Smart Supply by AI Digital is the supply intelligence layer built for programmatic buyers. Powered by AI and 450+ media experts, it scores and gates every bid request across 12 SSPs before your DSP bids, so inventory arrives optimized for your KPI, with every AI decision visible. Learn more at aidigital.com.

[00:00:00] Marcus Johnson: Behind every bid is a supply decision most buyers never see. That's why they turn to Smart Supply, AI Digital Supply Intelligence solution, scoring and gating every path before the DSP bids, so inventory arrives optimized for your KPI. Learn more at aidigital.com.

[00:00:22] Marcus Johnson: Hey gang, it's Friday, November 6th. Nate Palmer, and this is What's Behind the Numbers? An eMarketer podcast made possible by AI and digital. I'm Marcus. To join me for today's conversation, we have joining us from, he's still in New York, principal AI analyst, Nate Elliott.

[00:00:38] Nate Elliott: Hi, Marcus. Good to be here.

[00:00:40] Marcus Johnson: And joining us from New York, uh, as well, consumer insights analyst, Paola Flores-Marquez.

[00:00:46] Marcus Johnson: Hi. Hello.

[00:00:46] Paola Flores-Marquez: Hello. Thank you for having me back.

[00:00:48] Marcus Johnson: Of course, of course. Thank you for being here. Today's fact

[00:00:55] Marcus Johnson: Who invented the modern staff notation system? [00:01:00] Basically, the way sheet music looks.

[00:01:02] Paola Flores-Marquez: Oh.

[00:01:02] Marcus Johnson: So what I found, the modern staff notation system was invented by Guido of Arezzo, an Italian Benedictine monk and music theorist around 1025 AD. Um, almost every source actually said it was this chap. Uh, by the late Renaissance, uh, Renaissance, the five-line staff became the consensus standard.

[00:01:23] Marcus Johnson: So he invented one with four, but they added a fifth, uh, for most vocal and instrumental music across Europe. So the fifth one got added then. Uh, during the Renaissance and Baroque periods, uh, as keyboard instruments, think organ, harpsichord, later the piano, became popular. The grand staff became standardized.

[00:01:43] Marcus Johnson: That's the two sets of lines stacked on top of each other. Mm-hmm. And that's where it came from.

[00:01:48] Paola Flores-Marquez: Hmm. Can

[00:01:48] Nate Elliott: you read music? And what's the top set of lines? What's the bottom set of lines?

[00:01:53] Marcus Johnson: The top set is the treble... You mean like the G clef? What's it,

[00:01:57] Nate Elliott: treble and bass?

[00:01:58] Marcus Johnson: Treble and bass or G and F [00:02:00] clef, yeah.

[00:02:00] Nate Elliott: Okay.

[00:02:01] Nate Elliott: Mm-hmm.

[00:02:01] Marcus Johnson: Yep. Um,

[00:02:03] Nate Elliott: so- How I used to be able to read music. Do you?

[00:02:05] Paola Flores-Marquez: I tried to learn the flute when I was, like, 12, and very quickly was like, "I- this is not coming easily to me. I'm bored. I'm leaving."

[00:02:12] Marcus Johnson: She's out.

[00:02:13] Nate Elliott: She's out. I played so many instruments when I was in school, and now can play none of them.

[00:02:18] Paola Flores-Marquez: Hmm.

[00:02:18] Nate Elliott: No. And the world's probably better off for that.

[00:02:22] Marcus Johnson: Perhaps. Uh, anyway, today's real topic, the big three questions for OpenAI.

[00:02:34] Marcus Johnson: All right, we typically cover the big tech giants, Google, Meta, Amazon, Netflix, sometimes Spotify. We thought it would be time to add OpenAI to the mix. We're also doing a Walmart one, so the Reimagined Retail Show will have that. Um, so here we go. Big three questions for OpenAI. Nate, I'll start with you.

[00:02:49] Marcus Johnson: What's one of yours?

[00:02:50] Nate Elliott: Yeah, I think my first question is which of the moves that they're making in 2026 are in the long-term interest of their [00:03:00] business, and which are simply to prepare for an upcoming IPO?

[00:03:04] Marcus Johnson: Hmm. Uh, why now? Why this question?

[00:03:07] Nate Elliott: Well, I, I, I was thinking about their recent announcement that, uh, they would be launching advertising into a couple dozen more countries.

[00:03:17] Nate Elliott: Uh, that brings up the number of countries in which they are selling advertising to somewhere in the low millions. Uh, and, uh, all of this from something that, uh, only a few months ago they were still labeling as an ad trial or an ad test. It seems to me they're racing full speed ahead, and of course, part of that is they're spending a lot of money, and they want as many opportunities to make some money back as possible.

[00:03:42] Nate Elliott: But it also strikes me that if they were really looking after the long-term best interest of the business rather than rushing out an untested, unproven product into as many markets as possible, they would put some of those resources into actually testing and learning and iterating on that product, uh, in [00:04:00] a couple of markets.

[00:04:01] Nate Elliott: Um, they're not doing that. Uh, and you know, it, it, it's not beyond my recognition that, uh, being in more countries will look better on an S1 than saying we did lots of testing and iteration on those ads in the US would look in the S1.

[00:04:20] Marcus Johnson: Yeah. Um, do they have time to though? They're, uh, what they should do and what they're f- financially, um, able to do, they, they might be at odds with each other, right?

[00:04:33] Marcus Johnson: 'Cause to your point, they need money and they need it quickly, and so do you think their kind of hand is, um, they're being forced here to a certain extent in terms of rush- kind of rushing to, to lack of a better word, the, the, through their advertising strategy?

[00:04:45] Nate Elliott: Well, I, I think about it this way. Um, the, you know, the, the 20 most recent markets that they've launched into, uh, the combined total advertising markets in those 20 countries is probably a fraction of the total advertising [00:05:00] market in the US.

[00:05:01] Marcus Johnson: Mm-hmm.

[00:05:01] Nate Elliott: Does it make more sense to launch into all of those markets or to do some testing and learning and iteration and develop some new ad formats and placements and opportunities, uh, add some additional targeting capabilities and measurement capabilities so that advertisers could test and learn and optimize for themselves, and ultimately driving up the value of the ads that they sell in their larger markets?

[00:05:26] Nate Elliott: Mm-hmm. Basically, is it worth being in 20 tiny markets instead of making 50 or 100% more potentially in some of your bigger markets by offering a better product there?

[00:05:36] Marcus Johnson: Yeah. Um, one of the ones I had just related to this, which was how much can OpenAI realistically expect to make in advertising by 2030?

[00:05:48] Marcus Johnson: So their CFO, Sarah Friar, recently told investors the company's ad business is approaching a $1 billion run rate. The company estimates it will reach 2.5 billion in ad revenue this year, [00:06:00] but they're also- Run rate ... expecting to surpass... Yeah, run rate. They're also expecting to surpass 100 billion by 2030.

[00:06:08] Marcus Johnson: We estimate the entire US chatbot ad market will be just 5.5 billion, so under 6 by 2030. And America, as we've just been talking about, OpenAI's largest market, so expanding into new markets, which is fast doing to what Nate was saying, likely won't contribute nearly as much as, as they're hoping for. Um, so that was one of my big questions is where can they hope to, to get in terms of this $100 billion goal given where they are today?

[00:06:35] Marcus Johnson: I looked... Barclays seems to think that they can get close-ish maybe. Nate, I don't know if you saw this data from Barclays from, uh, October of this year. They think that $25 billion in US ad revenues for OpenAI is on the table, 25 billion by 2028. They forecast 2 billion this year, jumping by nearly, uh, to nearly 11 [00:07:00] next, and then 25 billion the year after.

[00:07:02] Nate Elliott: Barclays is saying that for the US or for global?

[00:07:05] Marcus Johnson: US OpenAI ad revenues, yeah.

[00:07:07] Nate Elliott: Okay. If they get to 25 billion in US ad revenues in 2028, then they'll blow well past $100 billion globally in 2030. I've barely talked to anyone in the industry who thinks that 100 billion is even remotely achievable. Mm-hmm.

[00:07:25] Nate Elliott: Uh, I would wanna double check those Barclays numbers, uh, because- They surprise

[00:07:29] Marcus Johnson: me.

[00:07:29] Nate Elliott: Yeah, I mean, if you look at, uh, Meta, they're making, I think, 40 something percent of their ad revenue in the US. If you look at Google, they're making nearly 50% of their ad revenue in the US. So, you know, if OpenAI is talking about getting to $100 billion in global revenues in 2030, we can assume that they would be between $40 and $50 billion in the US at that point.

[00:07:50] Nate Elliott: Yeah. That's in 2030. If Barclays is saying they can get to 25 in the US two years before that, then presumably, you know, they'd be at nearly 100 [00:08:00] billion just in the US. But again, I don't know anyone who thinks 100 billion globally is very achievable. On the table. Mm-hmm. And our numbers, we don't have a global number, but as you said, our, our entire US chatbot ad market in 2030 is a tiny fraction of that goal.

[00:08:18] Marcus Johnson: Yeah. Yeah. Uh, Pal, what kind of questions do you have? What you got for us?

[00:08:22] Paola Flores-Marquez: So my first one is, is OpenAI doing enough to protect its youngest users? Hmm. Um, we've spent a number of years talking about the negative effects of social media on children. We're seeing a lot of legislation coming out, and I don't think we've fully grappled with where, where OpenAI is headed, or where AI in general is headed when it comes to falling into the same pitfalls, right?

[00:08:42] Paola Flores-Marquez: Is it going to be proactive about how it approaches the situation- Hmm ... or is it going to react re- like, in, as things occur? And chatbots introduce a very different set of problems than social media, right? Because we're talking about more of a simulated one-on-one relationship. And [00:09:00] so-

[00:09:00] Nate Elliott: Mm-hmm ...

[00:09:00] Paola Flores-Marquez: it, um, raises a lot more questions.

[00:09:03] Paola Flores-Marquez: And even though OpenAI has introduced some additional safeguards for younger users, there is some recent testing by Common Sense Media that questions how effective those protections actually are. Um, an investigation from earlier in the year in October, not earlier, from last month , found that, um, it kind of failed when it came to suggesting crisis hotline recommendations, um, recommending specific medical or mental health support.

[00:09:31] Paola Flores-Marquez: There was a decline after they installed the safety regulations . Um, and there are, like, zero alerts on newly created parent-linked accounts during conversations about self-harm. And so, um, how are they going to approach this in a way that builds trust in the long term if they are trying to sort of ingratiate themselves into these spaces to become a reliable daily activity?

[00:09:55] Nate Elliott: Mm-hmm. I wanna know the answer to that.

[00:09:58] Paola Flores-Marquez: I do too. Yeah, it's- We're [00:10:00] gonna find out together.

[00:10:01] Marcus Johnson: Yeah. It's a good one. Now, I was reading that, um- Between models 4.0 and 5, they had taken steps to reduce the kind of sycophant fancy, um, component of its model. So, you know, saying, responding to everything with, "That's a brilliant idea.

[00:10:20] Marcus Johnson: You absolutely should." Uh, so they have seemed to take some steps towards, uh, making it more, uh, I don't know, safe, I guess. Um, but speaking to children in particular is interesting, Pal, given the recent headlines we've seen around, around meta and social media addiction. So yeah, curious to see what they do, um, to help alleviate concerns in that area.

[00:10:46] Paola Flores-Marquez: Mm-hmm.

[00:10:46] Nate Elliott: This is- I think the, the knack that these companies seem to have for firing their own safety researchers probably isn't helping anyone's confidence in them- Mm ... reducing harm.

[00:10:57] Paola Flores-Marquez: No.

[00:10:59] Nate Elliott: Going back to

[00:10:59] Paola Flores-Marquez: the other- Oh, [00:11:00] I was just gonna point out that this is definitely a Gen Alpha concern more than a Gen Z confirm 'cause-- concern 'cause I think Gen Z kind of has some sort of a grasp- Yeah

[00:11:08] Paola Flores-Marquez: of how dangerous these situations can be. And obviously it can, they can still fall. We can all fall privy to it, but, like- Yeah ... Gen Alpha is growing up alongside these models, and yeah. And

[00:11:18] Marcus Johnson: for folks l- listening, uh, how do we define, uh, Gen-- What's the cutoff between those two?

[00:11:23] Paola Flores-Marquez: So in '20-- Well, this year, I should say.

[00:11:26] Paola Flores-Marquez: I'm already thinking in 2027 terms. But this year, um, Gen Alpha will be between the ages of two to 13. So next year, the oldest will be 14. Okay. Wow. Um, and Gen Z in 2026 is between the ages of 14 and 29. So they'll- Okay ... they'll be turning 30 next year. So that's a whole- Yeah ... they're exiting the youth market in their own way.

[00:11:45] Marcus Johnson: Wow. Yeah.

[00:11:46] Nate Elliott: Okay. And my favorite fact from our AI user adoption forecast is that it's Gen Alpha and not boomers who are least likely to use AI, uh, because about half of them have not yet learned how to read and [00:12:00] write.

[00:12:00] Marcus Johnson: Aw. Yeah, that's a good one. Um, Nate, what you were saying though about safety researchers, this was leading into one of my one, which is, will anyone who leaves an AI company to speak out about the risks actually be taken seriously, or is the AI train moving too fast for any voice to slow it down?

[00:12:18] Marcus Johnson: So, uh, Huo Jinghan, uh, of NPR was explaining that in early October, three former OpenAI employees alleged OpenAI fired them over pretexts and punished them for either being outspoken about safety or working with outside researchers. They also expressed worries the company may walk back a recent safety commitment.

[00:12:37] Marcus Johnson: I think it's the one they're talking about from September 12th, where CEO Sam Altman said, uh, they would follow rival Anthropic's, uh, in expanding access to third-party evaluators who assess the safety of AI systems and practices of developers. Um, OpenAI has refuted the three former, uh, employees' allegations.

[00:12:53] Marcus Johnson: But then there was also David Robinson, uh, who led the writing of safety reports that accompanied the, uh, ChatGPT developers' product [00:13:00] releases. He also quit release, uh, recently, warning its culture was broken and that AI firms were not being nearly careful enough, that's a quote, about developing the technology.

[00:13:10] Marcus Johnson: And then Dan Milmo of the British newspaper The Guardian, uh, pointing out that these departures followed the resignation, we talked about this in the podcast, of Jacob Coxon, researcher at OpenAI's rival Anthropic, who quit the Claude chatbot developer last month. He said he believed AI could kill us all by the end of the decade, that's a quote, and was backed up by an Anthropic employee warning that there was more of a 10% chance that AI would wipe out humanity within the next decade Is there a voice that's gonna speak out with people that are gonna actually take seriously?

[00:13:37] Nate Elliott: I'm not sure who the people are that you're hoping will take these voices seriously, but there seem to be, you know, maybe a half dozen people who are in a position to do anything about this. So- Hmm ... you know, we're talking about the CEOs of OpenAI and Anthropic, maybe of Google and Meta and [00:14:00] Microsoft, and, uh, you know, a couple of voices at the very top of, uh, of the US government or the European government.

[00:14:07] Nate Elliott: Probably the European government. They seem to have- Mm-hmm ... a better handle on introducing responsible legislation to try to pull back some of the worst excesses of the tech industry. Yeah. But

[00:14:17] Paola Flores-Marquez: yeah,

[00:14:17] Nate Elliott: that- I mean, that- that's, that's the list, right? Yeah. It's, it's not about whether you- anyone on this podcast takes them seriously.

[00:14:25] Nate Elliott: I suspect most of us do. It's, it's about whether anyone who's in a position to make change takes them seriously, and the fact that the companies themselves are, uh, as I said and as you pointed out, firing lots of safety researchers, tells me that the CEOs of these companies may not be likely to, to, uh, to take these voices seriously.

[00:14:47] Nate Elliott: Mm-hmm.

[00:14:47] Paola Flores-Marquez: I think they're also feeling bolstered because there are cultural attitudes that are changing. I know anecdotally, uh, the, a couple years before this one, I saw a lot more pushback on social media, particularly in [00:15:00] Mexico around AI and environmental concerns and all of that, and, like, very harsh, like, no AI-generated content.

[00:15:06] Paola Flores-Marquez: And this year, it's picking up to the point where people have been sort of subverting it and incorporating AI-generated content with, like, content creators in a way that it feels like they're embracing it, um, in a way that was- Hmm ... kind of unheard of e- even a year ago. And so there seems- Yeah ... to be more of a cultural and social acceptance that which can sort of, I think, decrease the ability to regulate it more.

[00:15:29] Marcus Johnson: Yeah.

[00:15:29] Paola Flores-Marquez: Or the desire to.

[00:15:31] Marcus Johnson: Yeah. Um, Nate, what else do you have for us?

[00:15:35] Nate Elliott: So my, my second question and, and perhaps, uh, the most existential one that I would ask of OpenAI is, is it worth you being in the consumer market? Uh, November 30th of this year will mark the four-year anniversary of OpenAI releasing ChatGPT, and, uh, I on a regular basis [00:16:00] sit down and think about whether if they had a chance to go back and only act as an enterprise-facing AI model and vendor, would they choose that path instead of going to the consumer market?

[00:16:15] Nate Elliott: Mm. Uh, the, the compute cost, the, uh, all of the costs, electricity, data centers, chips, compute capacity, staff, all of the costs that they have are so much higher because they serve both the consumer and the enterprise market, and the consumer market doesn't look like it's anywhere near the point of offering them a path to profitability.

[00:16:42] Nate Elliott: I mean, we've already talked about the fact that their incredibly ambitious, uh, advertising sales goals are very unlikely to be reached. Even if they reach those goals, there's some question about whether they would be able to make enough money to pay for the, uh, resources that [00:17:00] consumers are using. In the meantime, their nearest rival, Anthropic, uh, is, uh, filing to go public and, and, uh, folks are estimating it will be at about a $1.5 trillion valuation, which seems to be a higher valuation than OpenAI despite the fact- Mm

[00:17:16] Nate Elliott: that ChatGPT is many times bigger than Claude is- Yeah ... despite the fact that Claude has never sold a single advertising impression. So Claude, much smaller than ChatGPT. Anthropic, though, focusing exclusively or almost exclusively on the enterprise market seems to have built what Wall Street and investors think is a stronger business opportunity.

[00:17:39] Nate Elliott: Mm. I, I just, I, I have to wonder if, you know, if Sam Altman could rewind the clock four years, would he still launch ChatGPT to the public, or would he focus on the enterprise market?

[00:17:50] Marcus Johnson: Sorry, if you're OpenAI, do you think that it was more that Google surprised them in terms of they went into the consumer space thinking they can make something here, and Google was just so dominant [00:18:00] in, to what you were saying on a previous episode, acquiring or accumulating so many users in different, um, uh, AI, um, uh, tools, and that surprised them?

[00:18:11] Marcus Johnson: Or do you think that they just didn't see the opportunity that Anthropic's been able to carve out in the business world?

[00:18:18] Nate Elliott: The startup AI developers were all racing to get something ready for market as quickly as possible, and that's, you know, it's almost the opposite of Google. Google and- DeepMind or where the transformer was created, and they sort of slow-played their hand.

[00:18:34] Nate Elliott: But that's, that's always been a Google move, right? Mm-hmm. Um, watch others go to market first, prove the viability of the marketplace, and then come in a little bit late with a tool that's maybe as good, maybe a little better, maybe a little worse, uh, and leverage your, you know, your reach- Mm-hmm ... your distribution, and all of the data you have on users, uh, to take leadership.

[00:18:56] Nate Elliott: That seems- Yeah ... to be what's happening in consumer AI. Um, but [00:19:00] aside from Google, where all of this started, uh, all the other tools, you know, all the other companies, whether we're talking about OpenAI or Anthropic or Perplexity or anyone else, they all seem to be racing something out the door as quickly as possible.

[00:19:12] Marcus Johnson: Mm-hmm. Yeah. That's a really interesting one. Would they, yeah, change course if they had the chance to go back in the time machine four years? Paol, what else do you have for us?

[00:19:20] Paola Flores-Marquez: I mean, I think it's an extension of this conversation already, which is can OpenAI make ChatGPT indispensable beyond a chatbot, right?

[00:19:27] Paola Flores-Marquez: Like, Google has so much more reach. It has the advantage of bringing AI to where consumers are, and OpenAI has to convince consumers to come to it. And so- Yeah ... it's one thing for, to trust, like, ChatGPT to give you suggestions on what to do. It's quite another to trust it to make the purchase for you, to, like, make the decision for you.

[00:19:45] Paola Flores-Marquez: And so I think it's also where what we're seeing is that people have established habits, right? They go to Spotify for their music. They go to Amazon for their shopping, right? And so how do you make this piece, like, something integral to those habits and that [00:20:00] process as opposed to just an add-on, which is what it is right now.

[00:20:03] Paola Flores-Marquez: It's just an- Mm ... additional tool that we use.

[00:20:05] Nate Elliott: Yeah.

[00:20:06] Paola Flores-Marquez: Um, I do think that they have a bit of an advantage when it comes to millennials because millennials surprisingly are big users when it comes to using it for, like, shopping-related tasks according to our lovely primary research con- that Nate organized.

[00:20:20] Paola Flores-Marquez: Um, but it is an uphill battle, right? When we're talking about generations that have kind of grown up in these digital spaces and are, are sort of cementing their habits.

[00:20:30] Marcus Johnson: This is interesting one. Do we think-- I mean, are they relying on, to your point, are they relying on a younger generation who maybe haven't got, had some of those habits, uh, as, as firmly cemented?

[00:20:43] Marcus Johnson: Are they hoping those older generations can form new ones just centered around ChatGPT? 'Cause it's an interesting question, you know. Like, if I can use almost the same service, why wouldn't I go to one that I'm familiar with? Why wouldn't I go to one where, you know, I know that how to use their maps, their video platform, [00:21:00] their email client, all of that stuff?

[00:21:02] Marcus Johnson: So, Eddie, what do you think-

[00:21:04] Nate Elliott: And, and that has the contextual data that will make its perform- even if it doesn't perform- Yeah ... well on the benchmarks and, and, you know, those numbers go back and forth between vendors. But even if Google's models aren't as good as some of the other models, they have enough contextual data from people's previous maps usage- Mm-hmm

[00:21:20] Nate Elliott: and their Gmail and their- Right ... search and their browser and their Android phones that, uh, that the responses are a lot better suited to individual users in many cases.

[00:21:31] Paola Flores-Marquez: And there's- Mm-hmm ... so much more trust there because those people already have a relationship with it, whereas this is a new player, you're not really certain, right?

[00:21:38] Paola Flores-Marquez: Like, one of the biggest hurdles about building out the advertising spaces in there is gonna be, like, how do you get people to trust that, like, your recommended product isn't just something that's being paid to be recommended, and it's actually something that you genuinely, that is, is what you need.

[00:21:51] Marcus Johnson: I, I mean, going back to, this is a great one, but going back, Nate, to kind of, 'cause they are linked to what you were just saying, is there a world where in two years, [00:22:00] OpenAI has pivoted, and they're like, "We're not doing consumer anymore"?

[00:22:02] Marcus Johnson: Like, do you think that's, are they too far gone down that path to wind it back or slow the investment on that side and say, "We're gonna take our chances in the business world"?

[00:22:11] Nate Elliott: So the, their biggest challenge is, right now at least, they need whatever revenue they can get, to a point you made earlier. And, you know, we can debate whether the best way to get advertising revenue was to launch into as many markets as possible or to focus on a handful of big markets and make the product as good and valuable as possible.

[00:22:30] Nate Elliott: But either way, they're counting on advertising in 2030, according to their own leaked estimates, um, to account for $100 billion out of less than $300 billion in revenues. And they're counting on consumer subscriptions to account for another about $40 billion. So, you know, their own numbers are estimating that in 2030, half of their revenue will come on the consumer side, and half- Mm-hmm

[00:22:55] Nate Elliott: their revenue will come on the enterprise side. Mm-hmm. Again, we have [00:23:00] significant and well-supported reasons to doubt that they can come anywhere near those consumer numbers, whether from subscriptions or from advertising. I haven't spent enough time looking at the enterprise side to know if the 140 or so billion dollars they think they can make in enterprise money in- Mm

[00:23:14] Nate Elliott: 2030 is, is achievable. But-

[00:23:16] Marcus Johnson: Yep ...

[00:23:16] Nate Elliott: you know, they very clearly have laid out a plan that requires huge amounts of revenue from both sides of that equation. Now, again, if the enterprise side makes a lot more- Uh, a lot more money or is a lot more profitable than the consumer side, it does raise a question of whether it was a good idea for them to get into the consumer business in the first place.

[00:23:38] Nate Elliott: But I've, you know, been around this industry for longer than I care to admit, and one thing I know is that, uh, companies don't often just dispose of businesses that they think might generate $140 billion in revenue in a couple of years- Yeah ... even if it was a bad idea to get into that business in the first place.

[00:23:54] Marcus Johnson: Yeah. Uh, all right, gang. They're the questions we have. Let's put together a top [00:24:00] three.

[00:24:00] Paola Flores-Marquez: So it's- I love them both, but I think that, um, the question about whether the consumer market is worth it is a really strong one. I feel like it's- Mm-hmm ... something they're gonna have to wrestle with in the next couple of years.

[00:24:11] Paola Flores-Marquez: Um, and my immediate instinct is like, I don't... For the sake of all of us, no. No. Mm-hmm. On a selfish level, no. But, um-

[00:24:22] Nate Elliott: Yeah ...

[00:24:22] Paola Flores-Marquez: but that decision's out of- All right ... out of the bag already, so.

[00:24:25] Marcus Johnson: Mm-hmm. Nate, how about you?

[00:24:27] Nate Elliott: Uh, I love Pal's question about kids and safety. Um, it's, I mean, the question of harm in technology is such an important one in general, and one that the industry itself doesn't spend nearly enough time thinking and talking about or addressing.

[00:24:42] Nate Elliott: And as someone who has covered social media for many years, and now is covering AI, I've watched a lot of those harms play out, and I really hope that we can do a better job of getting a grip on it as this technology evolves compared to the last big change we saw with social.

[00:24:59] Marcus Johnson: Mm-hmm. [00:25:00] Um, I wanna pick the one about what, what can they really, realistically expect to make in advertising?

[00:25:08] Marcus Johnson: 'Cause there's such a gap, so I'm curious to see more estimates about where that number actually lands between a few billion to 100. There's one estimate from Barclays, uh, Barclays, but, yeah, I think this is a, a, a really interesting question, um, which I guess kinda ties into that IPO of, um, is this gonna be almost nothing compared to what they're hoping or, or absolutely everything?

[00:25:31] Marcus Johnson: 'Cause it's quite a big range at the moment. Um, all right, that's our top three for today. Thank you so much to my guests for hanging out with me today. Thank you first to Nate. Thank you, sir. And to Pal.

[00:25:40] Paola Flores-Marquez: Thank you.

[00:25:42] Marcus Johnson: And to the production crew. Luigi, nope. Is he still here? Who knows. John's definitely there. I think Lance was in the background as well shouting things.

[00:25:52] Marcus Johnson: Thank you to those guys. Thanks to everyone for listening in to Behind the Numbers, an eMarketer podcast made possible by AI Digital. We'll be back on Monday with the big three questions [00:26:00] for Google. But until then, happiest of weekends.