The AI Spending Trap

Neil Patel and Eric Siu break down the rise of “token maxing,” why AI token spend without ROI is dangerous, and how companies like Anthropic, OpenAI, Google, and Microsoft are battling for AI dominance. They also cover TBPN’s X growth strategy, AI-powered advertising, enterprise AI services, and the marketing playbook behind Grüns’ $1.2B exit. A must-watch for marketers, founders, and AI operators looking to scale with smarter distribution, AI adoption, and performance marketing strategies. Key Takeaways: ⬛️ Token maxing without ROI creates dangerous incentives ⬛️ Google and Microsoft may dominate AI through distribution ⬛️ Grüns scaled to a $1.2B exit with message-match funnels Chapters: (00:00) TBPN’s X Ad Strategy (00:33) Mid-Form Content Growth (01:41) Monetizing Podcast Impressions (03:00) What Is Token Maxing? (04:08) AI Spend vs ROI Debate (05:10) Cutting AI Token Costs (06:16) Anthropic vs OpenAI (09:50) Why Distribution Wins AI (10:49) Anthropic’s $1.5B Venture (12:03) Why Services Businesses Win (13:32) OpenAI Enterprise Growth (14:21) Grüns’ $1.2B Marketing Playbook

• 0:00Speaker 00

  Finding customer information shouldn't mean digging through emails, call notes, and spreadsheets. HubSpot Starter Plan brings all of those details together, so it's easier to track what's happening with every deal and what to do next. It pulls information from calls and emails. Then it gets to work, logging and summarizing them. Right after your call ends, HubSpot pulls next steps from your call notes and drafts suggested follow-ups. And it stays up to date for you, so you can build demand, win more deals, and spend less time on repetitive work. Your records, insights, and next steps are ready for you to act on. And you don't have to miss a beat searching for any of it. Try the Starter Plan for $10 a month. Learn more at hubspot.com. Remember I was talking how TBN does the ads at the end? So this time I found the ad, so I'm gonna show it to everyone and then show kind of what's working for me on Twitter right now, and then we can move over to TokenMaxx.

• 0:44Speaker 01

  TBN's daily newsletter has op-eds that are written by Jordi and I. In addition to top tech headlines and the timeline's best posts. Sign up for free at tbpn.com.

• 0:55Speaker 00

  Okay, so I want to explain what I just played over here. So we went, we did the eBay video first. There's that interaction, and then they basically clipped that part. And this is, I call this a mid-form clip, where it's like 3 to 10 minutes or so, and it's horizontal, it's not vertical. These tend to do pretty well on Twitter or X. You see, this one has 644,000 views. So that's 644,000 impressions that you have for a post-roll over here to get people onto their newsletter, and it kind of advertises their show too. And so I think we're gonna see more of this. And I wanted to test this as an example for kind of the work that we're doing right now, right? So if you go to, I'm looking at my page right now. I think this one is, maybe it's with you as well. Not bad, okay. Look, Neil, so

• 1:39Speaker 00

  there's a, there's a video of Neil here with his kid's laptop decorating it. And then this is us talking about Anthropic as a single-growth marketer. This is from this podcast, Neil, okay? 2 minutes over here, 54,000 views. That's not bad, right?

• 1:51Speaker 02

  Yeah, it's not bad.

• 1:53Speaker 00

  Yeah, so you got you got you got two here. So my point of saying this is like you have 1100 bookmarks, 535 likes. I think there is, this was 1.7K. I'm looking here, 9, 10K over here, 2.2K. In fact, whoever wants to advertise on this podcast, the next advertiser will have to pay for these impressions as well, because we will also stack these on. So 18K on that one. Not bad, right? So my point of saying all this is that if you're trying to grow on like X, for example, the reason I think TBP and did so well is because they were selling these impressions, especially on X. And you know, you have very tech forward people listening to this stuff, right? So what do you think about this?

• 2:30Speaker 02

  It's not a bad way to monetize, but when their ad, I always saw go subscribe to their website. I didn't hear them pitch another company or product or service.

• 2:38Speaker 00

  I saw other companies, so I think it just depends on what you see. Like last time I tried to show, I couldn't find anything.

• 2:43Speaker 02

  No, no, the one you just played right now, it wasn't a pitch of another company, it was to subscribe to their own channel.

• 2:46Speaker 00

  I've seen them pitch other companies.

• 2:48Speaker 02

  Oh, I got it. You're saying they rotated up, so it's random.

• 2:52Speaker 00

  Yeah, exactly. Yeah. So we should definitely do that. You know what's funny for this podcast, Neo? I have to record the podcast for you, or the podcast ads for you is promoting your company, because you won't do it for yourself. So if you ideally do it for yourself, we can include yours as well. So.

• 3:06Speaker 02

  I can do it. Just tell me what you want me to record.

• 3:08Speaker 00

  Uber suggest. I'll just tell you right now, Uber suggests answer to public and MP Digital. Yeah, you can just make like three 30 seconds at the end. Please don't do it when you're traveling. Do it in like a nicer.

• 3:20Speaker 02

  Three thirty.

• 3:26Speaker 02

  Okay, I set myself a note. Oh, with video or no video?

• 3:31Speaker 00

  I think you should do video, because we want it to be horizontal.

• 3:34Speaker 02

  Okay.

• 3:35Speaker 00

  Cool. All right. Because I think we're gonna get a lot more views this way. Anyway, okay. So let's talk about the next topic here. So a lot of people are talking about token maxing. Okay, so you know, I've been watching some videos. Everyone's like, yeah, you should token max. What token maxing is first and foremost is spending as much as you can on tokens, okay? So here's the uncomfortable truth about token maxing. Simple execution beats clever strategy when timing is right. So what that means is if you spend on token maxing, like let's say Neil says, oh, Eric, if you work for me, you have to spend as much as possible. I would just run my tokens on stupid tasks all day. I may just have an agent just to hit that number so I don't get fired. Okay. But if you say just token maxing alone is pretty stupid because it creates perverse incentives. When you create a metric like this, you have to have a pairing metric.

• 4:20Speaker 00

  And so a pairing metric means sure, you can try to token max, but maybe every time, every, at the end of every month, you have to go to your manager and say, hey, here's all the things that we built. Here's how it helps customers or here's how it gets us customers. Cause if you don't have a pairing metric there, it's going to just drive a bunch of compute up. You're going to pay an arm and a leg and you're going to get nothing from it from your organization. Frictionless Growth Marketing hosted by Sonya Thompson is brought to you by the HubSpot Podcast Network, the audio destination for business professionals. Frictionless Growth Marketing hosted by Sonya Thompson shows how modern brands grow by removing the friction in their marketing and customer experience that pushes customers away every single day, often without leaders even realizing it. Each episode delivers practical insights, real brand examples and conversations with

• 5:04Speaker 00

  marketing leaders, showing brands how to win with the new rules of growth, identity-driven marketing, cultural relevance and frictionless customer experiences. Listen to Frictionless Growth Marketing wherever you get your podcasts.

• 5:17Speaker 02

  Dude, I totally agree. I don't understand why people would wanna just juice up their cost. Like it doesn't make any logical sense other than their bosses looking at it as a way of, did you use AI? And we can see based off of your spend. I think that's just a terrible way. They should start looking at things like what task did you actually complete? What were the results from it? What's the ROI? Did it reduce churn or increase LTV or repeat purchases? And I genuinely believe that's what a lot of the data and analytics teams will start looking at or CIO or CFO, they'll start grinding into the numbers because you can't just say in marketing, oh, we're all spending five or 10% more of our marketing costs in total because of AI. Oh, what are we getting for that?

• 6:02Speaker 02

  Who knows? But look at this spend, like people are just gonna be pissed.

• 6:06Speaker 00

  Yep. So again, I don't think we're saying, don't look at your tokens. I think we're both saying that look at your tokens absolutely, but your token spend. But you have to make sure that you're working closely with your team because if unmanaged, it's just going to go all over the place. Speaking of which, Neil, my token cost got up to basically like a $12,000 a month run rate. And I was like, oh my God, this is crazy. So what I did was I literally, all you have to do now for Neil, if you want to cut it to zero for yourself, is you're paying for the Claude Max subscription, $200 a month. You can even pay for a ChatGPT, $200 a month. The important thing is you create a long-term token, which will last for a year and use the Claude CLI command line interface for that. You literally, you can just go to your client, say, how do I make a one-year token?

• 6:48Speaker 00

  You create that token and then you just run your, whatever you're using, like an OpenClaw or whatever. And that will basically cut your cost. I cut it down to 87 cents from $400 a day, Neil, to 87 cents a day.

• 7:01Speaker 02

  Yeah, that's awesome. You shared that last week or was it the week before? Where the issue and for a few day period, it started or one day or two day period, it started rising again. There was, I think.

• 7:11Speaker 00

  It's up to 400. So I fixed that, fixed that. And then now officially it's on the Claude CLI or you can use OpenAI OaaS. I will say another thing, Neil, that's funny about this stuff is, let me just ask you right now, let's see where you're at. What do you think is like the model that's ahead in the world right now? The number one model, AI model?

• 7:29Speaker 02

  I would say it's Anthropic is winning right now, just based off of your valuation and how many people want to buy their stock and how many people in my organization use Claude versus any other company's products. Like they have access in my organization, you have access to Microsoft's products, Google's products, OpenAI's products, but yeah, majority of my employees are using Claude.

• 7:54Speaker 00

  Yep. So here's the thing, you would also agree that in the world, there's a little bit of a lag when it comes to adopting like the internet or email, things like that. Yes?

• 8:03Speaker 02

  Yes, but the lags have gotten shorter and shorter over time. So for example, when it comes to AI, the reason, or let's actually go with e-commerce. The reason it really started picking up is everyone started having these devices and it made it easier to buy. If you start looking at things like AI, well.

• 8:23Speaker 02

  You don't have to get people on the internet as or when people had to get on the internet, not a lot of people had computers. Now, everyone already has devices that plug right into the AI. So it's easier for human beings to adopt the technology at a much quicker pace.

• 8:37Speaker 00

  Yeah, so let me tell you where I'm going with this, Neil. So a couple, like very, until very recently, everything was about Claude. They were marketing very quickly. It's like they could do no wrong, right? Mythos came out, you know, the Opus four seven, amazing, right? Oh, my God, they're shipping constantly Claude, Co-Work, all this stuff. Here's the thing, Neil, in the past couple of weeks or so, because I tend to live in this world, OpenAI is slowly taking the crown back, right? Everyone now on X is talking about Codex, they're talking about ChatGPT five point five, they're talking about the slash goal commanders, all these things. And so I think it's it's pretty funny that in AI, we're very promiscuous. There's a there's a tug of war. All that matters in this world, when we're, when you're talking about trying to stay at the cutting edge of of intelligence, is the product itself, right? It like nothing matters more like I like one or two weeks, three weeks ago,

• 9:22Speaker 00

  I would just be talking about Claude all day, okay? And now I'm like, oh, man, this OpenAI has made a comeback. I will gladly switch over to it. So it's like there isn't really like a defined moat quite yet. I'm not seeing it yet.

• 9:35Speaker 02

  It's coming soon. I genuinely believe it's coming soon, and I think the winners are going to be Microsoft and Google. Because i- i- if you look at the technology right now, it's improved over time, okay? When AI first came out, it's drastically improved from then to now, right? But in the earlier days, you saw much bigger leaps in the technology. You you give it two, three, four years, the the versions that keep changing, I don't think they're going to be leaps and bounds better each new cycle. And I believe the Microsofts and the Googles are going to be like, it's included with your subscription for your company, and it's free. And I believe that they will just dominate and take over a lot of the corporate spending,

• 10:19Speaker 02

  which is where Anthropic is growing extremely fast, due to the fact that they already have the customer base and they will pay, they'll take free over pay.

• 10:26Speaker 00

  Real quick, if acquiring customers has been a struggle for you and you are trying to figure out how AEO works, Answer Engine Optimization, or the new version of SEO, how paid works, how all this AI stuff is going to play into your customer acquisition strategy, then check out my ad agency, Single Grain. It's singlegrain.com. And if it looks like it's a good fit, we'll help you with a free marketing plan. So again, go to singlegrain.com and we'll see you on the other side. Yeah, I'm trying to pull up a chart where it shows AI intelligence compounding over time, where basically it's right now at a genius level, and it's going to continue to get stronger and stronger. So I'll try to pull that up in a little bit, but let me just show you something, Neil. So we're talking about Neil talking about Microsoft and Google being

• 11:06Speaker 00

  the winners in AI. There's a big reason why that's important, because when you look at Google and you look at Microsoft, they have the most distribution, right? It's distribution that matters, especially if distribution is getting harder and harder in the AI-driven world where you can make whatever you want. You want to be able to control that distribution. And Microsoft, they have those enterprise contracts locked down, Google as well. In fact, Google kind of has the whole stack. They have the tensor processing units, they have search, they have Android, they have Gemini, they have, what else do they have, Neil? They have Gmail, they have YouTube, they just have all the distribution in the world, they have all the data in the world, and they're going to make use of that. So obviously they have Google Ads too. So, and then Microsoft, not too bad themselves, they're

• 11:47Speaker 00

  both multi-trillion dollar companies. So you were just speaking about Neil, Anthropic and OpenAI. I just want to call something out here. So you look at this here, here, I'm going to share my screen real quick. So let me share my screen. Let's go over here. So Anthropic has decided to start a 1.5, they've decided to start a 1.5 billion dollar joint venture with multiple Wall Street firms. So you got, you got Blackstone in there, you got Goldman Sachs in there, they're all major investors. And basically, this is from Wall Street Journal, Anthropic nears 1.5 billion dollar joint venture with Wall Street firms, okay? So Anthropic, Blackstone, Hellman, and Friedman, and Goldman Sachs announced the formation of a new AI services company. The organization will work with mid-sized companies across sectors to bring Claude into their most important operations.

• 12:30Speaker 00

  Applied AI engineers from Anthropic will work alongside the firm's engineering team. Whether, where? Claude can have the most impact, build custom solutions and support customers over the long term. Um. So the company is backed by just these leading companies. Oh, look, Apollo's in here, Sequoia's in here as well. These are uh general Atlantics in here. These are major firms. Okay. So why is this important? People might be thinking, oh, well, first of all, first and foremost, Anthropic and OpenAI are doing this. That means that everything still comes back to services at some point. And you might be thinking, oh, I run an agency. Now I have no chance. No. This actually reinforces that people need agencies, they need services companies to help with implementation here. So I think this is a net positive thing. It's not a negative thing, right?

• 13:12Speaker 00

  There's a lot of things what I that I see online right now where people are talking about building these single brains or company brains and all these things. And that reinforces the path that I'm on. But also, when I look at Neil's agency as well, it's not like this stuff is going away. In fact, they're gonna need it more because I I think Anthropic and and OpenAI, they're they're gonna try to go with with the their engineering resources. Are gonna focus more so on the enterprise market, right? It could be lower market enterprise or or or higher enterprise, but there's still a lot of room. There's still more customers than ever. And there's gonna be a lot more entrepreneurs, I believe, as well. So I would I would try to double down, triple down, especially if a services business.

• 13:47Speaker 02

  Yes, and uh services now, I think, is gonna explode more and more because of all the noise that's around it right now and all the announcements, right? Like, it's becoming quote unquote sexier, at least temporarily.

• 14:00Speaker 00

  Yeah, I mean, look, what what is it? Sequoia, Andreessen Horowitz, uh, why comment or talk about this? Uh, the the new one trillion dollar company is a services firm, right? Services as a software. So that's why I'm more excited. I'm excited for Neil. I'm excited for myself. This has never been a better time to do business, I believe.

• 14:17Speaker 02

  Yeah, and uh what was the amount that they all raised? OpenAI did a joint venture with someone. I forgot what the dollar.

• 14:23Speaker 00

  I forgot how much they raised. Yeah.

• 14:25Speaker 02

  But it was in the billions and I'm I'm assuming, I think Anthropic was five, right? Five billion?

• 14:30Speaker 00

  Uh, for the joint venture, one, one point five.

• 14:32Speaker 02

  Oh, one point five. Maybe the other one was five billion.

• 14:35Speaker 00

  Yeah, let me see. So OpenAI raises money for services.

• 14:37Speaker 02

  Yeah, that's right. Anthropic was one point five. The other one was, which you just covered, but the other one was five billion, I think. I could be wrong.

• 14:43Speaker 00

  The deployment company. Okay, yeah, I see it over here. So OpenAI has raised over four billion for a new joint venture focused on helping businesses adopt software. So who's in here? You got the other side. TPG, Brookfield Asset Management and Bain Capital. All the biggest, all the biggest companies, right? In terms of, uh, investment companies.

• 14:59Speaker 02

  They're spending.

• 15:00Speaker 00

  Yep. Enterprise growth, by the way. Enterprise revenue for OpenAI now makes up over forty percent of total revenue with targets to reach parity with consumer revenue by end of twenty twenty six. So I, I think we thought, like, at least, I, I was like, oh, man, OpenAI is really falling behind. And then boom, before you know it, they're back, right? Um, let's just see how that lawsuit goes.

• 15:16Speaker 02

  Yeah.

• 15:17Speaker 00

  Yeah. All right. So let's, here, I'm gonna pick another one here. I, I'm, I'm good at the screen shares, guys, when we're, when we're on the go, so I'm just gonna keep calling these out. Um, Neil, I'm gonna let you pick one. Which one do you like to pick?

• 15:30Speaker 02

  All right, so let's go with...

• 15:35Speaker 02

  Oh, I wanna do the Groon's one, right? The ad that helped Groon sell for one point two billion. Or using your hook in the sheet, this brand sold for one point two billion off of landing pages.

• 15:48Speaker 00

  Yep, here we go, Neil. So this is Groon's, this is Chad Janis or Chad Yanis. Um, dude, I'm calling this out again, Neil. Look at this. This, this mid-form video, three hundred sixty seven K views, okay? So he, the Groon's founder Chad, sold his three-year-old company Groon's Unilabor for one point two billion dollars. And here's what he did that most brands don't. Most brands send every ad to the same generic homepage. That's why they bleed cash. Chad builds an entire world around every winning ad angle. The system is you test hundreds of ads per month to find the unlock. Once an angle sticks, so gut health, energy, focus, you rebuild everything around it.

• 16:26Speaker 00

  If someone checks a gut health ad, landing page is one hundred percent gut health, a pop-up asks about their gut concerns, emails, SMS tailored to gut health. His team is thirty people. That's pretty crazy. Thirty people, five to six creators. Four to five media buyers, three on retention. Look, the media strategists or the creative strategists outnumber the media buyers, right? Three on retention. The tool is use Replu, which is a Shopify plugin. Chat says they don't charge enough. You can now build a funnel, build a new funnel today. So I guess it's a funnel builder. Winners don't just test ads, they rebuild the entire experience around each winning message. Message match is the difference between $10 million and $1.2 billion exit. I'll let you go first.

• 17:02Speaker 02

  You know, I think Grun did amazing, amazing job marketing. You know, one thing to mention is he's also ex-Unilever, so he understands marketing really well. They're one of the biggest spenders of marketing globally and he came from there, built a good product. And it wasn't hard to get that product to go viral because that category just tends to do well. He grew really fast and you know, I think it was a smart playbook.

• 17:27Speaker 00

  Yeah. So by the way, I don't know if you know this Neil, but he, when he started the company, his goal was to sell to Unilever. Because he knows their playbook, right? And the other thing, I mean, let's, I think it's always a lot of marketing stuff. We haven't seen a lot of new marketing channels in the last 10 years or so. But a lot of, we often forget about the fundamentals. If someone sees an ad, let's say it's a green colored ad talking about, you know, your gut health. And then they land on a landing page that's orange and it talks about sleep. That's, it's not going to be aligned and you, you lose trust that way, right? And, and people have very short attention spans. So whether you use something like a Replu for Shopify or on the B2B side right now, we still have Carrot, K A R R O T. Carrot allows you to personalize your LinkedIn ads and then the landing page is also personalized as well.

• 18:11Speaker 00

  Cause you want those things to look relatively the same. And in this case, we kind of personalize the company name, the name and things like that. So. But I think this is smart. Good job on him. I think he get right off into the sunset. And sometimes when you work at a company, you know their strategy and you're focused on it. And, and look, only 30 people, this again reinforces that you don't, the rise of these, these small, small teams, big businesses. You also look at perplexity, only 30 people on the team.

• 18:38Speaker 02

  Is it only 30 people at perplexity?

• 18:40Speaker 00

  Yeah.

• 18:40Speaker 02

  I didn't know. I didn't, I did not know that. But one thing about Groon's is I'm pretty sure, and when I say pretty sure, if I had it, I'm guessing. I don't have any inside information.

• 18:50Speaker 00

  Oh, no, I'm wrong. It says for fifteen hundred employees.

• 18:54Speaker 02

  That sounds about right with complexity. Cause it's an AI company. I'm like, that just sounds too low. But with Groon's, that one point two billion number is the flash number, like what they go out with the release with. Majority of these deals, what happens is you get X amount of dollars upfront. The rest is hit on earnouts based on the company performing and hitting, you know, A, B, and C milestones. That's typically how almost all acquisitions work.

• 19:19Speaker 00

  Yeah, I'm looking at it right now. So here's a breakdown of the structure. These reports suggest it's a cash acquisition designed to scale the brand under. So yeah, there's probably a mix of multiple things.

• 19:30Speaker 02

  Yeah, it's, it's usually not simple. Like here's a one point two billion dollars. Cause then people can just leave and then you're not, you don't have the founding team and a lot of people are just like, I got my check, I'm done. So they incentivize incentivize not by just having you sign a piece of paper saying that you'll stay for two or three years. Incentivize on which you need to keep pushing harder to grow this thing. And that's how you'll unlock more money. If not, you won't get the full one point two.

• 19:53Speaker 00

  With that being said, yeah, we'll talk to you guys later. Bye.