Eric Siu and Neil Patel break down Hamilton Helmer's 7 Powers framework (branding, switching costs, scale economies, network effects, and more) and show how Google, Microsoft, and OpenAI use them to dominate. They cover building AI fluency through hackathons and metrics, why vertical integration matters, and what separates startups that survive from those that don't.
Chapters
I'll give you guys a little fun book that you can read. There's a book called 7 Powers and it's from this guy Hamilton Helmer. And the whole book is this concept of these 7 powers, these 7 strategies, these moats that you can have. 1 is branding. 2 is process power. 3 is switching costs. And I'm going to explain each of these. 4 is scale economies. 5 is cornered resources. And 6, network economies. 7 is counterpositioning. So I'm going to start from the bottom.
Okay.
So counterpositioning might be like a performance-based pricing versus retainer. So that's different than most agencies as an example. Okay. Network effects would be like a Meta, for example. The more people that use the network, the more valuable it becomes, like a Facebook or Instagram, for example, or even a YouTube. Okay. So corner resources would be an example of, let's say you own like a copper mine or you own all the chips. Like, for example, I think 97% of the Like making these really small chips are actually, you can only do that in Taiwan, for example, like these nanometer-based chips or whatever. And then switching costs, agencies. So we can talk about this for a second, but switching costs, when you have, when you do a lot of things like in an agency and they sign a long-term contract with you and you're just kind of locked in, it's okay if you do everything at like a B. You don't have to do everything at an A+, right? You do everything at a B. Ideally, you're doing everything at an A, but it's hard. And then you have process power, like a Toyota, for example. They were— everyone, every car manufacturer tried to look at Toyota back in the day. Um, like, how do they do it? How do they pro— how are they so efficient? Right. Um, I would say Tesla kind of has that with their cars. Like, that's the thing that Elon's actually most proud of, the manufacturing piece. And then finally, you have branding, like a Coca-Cola, right? Now, when we look at agencies, um, what do we have? We might have branding. Okay. You might have like a WPP or an Omnicom. There might be branding, but I think that's kind of slowly eroding. Not yet, but it's slowly over time. I think there might be process power, but I think there's switching costs for sure, which is why they can hang on to these deals for a long time or potentially counterpositioning, but that's pretty rare in agency world, like a center field, for example. What do you think about this?
No, I'm with you, but it's hard to create that.
Yeah, it is very hard.
Really hard. That's the hardest part about it.
Yep. Yep. But you know, sometimes the hard stuff is worth it, right? It's just, but if you have an engine that's working already, you should just double down on that. So.
Did you hear about the Elon CyberCab video? I think he was talking about like the production line, like 5 seconds or 5 minutes or something crazy like that. They can start building your car soup to nuts. Um, I don't know if that clip was AI created or not. Sounds too good to be true, but it's just like, all right, that's kind of crazy to think about.
Wait, what do you— they can make a car in how long?
It was like 5 minutes or something like a CyberCab. And what he was saying is it's more like electronics company. Manufacturing, and they've just speeded everything up.
I believe it, dude. I think in a couple, maybe in 5, 10, 15 years or so, you and I, I don't know about you and I, we're not going to be the first round of this, but we're going to be flying to China in 15 minutes in a rocket.
Yeah, something like that. I've seen that.
Yeah. And that was, I think it was Gwynne's her name, Shotwell. She's the president of SpaceX. She said, this was like 5, 10 years ago that she said this, and it's going to happen. So all that to say is the world's going to be crazy. Um, and I think, speaking of moats real quick before we move on from the topic of moats, who do you think is the AI leader right now that has, uh, the best moat?
Google.
Okay, so check this out. So this is done from Battery Ventures, okay, so a VC firm. So early leaders are racing to vertically integrate. So what we're talking about, what Neil just actually mentioned, is why not offer a bunch of things? Like if you're a marketing agency, why not offer a bunch of things? You're basically You want to control as much of the stack as you can. Okay, so Google, at the application layer, they've got Gemini. Okay, they've got AI search models. They've got Gemini, Gemma infrastructure. They have Google Cloud. Okay, and then silicon, they have TPUs. So they have everything. They have their own chips. They have their own infrastructure. They have their own LLMs. They have their own applications. Microsoft is emerging. They have Copilot, the models they use OpenAI, right? This is a partial presence. That's why it's in yellow. Infrastructure, they have Azure, that's their cloud provider. And then silicon, they're making this right now, Maya 100. Okay. But you just kind of go down the list over here. AWS is here, OpenAI is here, Anthropic, Cursor. The only one that's completely vertically integrated right now is Google.
Yes. Although Google does use NVIDIA chips as well, but yes, they also have their own TPUs.
Yep. Which is, I mean, this is interesting to me because this is, End of year, I really recommend reading Battery Ventures or all these kind of state of the AI and like how Tomáš Tunguz had this predictions for next year, predictions last year. I think it's really nice to stay on top of these things to see where things are going. And I just, talking to Wade the other day, you wanna know what's interesting?
Yeah.
Okay, they started this AI fluency program and they actually, I said, how many people did you have to let go that where they weren't into this stuff. How many, what percent of staff do you think he had to let go? 700-person org.
15, 20%.
10%. So still pretty high. Okay, so 10%. Originally their AI fluency was maybe 10% or so. Once they, he said they launched a code red and he said, look, after maybe 12 months into it, our AI fluency went from 10% to 80. Now it's at 87%. Now it's at 97% actually. And so all that to say, I was like, okay, what did you do exactly? He's like, well, we launched these hackathons. I was like, what did that look like? He's like, we basically shut the entire company off for a week. Nobody answered any tickets, nothing, right? And we just focused on driving AI fluency higher. I was like, how often did you do that? He's like, every 4 to 6 months or so. So I was like, okay, obviously one week is not practical for a lot of companies. And he's like, well, you can at least do 1 to 2 days and do this every 3, 4 months or so. And I'm like, okay, that's a little more practical, right? But You know, you like, I don't know what the AI fluency level is for your company, but when I talked to one of our mutual friends, I was like, hey, what does it look like? And they have, let's just say they have 1,000 people or so. He's like, maybe 20%. And I just think that's unacceptable going into next year. And I've talked about it a lot. I just, now it's time to really, you know, make sure this is where the rubber meets the road.
Yeah, 20% is low. You want it much higher, but keep in mind, More important than just— I wouldn't say it's more important. Just as important as fluency is how quickly they can adapt. And the reason I say this is AI is moving so quick with the different platforms doing different things. People need to be able to adapt super fast because if you're great at AI right now and you're stuck in using the same, you know, tools and you don't want to change, a year from now, a lot of them could be obsolete and you could just be doing things really inefficiently.
All right. So I wanted to take a moment to tell you about my podcast co-host, Neil. Agency called NP Digital, and they work with a whole host of global companies or global organization. Also, Neil has SEO tools such as Ubersuggest and Answer the Public. All you have to do is go to npdigital.com to learn more, and we'll see you on the other side. When did you start using the internet? What year was it?
Dude, I don't know. I was a little kid. I had similar to you probably, but I think we had a similar story. I was using the Battery dial-up internet that was connected to the phone line.
So I stole my mom's credit card to sign up for MSN, and I waited for her to leave, and then I moved my computer over because I didn't have the extension for the phone cord, and I plugged it in, right? So this is 1995 or 1994. I was 8 or 9 years old. Um, and at that time, I couldn't fathom anybody not knowing how to use the internet, okay? Yet at the same time, you know, you have people that were resisting, people saying, oh, the internet's going to be a fad, you know, um, you know, Newspaper is still the way to go, like these online publications, blah blah blah. I see this as this is how a lot of people are behaving right now. This is a fad. It's going to go away. Oh, it hallucinates. It makes mistakes. All these things. And so, you know what we can do as leaders? If you're a leader listening to this right now, is how can you bring people along for the journey? And you can bring them to water, but you can't make them drink. And so that's why we're doing these office hours right now. We're doing these hackathons, and we realized with these office hours we made them optional. The first one didn't have that many people show up, and so my CTO was pissed, right? And so. Um, I don't know how you guys are doing it right now. You want to talk about your AI fluency program if you, if you have one?
I have no idea how we're doing it. I know we have one, but I have no idea how we're doing it because I'm not in charge of it.
Yeah. Well, did you do that ChatGPT thing, by the way, the end of year review? Did you see what kind of user you are?
Uh-uh. I did not.
Oh dude, you got to check it out. You have it on your phone, right?
Yeah, I do.
I say you're like a top 10 to 15% user. That's what I think you are.
No. Really?
You think you're higher than that?
Less.
Open up your ChatGPT app on your phone, and then it should have a year in review thing and then hit play. But look, I think all this to say is you don't—
And the reason I say that is I use Gemini more than I use ChatGPT.
Oh, you do? Okay. Yeah, I don't know. But all that to say is like, guys, I think speaking from 2 leaders' perspectives, I don't think we expect people to necessarily be power users, but come on, you have to help us help you at the end of the day. Because to be able to automate away the work that you don't want to do, like we were talking to one of our designers this week and he's like, dude, with the work I can do, I like, if I can focus on this, I can do the work of, you know, he's maybe exaggerating a little bit about 50 or 100 people or so. Um, and he's just, I want to automate a lot of this stuff. I'm like, so, so our CTO is like, why aren't you automating away this stuff? He's like, because, because I'm being told to do all these other things right now, which is kind of the old way of doing it. And I'm just like, dude, how do you not push back? Um, and so. All that to say is, um, if you're within the organization too, you got to push back. If you guys are doing things an antiquated way, just say, hey, why are we doing it this way? Like, why don't we try this way? And try to drive— AI fluency can't just come from the top, it's got to come from the bottom too. And one more thing I'll say about Wade is I was like, hey, who's driving a lot of this right now? He's like, actually, it's the people team. I was like, really? The people team is driving innovation with, with AI stuff? He's like, yeah. So, um, anyway, food for thought.
Yeah. And When it comes to AI, just make sure that you're helping out your team and not just putting it on them and just being like, you're expected to know everything. Like, guide them, encourage them, set up internal groups. So I do know we have little pods and the pods, call it, let's say, 5 people max. They share some of the things that they're doing that are interesting every single week on a Friday. For just 15 minutes and it helps everyone learn and it becomes more of a support group. And we've really found that that improves the percentage of people using AI to really move your KPIs. I'm not talking about efficiency and like, oh, you got more done in a time. Yes, that's cool and all, but what we really measure is KPIs. What are your key performance indicators? Are they going up and to the right because of AI?
Yeah.
And we really put people in pods and we really have them focus on helping each other for their individual or division KPIs.
Dude, and by the way, if you're a founder listening to this, so one thing that really, my CTO doesn't like it when I say this, but I'm like, dude, this stuff is so easy. Like when I open up a Zapier or a Lindy and I just dictate out what I want to build and it builds the agentic workflow. And Wade agrees with that when I asked him, because they have something very similar with Zapier. He's like, yeah, it's super easy. So how do you, You know, you as a founder or a leader listening to this right now, you might be able to build these things. The challenge is you need someone to actually sit on top of and manage it. So I don't know if it's that GTM engineer or a workflow automation specialist. You need someone that's kind of walking around the organization just asking people and almost pestering people saying, hey, like, what is, what is something you're spending a lot of time on that you're repeating every week that's taking a lot of time? And we should be thinking about every week how we can at least create some type of workflow to automate. And a lot of it's not going to work. Some of it will. But to Neil's point, You don't want a lot of this stuff to be theater too, right? Where it's just like, oh, it's just to appease the leaders to make it feel like we're using AI. That's not helpful either.
It has to be related to your KPIs. When AI first started coming off, we had a lot of people that were doing all this cool stuff. And then we fast forward 3 to 6 months and we're like, wait, we're not seeing any change in our organization.
It's theater.
Yeah, it's straight up theater.
Neil, look, I used this tool, but like, okay, what did it do?
Yes. And then it's like, look at all this cool stuff. Okay, there's no new revenue, no extra profit. Like, yeah, this is cool, but it's costing us money, dude. At an actual point, no joke, I just, I started seeing costs go up because we weren't seeing revenue go up from all these initiatives. And what we were seeing is them spending a lot of money on all these AI tools. And this is like, whoa, whoa, whoa, whoa, something needs to change here. Yeah, dude.
One thing I want to talk about too is, um, You know, I was listening to Jensen Huang talk with Joe Rogan, okay? And I think it's worth it to talk about the secret to wealth and it's the secret to guaranteed wealth really. So you listen to a Jensen talk, for example, he's like, 2 things you need to succeed. One, don't get fired by your board if you have a board, okay? Second one is don't give up. And then I look at UiPath, you look at UiPath, you know how long they took to hit $1 million ARR?
No.
10 years. 10 years.
UiPath's that old of a company?
Yeah. It took them 10 years to hit $1 million AR. Figma, how long did it take them to get product-market fit? I think it was like 7 or 8 years or something like that. So the key thing is really you want to stay alive long enough to get lucky. That's all it is ultimately. If you can just hang in there and keep compounding, if you're listening to this right now, you're definitely not a stupid person, right? And actually I read a book on human stupidity. I want to talk about that in a second too. But, um, I think Neil and I have realized that you don't, again, you don't necessarily need to be the smartest person in the room. I'm not, I'm speaking for myself really. As long as you keep going, you will get lucky enough to hit it big. And I don't, I'm not saying you're going to be a billionaire or whatever. I don't know, but you're going to do, you're going to have outsized returns that your 18-year-old self would really be proud of.
But I think it's not just doing it long enough. And yes, I don't, don't get me wrong. I do think a lot of people get lucky. You just have to also have some sort of common sense or, you know, it makes it hard if you don't. Yeah.
Do you?
Okay.
Let's talk about this. The human laws of human, the laws of human stupidity. So you'll like this, you'll like this. How would you define stupidity, Neil?
Someone not having common sense. And that's why I said it in the earlier one. I'm like, dude, I see a lot of people who go after it for 10 years and I know they'll never do well because they just don't have common sense.
Okay, so this definition is really good. You'll like this. I didn't think of it this way. So in this book, the definition of a stupid person is someone who harms other people for no gain to theirself, or actually they're detrimental to themselves. And if you think about it, there's 4 different people in life. You have stupid people, which we just talked about. You have intelligent people. So intelligent people will produce for others, they'll help others, but they'll also help themselves. Then you have bandits. Bandits will steal from you. So bandits will help themselves, but they'll steal from someone else. A helpless person is someone who helps others but can't help themselves. So for example, you have someone who likes to play hero bar that doesn't know how to delegate in your company, and then they burn themselves out. That's a helpless person. So of these 4 people who I mentioned, Neil, which one actually harms society the most? Is it the bandit? Is it the helpless person? Is it the stupid person? It's definitely not the intelligent person, but which one is it?
Okay, I need you to slow down a little bit.
I'll slow it down because I think I spoke too fast. I'm going to give it the 4 again. Okay, ready? Okay, so you have 4 people in society. You have the stupid people who, again, they harm others, but they don't help themselves or they even harm themselves. Okay, got that? Second one is you have the bandits. So bandits, like, I steal from you to benefit myself, okay? And then you lose, okay? So I win, you lose. Then you have the intelligent people. The intelligent people will help themselves and they'll help other people, okay? And then the last one is helpless people who help other people, but they don't help themselves. So which people hurt society the most?
First one is because they hurt themselves and they hurt others as well.
The biggest problem though is the You can't triangulate, you can't predict. Like with a bandit, you can predict their behavior. With a stupid person, you can't predict their behavior. And the other thing with this book too is it doesn't matter if you're a Nobel laureate or you're in a group of billionaires or deca-millionaires or whatever, there's always stupid people at every level. There's an exact fraction at every level. So it's the stupid people that hurt society, not the bandits. Because if you're in a bandit society, everyone just steals from each other. The society doesn't get better. So it's a great book.
I'll give you an example of common sense. This is what I mean in business. I was talking to a friend who has a SaaS product, okay? And he's trying to optimize his plans and pricing to get the most conversions and generate the most revenue. So I ask him, hey, how many signups do you get on your highest plan? He said 2, 3 a day. All right. I think his highest plan is like $499 a month. It's a free trial though, right? So they don't all convert. I was like, okay. I was like, what are your total signups per day? I was trying to equate the percentage. He then made a price change and then I was trying to look at the numbers. I'm like, just send me the total, the total amount of signups and the percentage that are choosing the highest plans and the trial conversion rate. And he's like, well, on a weekly basis, we get around 100 and something people who choose the highest plan. And I'm like, new signups, right? I'm not talking about existing customers. He's like, yeah, new trial signups. I'm like, so how do you go from 2, 3 a day, right? Because 2, 3 a day times 7 is not 100+ signups at the highest plan. And they're showing me data from their database. of hundreds a day. And I'm like, this is what I mean by common sense. If you as an individual can't just see there's something wrong with the data— I'm not saying you get 100+ a week or you don't. I'm not saying you get 1, 2, or 3 a day. Something is off, right? But common sense will tell you that something is off. When they send it to you and they don't question the data, this is like, oh, you lack common sense.
Got it. Anyway, I thought that was entertaining. And by the way, that book, it'll probably take you like an hour to read it. So it's a really entertaining book. In fact, I'm going to bring it next time so you can read it. It's fun. You'll like it.
Hope you guys have a happy holidays and amazing New Year's. Eric and I wish you amazing 2026.
Yeah, we'll see you in the new year. Bye.
