Doug Leone & Diego Piacentini - Italian Tech Week 2025

Okay, this is the last time you're going to see me, I swear. You won't see me anymore. So, Doug, let's do it in Italian. Yes. He knows more than that, right?

Yes, a little bit more.

Because Doug was born in Genoa, right? Why don't you tell us one minute about yourself?

So I was born in Genoa in a very tough street called Via del Prè, which is the first street off the port, where you couldn't walk at night unless you were a protected kid. I was a protected kid, so I could go buy milk. If you showed up, you may not come out. And we emigrated to America because we were poor, basically, in 1968.

And how old were you?

11.

11. OK, good. And he has both passports, Italian and US, right?

I do.

I do. OK, so AI, dot-com era.

Yes.

We're both old enough. We've seen them both.

Well, speak for yourself.

OK, right. Do you think that, what's the difference between the AI wave versus the previous wave? You saw at least two, actually you saw three.

Yeah, I saw many waves. So I think there's two differences. One is the violence of the wave, the steepness of the slope. As waves increase, and let me take you back, semiconductor wave, systems wave, LAN waves, dot-com wave, mobile wave, AI wave. The slope of the curve is higher and higher, so they're more violent. Second, the amount of time between head fake and reality is shorter and shorter. So, just take you back to 1999, there were a couple of great companies built in 1999, you were one of them in Amazon, the other's Google, there were a lot of others, to me, that came and went. But then we had four or five years of lull, and then lots of companies. In a mobile wave, Netscape went public in 1995.

Sorry, in a mobile wave, the iTunes Store. It came out in 2008. We had all these little apps, and we didn't know how to monetize little apps. And little did we know of DoorDash, on which you're on the board, or Airbnb. That took two or three years. I think that time is compressed in the AI wave. I think the time between initial couple of companies, a couple of years of lull, or sometime a lull, and then lots of things is going to be even more compressed. So I am wildly optimistic. While I call those waves, I call AI the industrial revolution. That's my view. Now, in your case, you live through a wave and that wave is composed of high waves. The amplitude is the highest and trough. And Amazon was, if you will, the first company that decided to burn a ton of cash

and grow almost at all costs. And then the trough hit, and you had to survive. Talk a little bit about what caused you to be so aggressive at a time when all analysts thought you were crazy, lots of investors, and how you survived the trough.

Yes, and everybody loved eBay and didn't like Amazon. So I joined January 2000, and the historical context, the bubble burst in, I don't know, March, something like that. And still remember, we went from Get Big Fast to, you're going to laugh at this, MMSSGG. What does that mean? We all collected our efforts around MM, Make Money. At that time, my company was unprofitable. SS, which was Single Store, needs the context. Amazon had a marketplace separate from the retail store. We merged the two and we created the marketplace on Amazon. And the GG was touching directly my job, which is Go Global. So we did not stop the investments. But we definitely focused on, you know, we closed huge fulfillment centers.

We did this deal with Toys "R" Us and Target so that we could monetize our expertise. We did things like, oh, we like acronyms. We did something called CRAAP, which is Can't Realize a Profit. We knew that we're shipping products that would cost us more to ship than the profit of the product itself. So we really gathered all the efforts and every team was about those things here. But don't give up the growth. We decided that we would go global, we would do the deal with Toys "R" Us and Target, we would do the Single Store because if you give up, I mean, it's easy to cut costs and give up growth. So on your side, lots of AI companies, different LLMs. How are you going to differentiate? Asking the winners from the losers is very, very hard.

But how do you differentiate in this market? Do you see that as an investor?

So we have a point of view, and that is that ultimately value is going to accrue to the application layer. That has happened in the past. It doesn't mean the future is going to be like the past, but that's our point of view. So we think the LLMs are important. We're investors in OpenAI and some other companies we don't talk about, but we also think there's a way to differentiate. So I'll take a couple examples. We're investors in Harvey, legal. Think of that as a vertical type of LLM. We are investors in Open Evidence, where about half the doctors in America are now using it. So to us, specialization in vertical markets, proprietary workflow, proprietary data, trust and safety. To us, there's a million ways to start companies, get a lot closer to the customer, get a lot closer to the value, get a lot closer to the person with money.

The person with money is not using LLMs. He needs an application on top of it. And so that is, in our opinion, what you should do. And we think there is a ton of opportunities in ways we can't even begin to fathom. Now, in your case, though, how do existing companies survive this AI wave? What should they do?

Well, again, if they have to do something new from scratch, it means they've already failed. Because you're not going to change strategy just as a reaction to the fact that things are moving so fast. First of all, I think that you need to realize as a startup, are you in a land-rush market, or are you in a place where your solution can wait for perfection? My suspicion is now it's a lot of land rush, and that looked a lot like in 2000, right? Amazon went global very, very fast because they did not want others to go before. The other one is—this is something rhetorical—but do you have a solution, and are you solving a real problem, or do you have a technology waiting for a solution to pop up, or waiting for a problem to pop up?

And if you're in this case, you're in a very unprotected area. The other one is, and this goes to the Theory of Constraints, what are your bottlenecks to growth and what are your bottlenecks to profitability? And you really, really need to focus on those bottlenecks and make sure that you understand what those bottlenecks are. Is the bottleneck the technology itself? Is the bottleneck the fact that your product doesn't have a market fit and you need to work on that? So it's this amazing awareness of the business. And the other thing, which is now very different, I think, from our day, is the build length. At Amazon, we built every single component of our stacks. We built our customer service software.

We built our CRM, everything. Now, I see companies that are building their web scrapers, their syndication layer, their agent-to-agent conversations. I mean, in the last three months, I saw companies doing orchestration layers. So the build versus buy now is the key question. And you need to make sure that you're not building things that are going to be available off the shelf in two months. In your case, we always talked about the war for talent, how talent is important. So how do you build the team that can thrive in AI?

So I think I should take you through five seconds of history. 25 years ago, we would get profitable quickly. In fact, I had an argument with someone a couple of days ago in Israel. He said, "Why should we get profitable?" You ought to get profitable as quickly as you can. We wouldn't travel to Europe until the US was profitable. And then we became interconnected. And things just moved a lot faster. The cost of capital came down. And as long as you're profitable at the unit economics level, you should go as fast as you can. And things move a lot faster. If you look at OpenAI and the adoption of ChatGPT, the cost of sales of that product is zero. The danger thereby being, the cost of replacement is zero.

And so what is the option you have when you live in this new world where you can adapt very fast, you can change very fast, and people can leave you quite fast? And the answer is you've got to build product as fast as you can, and you have to have a mindset of people that can move as fast as you can. And that, at the cost of sounding kind of inappropriate, that brings you to youth, because the older people get, the more they're used to the world they were living in.

I can say that. You can't because you're younger. Yeah, I'm younger, of course.

I'm saying that. Sorry for offending you. And so you have to hire people that are willing to move very fast and you have to be able to move even when you've got imperfect data. And the lesson I learned during down markets—and I lived through the down markets of 1999 and the down market of 2008—the lesson I learned: you never take your foot off the accelerator of building product. You keep on building. Maybe you slow down sales, which is very expensive to ramp up sales, but you don't stop building product because that becomes your wall. And so during up markets, cost of capital is cheap. You run as fast as you can. During a down market, cost of capital is more expensive. You maybe pull back on sales a little bit, keep on building that wall called product, because product is the center of a company.

Now, in your case, you built a global organization. Tell us the lessons that some of these folks or all these folks can take away from the experience that you had and how to go global fast and quickly.

First of all, the question underlies one factor. If you are in AI, you cannot be a domestic company. You can start there, improve the products, test the products. But in order to scale, you need to go beyond your borders. And that's absolutely the right thing to do in that direction. So the first question to ask is—it's an organization question before a type of hiring question—which is, what part of my technology, what is the technology that I don't need to change too much to go global? Because you have to tweak, change, try to adapt, and sometimes you adapt to something that really doesn't exist because some sales guys are telling you that the clients, they are different, then you're kind of losing the battle already.

So the point is, what are all the pieces of my business that can be touched as little as possible to go global? Once you have that, and especially, you need to make sure that your technology is expandable without too many changes. Obviously, the data sources are going to be different. Talking about Harvey, right? I mean, I believe that when they go global, obviously, the data ingestion about the laws are different, but the technology is going to be the same. It's not going to change. And if you have to tweak it too much... So that's an organization part and a technology part. The other is about talent. What is the kind of people that you need to run your business outside the US? And I think that the US or Germany or the UK, if you're a European company, a lot of those companies make the mistake of hiring through credentials.

They look at the CV, they say, 'Oh, this guy's been running Airbnb in France, so he has the experience of working with a U.S. company.' And most of the time, you need to realize, if you are the country manager of Airbnb, the country manager of Uber, of all those companies, it's a sales relationship job. So do you need that? If you need that, that's perfect. But if you need people that are also about technology, they're also about marketing because the marketing strategy is different, then you need to hire based on that and not based on the fact they've been working at similar companies before. And the other point is—which is the hardest one, obviously—you always want to hire missionaries and not mercenaries.

So you need to really understand who are the people that believe in the missions you're building, and they're there for the long term. Yes, you can make mistakes. Yes, people can decide to leave anyway. But you need to identify the key players that will be part of the expedition. That's so important that you see that.

And if I can add to that, so we can go off this boring script that we were given here. I can't tell you how many times we have hired the salesman or the sales manager that could sell $15 million for big company A, say Cisco Systems of yesteryear, and couldn't close a $100,000 order for a startup. So you have to interview. That's very interesting. You have to interview from first principles. It's not just a resume. First of all, you want to stay away from people that are well-traveled. I call them the two-and-out people. They last in a company for two years. They have excuses for every job. I now have given up just listening to those two-and-out people. And then that's not enough. That's a necessary but not a sufficient condition.

And the other one is that they lost a step. Because in these big companies, they're order takers in sales or in engineering. They're not doing very much anymore. They're administrators. And so we've learned, hire for first principles. Figure out the questions for your business and go get those people.

Let me reinforce this part, and I'm gonna be self-critical towards the company I work with, which is a lot of the people leaving Amazon, they fail outside of Amazon. That's right. Because they think they're good, and then they get lost in the startups. And they were not good, they were effective. They were effective because they were working in a machine that was super organized and everybody had precise tasks, hence they were successful. And this is why I actually mentioned in a conversation before, you need to make sure that you have a culture that understands where there are good mistakes and there are bad successes. When I see a CV of a salesperson that says, "I grew the sales of my region by 85%,"

I close the CV right away. First of all, because they use the word "I." And second, because there's no causality and it's confusing correlation with causation. That's all part of that.

Wonderful.

You've seen so many companies. You see companies going down, going up. What is the pattern or is there a common identifier for the entrepreneurs that are successful?

I think it may surprise you to learn in that world, nothing's changed. This is how you build, in my opinion, a great company. First of all, you need to have a wonderful founder or founders that have a very clear vision. And you know that by how they articulate it. If they can't articulate it, they can't think it. To me, what comes out of here is connected to what's in there. And the skills you need early on are engineering and product management. Then you have to hire—the first four or five people have to be A-plus people. Because if you start with A-minus or B-plus, you'll never move up to A-plus. That's a must. If you're in B2C, you've got to have a quality product. If you're in B2B, you can afford to give your corporate customers something that's halfway broken because they'll understand you're building a new product.

So you expose your vision, you get it in there, wonderful things happen when you have that communication. Then you build that product and then you have to expand in concentric circles because the best products look like utilities. They're very simple products to understand and to adopt. Give me a company with five $100,000 or euro-type customers, then two $500,000 customers. Lot more money? No. I want customer contact, want diversity of customers and so on. The simplest product, the better it is. Then you expand in utilities. I call it utility—simple product. And one day, they wake up and they figure out they had a platform, like Amazon. They first started with books, utility, CDs, everything.

One day you bought everything. You bought your underwear at Amazon. You couldn't believe you did that. But the interesting thing is then you build a second pillar, in your case, AWS. So first you expand as much as you can, then you build a second pillar. Which brings me to my question, how the heck did you find AWS? What gave you the right to build that product? Something, a little luck must have happened because I can't imagine Jeff, out of his brain that's selling all kinds of stuff, one day says, 'You know, we need data centers.' So how did that happen?

So first of all, I was among the skepticals. I was not the only one on the S team that says, why on earth are we doing this cloud computing thing? We should be focusing on building our retail business. I was talking like a McKinsey consultant, and I was completely wrong. Jeff, and this was not a Jeff idea. This was coming from some technology team that observed how 45% of our server power were used in two months, less than that. And based on that observation and how you can actually decide to sell the other months, that's how the cloud computing idea came out. So it is the inventiveness and innovation culture at Amazon that allow that to happen. I don't think this is replicable because, I mean, especially you start up here, don't go and jump in inventing another business.

Try to do well what you're doing. But it is something that is so hard. I mean, cloud computing should have been invented by Microsoft, not by Amazon. But it was this combination of giving the freedom of bringing new ideas and obviously the observation of jumping on it. I believe there was a lot of resistance. I was one of those saying, 'This is where we're wrong. We shouldn't be doing this.' You know, I missed the $500 billion business.

We have time for one more. According to that, we have time for one more question.

So markets that are protected and markets that are in danger, or businesses that are protected and businesses that are in danger?

So I've done a little chart for myself. I think productivity tools, creative tools—now we're investors in Figma and I trust that Dylan is going to AI that company. But productivity, single-use tools, SMB tools, because the SMBs look for best-of-breed, not all different categories, I think they're in danger. If I go to the middle tier, I think the systems of record, things are protected, but I'd rather have the workflows of a ServiceNow than the thin layer of CRM of a Salesforce.com. Workday's in the middle; you figure out the other functional areas. Things that I think are protected are things that are in deep vertical markets, proprietary data, and so on. I think in that order, when I look at companies, I think of: where are you in this thing and how are you going to play it?

But conversely, what should companies do, as the last question, in order to protect themselves with this barrage of new startups and AI companies that are coming?

From a macro standpoint, it's focusing, it's understanding what are the inputs of your business to drive the growth. Too many companies, too many CEOs, too many founders are focused on the ultimate outcome without understanding what the inputs are. A quick example: in Amazon, we're not discussing, during the weekly business reviews, revenue. Yeah, you mentioned revenue, it's important, right? You mentioned revenue, profitability, but you jumped immediately on the inputs of revenue. What was the in-stock? What was our pricing compared to competition? What was the supply chain defects? So, having this great understanding of your inputs, the outcomes will come. Wonderful.