numbers tell the story. Surge built exceptional products and charged premium prices from the start.

In 2024, Surge reported over $1 billion in revenue [1]. Scale AI, which raised over $1.3 billion in venture capital, reported $870 million [9]. Surge did it with zero external funding and around 121 employees [10]. Scale AI had thousands.

One former Google researcher recalled calling Chen on a Saturday night in May 2023 when Google's Gemini models were struggling. After a two-hour conversation, Google signed a contract worth over $100 million per year [8]. The researcher's explanation: "You feel like you're paying for quality in one case versus paying for man hours."

That's the positioning at work. Surge isn't selling labor. They're selling the difference between AI that cures cancer and AI optimized for clicks. The philosophical messaging came later, but the philosophy was always there.

The pattern emerges.

Surge's messaging evolution follows the same trajectory we saw with Clay's homepage journey. Three distinct levels of positioning maturity:

Level 1: Feature-first (2021)

"Data Labeling for the Richness of Language. Build powerful NLP datasets using Surge AI's So, I don't know. All right. All right. All right. All right. All right. All right. All right. All right. Oh.global data labeling workforce and platform."

This describes what the product does and how it does it. Workforce. Platform. Tools. Features.

Level 2: Social proof and trust (2024)

"Surge AI Powers the World's Leading LLMs. Trusted by the world's top Enterprises, LLM Labs, Startups & Researchers."

This lets customers validate through association. If OpenAI trusts them, maybe you should too. It's effective but still reactive. The company defines itself through others.

Level 3: Philosophy and mission (2025)

"What made Hemingway, Kahlo, and von Neumann extraordinary?"

This transcends the product category entirely. Surge isn't competing with other data labeling companies anymore. They're competing with ideas about what AI should become. They've manufactured a category where they're the obvious leader.

The contrast.

Look at the competitive set again with fresh eyes.

Others: "The data factory for AI teams."

Surge: "Our mission is to raise AGI with the richness of human experience."

One company describes a process. The other describes a purpose.

When a buyer evaluates data labeling vendors, most get stuck in a price comparison. But Surge enters a values conversation. Different buying process. Different budget. Different relationship.

Chen himself explains his philosophy about what differentiates Surge: "AGI will not be born from a pile of wrong data. We need to solve truly complex problems and achieve high-quality data code input" [13]. He's not selling annotation services. He's selling a seat at the table of building something transformative.

What founders can learn.

Most startups get stuck at Level 1. They describe their product's features because that's what they've built. They talk about what makes them technically different without asking whether customers care about those differences.

The progression Surge demonstrates is available to any company willing to ask harder questions:

• What do we actually believe about the future of our industry?
• If that future arrives, what role do we play in it?
• How would we describe that role to someone who has never heard of our product category?

Surge could have competed on annotation accuracy percentages. They could have published benchmarks. They could have written case studies about faster turnaround times. Instead, they wrote about Hemingway.

It's counterintuitive. It feels risky. It worked.

The question you should ask.

Look at your messaging right now. Does it sound like everyone else or does it sound like Surge?

If your positioning can be swapped with three competitors without anyone noticing, you're competing on features. You're in a commodity trap. And no amount of marketing spend will fix it.

The way out isn't just better copy. It's a harder question: 

What do you believe about the future that your competitors don't? 

And are you brave enough to build your entire positioning around that belief?

Surge didn't just find product-market fit. They found philosophy-market fit. From the beginning, they believed that the companies building AGI wanted partners who took the work as seriously as they did. They charged premium prices. They hired expert annotators. They turned down commodity work. The messaging evolved over time to reflect that underlying conviction.

The next time you're delighted by Claude's emotional tone or impressed by ChatGPT's reasoning, remember that somewhere behind those responses are thousands of humans who shaped those capabilities through careful feedback. Surge's bet was that those humans, and the philosophy guiding them, determine whether AI becomes Hemingway or spam.

That's a bet you can build a company around.

And apparently, a $25 billion one.