Sep 6, 2019 · 31m · 20vc
20VC: Scale Founder Alex Wang on How To Hire Incredible Talent Before You Are A Hot Company, Why Beating Competition Is Not As Clear Cut As Investors Believe & Why AI Is Under-Hyped Today In Terms of Total Impact
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The 20VC, host Harry Stebbings interviews Alexandr Wang, founder and CEO of Scale AI, discussing Scale's rise to a $1 billion valuation, the reality of artificial intelligence and high-quality training data, strategies for recruiting early talent, and long-term startup execution.
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 37.3% of the talking time here. How this is scored →
Harry as informed peer 3.0 Guest teaching 3.6 Guest disagreement 1.7 Harry pushing back 2.0
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 7:44 Reframing the AI hype consensus
Alex directly pushes back on Harry's question about AI hype, asserting that the technology is actually under-hyped when evaluated on total multi-decade impact across business processes.
Hardest push from Harry ▶ 11:46 Challenging core dataset utility assumptionsHarry refuses to accept that more data is always better, bringing up Founders Fund partner Aaron VanDevender's thesis on asymptotic utility drop-off to challenge Alex's core business value proposition.
Biggest teaching moment ▶ 12:11 Explaining exponential data curves and edge casesAlex educates Harry by explaining that model gains scale exponentially rather than linearly, and demonstrates that real-world AI reliability depends entirely on collecting rare edge-case data.
Harry holds his own ▶ 11:46 Citing domain experts on data utility limitsHarry demonstrates deep preparation and domain knowledge by citing specific theoretical arguments from Founders Fund regarding diminishing returns on large datasets.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome and Transition to Interview | 2 | 1 | 1 | 1 | Harry introduces Alex and asks friendly background questions about dropping out of MIT, his parents' reaction, and advice for younger people evaluating career paths. The dynamic is polite, standard interview banter with no real friction. | |
| Upbringing in Los Alamos and Cultural Impact | 2 | 3 | 2 | 1 | Alex discusses growing up in Los Alamos and how the non-commercial, knowledge-focused community shaped his company culture. When Harry asks if AI is overhyped, Alex gently reframes the premise by arguing AI is actually underhyped in terms of total long-term impact. | |
| Distinguishing True AI from Marketing Rebranding | 6 | 6 | 3 | 5 | Harry pushes back on dataset value by citing Founders Fund partner Aaron VanDevender's argument about the asymptotic diminishing marginal return of data. Alex educates Harry on exponential scaling research and explains how safety-critical real-world AI relies entirely on gathering data for rare edge cases. | |
| Evaluating Synthetic Data Platforms | 3 | 4 | 1 | 3 | Alex explains why synthetic data creation platforms fail in practice due to models picking up subtle artifacts and bias. Harry asks probing operational questions about how Alex evaluates candidate risk tolerance during early-stage hiring. | |
| Compensation Structures and Equity Allocation | 4 | 5 | 2 | 2 | Harry references Alex's tweet about competition and defends the investor view that market victories are clear cut. Alex explains why investor assumptions about 'just building a better product' are naive, arguing excellence is required across every organizational function. | |
| Fundraising Strategy and Selecting Investors | 2 | 3 | 1 | 1 | Harry asks standard fundraising questions about Scale's $100M Series C led by Founders Fund. Alex shares advice on selecting investors who are long-term believers rather than chasing brand names or fair-weather capital. | |
| Board Management and Transparency | 2 | 3 | 2 | 1 | Alex discusses board management and transparency before moving into the quickfire round. He offers a sharp critique of Silicon Valley PR hype culture, emphasizing business fundamentals over TechCrunch headlines. |
Statements from this episode (20)
Scale AI Achieves $1 Billion Valuation After $100M Series C
“With their recently announced a hundred million dollar series C, their valuation surpassed the one billion dollar mark, making their 22 year old founder one of the youngest to do so.”
Wang: People overweight the downside risk of joining or founding startups
“I adopt a very risk-seeking perspective, because the reality is that there's not actually that much risk. When you start a company, the worst case, obviously, is that it fails, but even in that case, it's not that bad, especially if you're young, because if yo…”
Wang: First-time founders should join a startup before starting their own
“I think the thing that I would do first is join a small startup. And the reason for that is a couple things. I think first, your first couple jobs really imprint you and like imprint how you think about jobs and how you think about business and how you think a…”
Alex Wang: AI is underhyped and will rival the internet's impact
“So I actually have this belief that AI is either appropriately hyped or even maybe a bit under hyped in terms of its total impact. I think that it really is, if you were to think about technology in terms of these giant waves, I think there's the internet, the…”
Alex Wang: AGI is overhyped, but existing AI already justifies hype
“In particular, I think AGI and the thought that we're going to have AI that is better than humans at everything, I think that's a bit overhyped, or at least it's unclear when that might actually happen, and some people say it's going to be sooner, some people …”
Alex Wang: Many companies are lying about using AI
“Today, a lot of companies and projects really do, they're using the term AI extremely loosely, and I think that is part of what impacts a lot of the perception of the technology, because so many people are lying about it”
Alex Wang: Scale AI counts OpenAI, Waymo, Uber, and Lyft as clients
“We at scale, we work with Very innovative companies are actually using AI and machine learning and deep learning in particular. So we work with folks like OpenAI and Waymo and Uber and Lyft and all these folks.”
Wang: Google research shows exponential dataset scaling yields continuous model gains
“Google sort of published a bunch of these papers, which showed that each time you can X the size of your data set, you're going to keep getting bang out of your buck in terms of the Performance of your model. So it's not just that it is true that like for each…”
Wang: Real-world machine learning reliability depends primarily on edge case data
“In the real world, the performance of these models, it really all comes down to edge cases, right? If you think about a self-driving car, Right now, everybody in the industry is sort of tuning and sharpening the performance against edge cases, and if you kind …”
Alex Wang: Synthetic data is nowhere close to impacting machine learning progress
“It hasn't really worked in practice in most situations, particularly when it comes to visual data or even textual data. It really hasn't worked, and a lot of the reasons why it's sort of limited in terms of its overall impact is that at the end of the day, whe…”
Wang: Convincing candidates choosing between Google and startups usually fails
“If you have to convince somebody between joining Google and your startup, you've kind of already lost a big battle, especially if they aren't looking at other small startups or they aren't even really trying to figure that out.”
Wang: Candidates asking about work hours lack startup readiness
“When candidates ask about, oh, how long do you guys work, and what is the policy there? I think the reality is that if you look at most startups, people need to realize that it is a fight for survival, and so it kind of needs to be in the spot where you're oka…”
Wang: Early startup hires should take big salary cuts for equity
“Now, I actually think that's negative. I think you do really want people who are okay to Taking a massive salary cut, and then really excited about getting more equity. And even though a lot of founders might view their equity as very precious, particularly ea…”
Alex Wang: Building a better product is almost never enough to win
“And so it's really naive to think that like, oh, we can just build a better product and then we'll win. That's basically, I think it's almost never true. I think you always need to build a better product. Yes. But you also need to realize that you need to be b…”
Wang: Scale AI Series C process with Founders Fund took under a week
“I think it was probably less than a week of due process.”
Wang: Founders should prioritize true believers over brand name when raising
“You want to optimize for the brand and you want to optimize for like the ability for them to help your business, but probably the number one indicator of that is just like, how big a believer are they in you and what you're doing?”
Wang: Building personal relationships with board members enables direct feedback
“The thing that's gone the furthest for me is just really building close personal relationships with the members of my board. And that's probably been like The number one thing that I think has helped me because I think it means that with this relationship with…”
Alex Wang Calls Seven Powers One of the Best Business Books Ever
“I really love this book, Seven Powers, by Hamilton Helmer, but it's probably one of the best business books ever, and the reason is that he sort of, it goes through Through what he called these seven powers, which are these seven qualities of businesses that a…”
Alex Wang: Early Startup Media Hype Rarely Helps Build Great Businesses
“I think particularly for early stage startups, there's a lot of focus around press and getting your name out there and being a sort of like hyped person. And I think the most important thing is to just build a good business actually, and being hyped very rarel…”
Alex Wang: Scale AI Will Build Full ML Infrastructure Suite Over 5 Years
“And so I think as we look at the next five years, we want to build out all of the pieces of infrastructure, all of the tools, all of the platforms that would allow more and more organizations to be successful, effective, and productive with machine learning so…”