Podcast Summary: Inside Datumo: The Startup Taking on Scale AI
Joe Rogan Experience for AI
Date: August 17, 2025
Episode Overview
This episode dives deep into Datumo, a Seoul-based startup that recently raised $15.5 million and is directly challenging the data-labeling giant Scale AI. The host explores Datumo’s origins, unique business model, growth trajectory, and emerging competition dynamics following the major Meta–Scale AI deal. The conversation also addresses larger industry concerns about AI safety, benchmarking, and the use of licensed datasets, offering listeners a comprehensive look at the evolving landscape of AI data infrastructure.
Key Discussion Points & Insights
1. The Scale AI–Meta Deal as a Backdrop
- Industry Context: The host begins by highlighting the “drama” and massive potential in the AI data industry, sparked by Meta's $14.3 billion investment/acquisition of Scale AI.
- “It is an absolutely massive market as we’ve seen with Scale and Meta doing their, you know, $14 billion deal there.” (01:04)
- Customer Fallout: Major clients like OpenAI left Scale AI post-Meta-deal due to conflicts of interest—emphasizing the volatility and opportunity in the space.
2. Datumo’s Company Origins & Business Model
- Startup Roots: Datumo started as a data-labeling company, founded by David Kim, ex-AI researcher for Korea’s Agency for Defense Development.
- Pain Point: Slow, labor-intensive data labeling prompted Kim to create a reward-based crowdsourcing app, allowing anyone to label data in their spare time for payment.
- “If you have free time, you can sit there and label data in your spare time and you get paid for it. I thought this was kind of funny, but it’s obviously a huge business.” (03:13)
- Early Traction: Even before the app was built, Datumo secured significant pre-contract sales and crossed $1M revenue in its first year.
3. Datumo’s Growth and Clientele
- Current Status: Seven years in, Datumo boasts 300+ clients, $6 million revenue last year, and around 150 employees in Korea.
- Major Clients: Well-known names such as Samsung, LG Electronics, Hyundai, Naver, and SK Telecom.
- “A lot of big companies in Korea are basically their main clientele and I imagine they'll try to push and expand a little bit more to America in the coming years...” (05:42)
- Global Expansion: New funding will fuel expansion into Japan, the US, and further R&D.
4. Service Diversification: AI Model Benchmarking
- Moving Up the Stack: Initially focused on data labeling, Datumo evolved to offer model evaluation and benchmarking services.
- Co-founder Michael Huang: “They wanted us to score AI model outputs to compare them to other models. That's when we realized we were already doing model evaluation without even knowing it. We started in data annotation and then expanded into pre training data sets and evaluations as the LLM ecosystem matured.” (07:10)
- Full-Stack Tooling: Launch of “Dadu Mo Eval”—a no-code evaluation tool designed for compliance, safety, and non-technical teams.
5. Licensed Data Sets and Competitive Differentiators
- Unique Datasets: Datumo licenses its own datasets, including data curated from published books, positioned as enhancing human-like reasoning in AI models.
- “Apparently reading books is a good way for [AI models] to…learn how to reason through problems, which I thought was absolutely fascinating.” (10:44)
- Ease of Use: The platform enables non-developers—policy, trust, and safety teams—to benchmark and evaluate AI models without technical know-how.
6. Funding Journey and Salesforce Ventures
- Networking Success: The host shares the unusual way Datumo connected with Salesforce Ventures.
- “Their CEO said that they previously hosted a fireside chat with Andrew Ng who's the founder of DeepLearning AI...they shared that event on LinkedIn and from that someone at Salesforce Ventures actually saw it and then...the whole funding process took about eight months.” (12:04)
- New Funds Usage: Accelerate R&D, automation in evaluation, and drive international growth.
7. Industry Takeaway & Host's Perspective
- Massive Opportunity: The confluence of big tech deals, customer churn, and rising concerns about responsible AI creates a fertile ground for new entrants like Datumo.
- “Right, Scale AI taking on $14 billion. This is obviously a very big industry. So excited to see what they do with it and how they're able to grow.” (15:49)
Notable Quotes & Moments
-
On AI Safety Concerns (02:01):
“Most companies...say that they're not prepared to use AI in a safe and responsible way...40% of the people that took that McKinsey survey said that they view it as a significant risk.” -
On Industry Turmoil Post-Meta Deal (08:26):
“A lot of the big customers of Scale AI pulled out, OpenAI being one of them. OpenAI actually stopped using Scale AI after the whole Meta deal because they're like, well, Meta's our competitor and, you know, how can we trust it?” -
On Benchmarking Pivot (07:10):
Michael Huang:
“They wanted us to score AI model outputs to compare them to other models. That's when we realized we were already doing model evaluation without even knowing it.” -
On Successful Fundraising via Networking (12:04):
“Hosting some big famous person and then posting it on LinkedIn is like the best way for a startup to raise money.” -
On Datumo’s Unique Data (10:44):
“Apparently reading books is a good way for [AI models] to…learn how to reason through problems, which I thought was absolutely fascinating.”
Important Timestamps
- 01:04 – Opening thoughts on market potential and the Scale AI-Meta deal
- 03:13 – Datumo’s reward-based labeling app and crowdsourcing origins
- 05:42 – Current revenue and major Korean clients
- 07:10 – Michael Huang on expanding into AI benchmarking
- 08:26 – Industry drama: Meta’s deal & OpenAI leaving Scale AI
- 10:44 – Datumo's approach to licensed data from books
- 12:04 – Salesforce Ventures connection via LinkedIn and the funding journey
- 15:49 – Host's bullish outlook on Datumo and data labeling industry
Conclusion
This episode paints a comprehensive picture of Datumo as a fast-growing, innovative competitor to Scale AI with unique offerings in both data labeling and AI evaluation. The host ties together global industry moves, competitive responses, and the entrepreneurial hustle behind Datumo’s journey—showing how even a Korean startup can find international traction and play a major role in the next generation of AI infrastructure.