OpenAI's Secrets to Growth in 12 Minutes

In this conversation, we learn the fundamentals of Blitzscaling and explore some examples with Chris Yeh, co-author of the book, along with Reid Hoffman. In the next video, we will delve into the detailed strategies for company growth and the tasks to be undertaken at each stage of that growth. So, stay tuned!

Hi, I'm Chris Yeh and I'm best known as the co-author of Blitzscaling. That's a book I wrote with my friend Reid Hoffman, the co-founder of LinkedIn. It's about how companies can grow into enormous, multi-billion dollar businesses by prioritizing speed over efficiency to win a winner-take-most market.

Everyone's very interested in OpenAI. But OpenAI is also a great example of how the techniques of Blitzscaling become important for building a great company. With OpenAI, they began by just building out the product. That's wonderful. But the thing is, the number of users your product has drives product value. It played out in terms of the launch of the product, tapping into global distribution in order to grow quickly. It plays out in terms of the alliance with Microsoft: they've been able to continue to keep that product up and running and to grow, even though it's the fastest growing product of all time.

The core principle of Blitzscaling, which is to prioritize speed over efficiency in the face of uncertainty, is something that is not found in any single business school textbook. I know because I went to Harvard Business School; that was not something we were taught. Whereas in Blitzscaling, our message is no, sometimes you have to sacrifice efficiency and risk making mistakes in order to move faster.

The important thing to remember about Blitzscaling is it's not a strategy you apply in every single circumstance. It applies when you are in a very particular kind of market with a very particular kind of opportunity. The way you can tell is you can focus on two key principles.

The first is a winner-take-most market dynamic. You need to be in a market where the first company to achieve scale achieves enduring market leadership. This can come about because of network effects, elements like community, or elements like a platform network effect. If you don't build that in from the start, you won't have that winner-take-most market dynamic.

The second part is equally important: you need a distribution advantage. Your ability to grow depends on your distribution. You can think of it as a race where you're running against your competitors, trying to get to the finish line first. Your distribution is the thing that determines how quickly you can run.

Companies should really focus on either building organic virality into the product, or engineering incentivized virality instead. Dropbox is the classic example, where referring a new user earns you additional free storage. That's something you have to consider as you're designing your business model.

A lot of people ask: is Blitzscaling forever? As we've described, it only applies to certain circumstances. But it's also the case that Blitzscaling is a difficult strategy to implement. It's not like, great, let me just grow and things will magically happen. As we all know, growth is hard, and Blitzscaling involves a level of growth that is uncomfortable and risky. When you are prioritizing speed over efficiency, it means that you might be in danger of running out of money.

If you do not need to Blitzscale, it may be easier to pursue a slower growth strategy. But that doesn't work if you are actually in a Blitzscalable, winner-take-most market. While you are taking your time, another competitor may be Blitzscaling. The key is spending money in a way that gives you a better competitive advantage. Once they win that market, it doesn't matter how well you execute. You will never be able to catch up.

When a founder decides they're going to Blitzscale a company, they have to prepare themselves for uncertainty. You are trying to figure out how to grow faster than your competitors. Your competitors are also aggressive entrepreneurs, and they're also finding ways to grow quickly. As a result, the aggressiveness that's required means you have to tolerate a lot of uncertainty and a lot of risk. In order to actually do Blitzscaling, you often have to do things that are counterintuitive compared to conventional wisdom.

There are a couple of these counterintuitive examples, and I'll share one that's been very popular because it's probably one of Reid's most famous sayings: "If you're not ashamed of your first product, you've launched too late." So launch a product you're ashamed of.

Now, this is really strange, right? We normally say we want to launch a great product, a product that people love. But Reid's insight is that we have the ability to rapidly iterate and improve our products. This is a lesson Reid learned with his first company, called SocialNet. What he discovered is they spent all this time perfecting little features they thought were important. When they actually launched it, customers ignored them and never used them.

He took that lesson with him when he started LinkedIn. Right when they were about to launch, one of the things people said was, we need to stop and delay the launch because we need to build the ability for people to find consultants on LinkedIn. That sounds interesting, but Reid's answer was simple. They would launch LinkedIn now, and if people clamored for the feature, the team could go back and build it later.

That's because Reid knew that what we think is going to be essential to the user might not be. The best way to improve the product is to get real-world feedback and iterate. They launched LinkedIn, and it turned out that's not what people wanted. The number one request that came in from LinkedIn users was very simple: I want to put my picture on my LinkedIn profile.

Launching a product that embarrasses you is a way of getting over your fear of what if it's not perfect. Guess what? It's not going to be perfect. It's going to have problems. The reason is none of us are geniuses. None of us is smart enough to know exactly what the market wants. But you'd rather be fixing real problems based on real user feedback than trying to fix things based on your own intuition, which could very well be wrong.

The question I always get when it comes to Blitzscaling is: how do I prepare myself to Blitzscale? Is it just a matter of being aggressive? No. No, no, hold on. Aggression is not the point. Speed is the point. The most important speed you can have is actually speed of learning, because the world around us is constantly changing.

The thing that worked three months ago may not work today. The thing you want to train yourself to do is to be what we call an infinite learner, someone who can constantly learn new things. More importantly, you have to let go of old lessons. We all learn lessons in our lives, often from the things that help us be successful. But that doesn't mean those lessons still apply. Learning to let go of the lessons that made you successful, in order to have room for the new lessons that will make you successful in the future, is one of the most important things you can do as a Blitzscaling entrepreneur.

There are a lot of misconceptions about how startups grow. Some people think you just raise a lot of money and spend it on marketing. You can buy growth, to some extent, but you're not going to buy sustainable success. That's why it's so important, as detailed in the book The Lean Startup, for companies to find product-market fit early on. Without product-market fit, no company can achieve long-term success. If you Blitzscale before you achieve product-market fit, that's very risky. You may just be spending money to buy temporary market share, because the product doesn't yet retain its users.

The best time for a founder to implement Blitzscaling is after achieving product-market fit but before competitors begin to scale up: that's the ideal scenario. You've achieved product-market fit and you know you can benefit from throwing money into growth, but at the same time, you haven't fallen behind your competitors. If your competitors are starting to grow before they achieve product-market fit, you may need to start growing earlier. That's the judgment call. Can my competitors achieve product-market fit as they scale, or not?

The issue becomes nuanced. Say your competitor is scaling up very quickly. You have to ask yourself, is that competitor going to be able to achieve product-market fit as they grow? Because if they do, and they've grown out in front of you, you won't be able to catch up with them. However, if you think they're far from product-market fit, you may let them grow quickly and swoop in to buy up their assets once they go bankrupt. This has happened many times in history. But if achieving product-market fit occurs simultaneously with scaling up, you may need to Blitzscale even before you achieve product-market fit.

One of the most fascinating stories about why Blitzscaling is necessary is the story of Airbnb. Think back to when Airbnb was a much smaller company. It had about 30 or 40 people at the time, primarily in the United States, and that is when the Samwer brothers of Germany decided they wanted to create an Airbnb clone. They created a clone called Wimdu, funded it with $100 million, and hired 400 people to build out the company. $100 million, 400 employees; Airbnb at that point had maybe 40 employees and $10 million of funding. Suddenly, Airbnb had gone from the leading company in the field to the David versus the Goliath of Wimdu.

The Samwer brothers were also interested in merging Wimdu and Airbnb. In the end, Wimdu shareholders would own 25% of the company, Airbnb shareholders would own 75%. This was a nasty shock to the founders of Airbnb, because now there was a competitor with ten times as much money and ten times as many people going after the same market. It would have been easy to just say yes to the Samwer brothers and accept this merger agreement, which would still allow them to keep most of the value. But it would involve giving up so much of their value, and it would involve rewarding bad behavior. We don't want to set a precedent where people who just clone products get to extort people for their equity.

So what Airbnb decided to do was Blitzscale, though the term hadn't been invented yet. We have to raise more money, we have to hire more people, we have to compete head to head with Wimdu in Europe so that we can outgrow them: that's exactly what they did. They raised $100 million of their own, launched 12 different offices in Europe over the subsequent six months, and ultimately outgrew Wimdu. Wimdu went bankrupt in 2019. Airbnb, thanks to the power of Blitzscaling and its focus on growth over efficiency, was able to become the dominant company we see today.

One of the things that is really new, and that I'm going around the world talking about, is the way that AI and Blitzscaling interact. Remember, the model of Blitzscaling is that technology innovation triggers new markets and scrambles existing ones. Well, guess what? AI is doing that left and right. We are looking at this tremendous change in the marketplace. That's going to create opportunities for Blitzscaling, for business model innovation, and for entirely new companies. Reid told me he believes there will be dozens of outcomes that are $10 billion or more just in the next couple of years, as companies build out AI solutions of various kinds.

The message I want to give people is: AI is coming, it's absolutely going to have a huge impact, and it enables even more Blitzscaling on several levels. For existing companies looking to Blitzscale, AI and its ability to amplify human productivity allows you to grow more with fewer people and fewer resources, and the companies that leverage AI will be able to outscale their competitors. On the other side, AI-based companies can really tap into the principles of Blitzscaling. AI has strong winner-take-most market dynamics. A better model attracts more users, and more users generate the data needed to build a better model, which attracts even more users. There is this incredible virtuous circle that really helps existing players like OpenAI. We're going to see this play out in many different AI-based markets, and we expect the Blitzscalers being established right now to ultimately dominate the world of AI in the next couple of years.