Alexandr Wang (@alexandr_wang) started Scale AI to help machine learning teams label data faster. It started as a simple API for human labor, but behind the scenes, he was tackling a much bigger problem: how to turn messy, real-world data into something AI could learn from. Today, that early idea powers a multi-hundred-million-dollar engine behind America's AI infrastructure—fueling everything from Fortune 500 workflows to real-time military planning. Just last week, Meta agreed to invest over $14 billion in Scale, valuing the company at $29 billion. Alexandr joined us on @LightconePod to share how Scale evolved from a scrappy YC startup into the backbone of some of the world's most advanced AI systems, how he thinks about competition with Chinese AI labs, and what it takes to build infrastructure that shapes the frontier. 01:15 - Alexandr’s early days at YC 07:25 - Dialing in on what worked 10:24 - Model improvements, evals 19:18 - The techno-optimist view of work 27:47 - The turning points for Scale AI 37:37 - Agentic workflows 41:55 - “Humanity’s Last Exam” 47:48 - U.S. vs China in AI and hard tech 56:57 - How to be hardcore
@ycombinator @alexandr_wang I believe prioritize building tools that don’t just automate but amplify human judgment, because messy real-world data won’t clean itself. Layering solid APIs on human-in-the-loop workflows is where the real magic starts.
@ycombinator @alexandr_wang Wow, that's amazing! I'm curious, what do you think sets Scale AI apart from other companies in the AI industry?
@ycombinator @alexandr_wang This journey is founder fuel. Want me to turn it into a landing page in 3 min?
@ycombinator @alexandr_wang This podcast is too good—I’m definitely clipping this podcast.
@ycombinator @alexandr_wang Wait is that Gary at the start He is incredibly expressive
@ycombinator @alexandr_wang Fantastic interview. I learned quite a bit. Also, thank you for the YC Startup AI event yesterday. It’s was so great to see the all the brilliant next gen thinkers and leaders gathered here together in SF.
@ycombinator @alexandr_wang That's a really inspiring story for all those who want to build in AI. Being first to build in AI comes with rewards as well as difficulties. It's difficult to predict future patterns and demands. Congratulations on the big achievement.
@ycombinator @alexandr_wang $29B from labeling tasks… respect.
@ycombinator @alexandr_wang Super cool! Transforming messy data to empower AI is a game changer. Keep pushing the limits!
@ycombinator @alexandr_wang I don't think we should compare ourselves to alexander, what he did is insanely impressive but it's a mix of right time and an outlier scenario, like the dot com bubble I feel. 99% of founders aren't the bill gates, the zuck, but wang is in that 1% for sure.
@ycombinator @alexandr_wang Gary Tan is so slimy and gross but I will watch this
@ycombinator @alexandr_wang Incredible journey. From labeling data to powering AI at national scale — true infrastructure play. @VendorApp, we’re on a similar path in vendor ops: turning messy workflows into structured, scalable systems that teams can actually build on. ⚙️📊 #AIinfra #StartupToScale
@ycombinator @alexandr_wang Incredible journey from a YC startup to AI infrastructure giant!
@ycombinator @alexandr_wang Have a look at this project- it is a real answering machine that is being livestreamed in TikTok that anyone can call and leave a message to the viewing audience @theB33P … not sure if that’s the type of real world data you are seeking though haha. theb33p.com