OpenAI Rolls Out Ads, AWS Weighs AI Content Marketplace for Publishers, How OpenClaw Agents Think

0:13Welcome everyone to the information's TI TV. My name is Akash Pashrich. It is Tuesday, February 10th. First on the docket today, OpenAI has begun rolling out ads across its platform. Our e-commerce reporter will weigh in and she will also share some of her new reporting on the tax challenges that could complicate OpenAI's commerce ambitions. Next up, we'll bring on our Amazon reporter to dig into her exclusive reporting about AWS's plans to launch a content marketplace for publishers to sell content to AI companies. It's also been another dramatic week for software stocks. Workday's CEO is stepping down and handing the reins to the company's co-founder, while Monday.com is trying to shake off investor concerns fueled by the rising anxiety around AI.

1:01We'll talk to an analyst who has a strong point of view on all this. And finally, we will wrap the show with the latest chapter in the Open Claw story. Our AI and robotics reporter published a new piece on how agents are improving so quickly and what that looks like on a technical level. It's going to be a fun show, so let's get right on into it. OpenAI this week said it is continuing its advertising rollout across certain accounts on its platform, and the next step after ads will likely be shopping features inside ChatGPT. My colleague Anne Guillen has done a ton of reporting on how difficult payments will be as part of that ambition, and a new story that she published today dives into how taxes are making all of this even more challenging.

1:45Joining me now is Anne to help us break it all down. Anne, welcome back to the show. It's great to have you here. Hey, Akash. Payments and taxes. Payments and taxes. Two of the most exciting season. And as tax season comes up, it's the thing that we love to talk about here the most on the show. Before we get to that, because I know you're on the edge of your seat and you just can't be excited enough about taxes, let's talk about ads, okay? So OpenAI said this week it's continuing its rollout for ads. It's across certain accounts. Some of the accounts won't see the ads. What are your initial reactions from people you're talking to about the ads they're seeing and also how brands are feeling about how the rollout is even going?

2:25Yeah, so the first ads, OpenAI said, went live yesterday. And so the ads will be shown to, you have to be a logged in user in the U.S., but they'll be shown to users that are logged in that are either on free plans or on OpenAI's new Go subscription tier, which is$8 a month in the U.S. So it's a little bit lower priced than kind of the full subscription. But OpenAI did also say yesterday that if you're a free user, you can opt out of seeing ads in exchange for fewer messages and a lower rate limit. So that's kind of an interesting wrinkle there. But yeah, I think most people that I'm talking to are still waiting to see how this will all play out.

3:18OpenAI is starting with a relatively small pool of advertisers. And most of the ads that they're starting with, we've reported, are all going to be mostly just kind of general brand awareness ads. OpenAI is charging advertisers per view, not per click, which is different than most other large ad businesses. So OpenAI has been pretty adamant that this is an early test. They are going to see what happens and see what the reaction is like. But certainly their need for revenue is growing by the day. And so it will be interesting to see, you know, both how users react to the ads and then also, you know, the demand from brands as well.

4:12I like that they doubled down on the Monday after the Super Bowl. We obviously saw that Anthropic came out with a fiery campaign that very much silently said that, hey, Hope and AI is doing ads, we're not doing ads. And Monday morning, we had this blog post from OpenAI saying, you know what, we're doing it and we're going hard at it. And I didn't know the piece, though, that you mentioned about the fact that users can opt out for a tradeoff there. And so that is certainly interesting. Let's talk about the next step after ads. So shopping is something that you've done a lot of reporting on. We've talked about how payments is so important in shopping.

4:50Your new story out today talks about how taxes makes all this even more difficult. What did you find? Yeah, I mean, so like you mentioned, I've done a lot of reporting around kind of the data wrangling that OpenAI has had to do with merchants in order to help them, you know, get set up to sell their products through the checkout feature in chat GPT. And so I think this, the payments and And the tax work and, you know, those questions that they're thinking about right now are kind of another leg of that, you know, getting into retail and running this thing that's almost like an online marketplace.

5:35You know, there's a lot of kind of back-end work that goes into that. And so kind of the main thing that OpenAI teams working on commerce are working through right now is what is their responsibility going to be in terms of sales taxes? It's an important piece of any online marketplace. Their Amazon, very famously in the early and mid 2010s, went through kind of this whole saga about whether or not they were responsible for collecting and remitting the sales taxes on behalf of the merchants selling on their platform. And ultimately, they were found to be responsible, even though their argument was kind of just like, oh, we're only connecting buyers and sellers.

6:27We're not taking an active role in the transaction. But most of the state legislatures under their laws, you know, found that Amazon was responsible as kind of a part of this transaction for facilitating, you know, tax payments. So it's just kind of another like it's another thing that OpenAI has to sort out in order to really scale up this shopping feature. And the other part, too, that I found interesting is I guess it's not just who has to collect the taxes, but then it's also trying to figure out, well, what state is the person in when they make the purchase? Which sales tax rules apply? I mean, that seems to be another intricate thing in all this.

7:17Yeah, I spoke to a couple of tax experts for this story. And kind of the main takeaway there was that the laws are all slightly different state to state. You know, they change depending on what kind of a product you're buying, what time of year it is, whether or not there's a tax holiday. So sorting all this out is quite complicated. And there's a reason why most other marketplaces like Amazon, you know, have taken steps to kind of sort this out on behalf of merchants. And then also, you know, the legal responsibility that they have to pay taxes. Now, what about, I'm trying to think of the payment processors like Stripe that have a role in all of this.

8:02I mean, do they not end up carrying some of the burden here? They do. But kind of the key distinction in a lot of these laws, they often kind of focus in on how much control an app or a payment processor has over collecting payments. And so a lot of the laws are written with this phrase, either collecting payments directly or indirectly. So you could make an argument that even though OpenAI, you know, isn't directly touching the transaction, you know, they still are indirectly facilitating payments. And so therefore, under these tax laws, they would be responsible. So, you know, yes, OpenAI is working mainly with Stripe, but also is laying the groundwork to potentially work with other payment processors down the line.

8:58I touch on that in my story today a bit as well. And so especially as OpenAI kind of lays the groundwork to work with a broader range of payments processors, I think that question of how much are they involved in the transaction, how much are they involved in payments specifically, I think that makes the tax question, you know, all the more pressing for them. And I guess big picture, Anne, here, I mean, the reason that this really, I guess, matters, you know, to the shopping store at large is I imagine, you know, as we look to the executives that OpenAI will have to bring on to help them sort through this.

9:38I mean, I imagine this is going to be a big hiring push for them. And I'm trying to think about the people that they might seek to poach from Shopify or Amazon, et cetera. I mean, these are very specific skill sets that I think they're probably going to need from these larger companies. Sure. I mean, yeah, it definitely means that they will have to think, you know, about what their needs are in terms of people that know compliance, that know taxes. And so I think kind of this vision that maybe they started with of being this assistant layer that just sits on top of existing marketplaces, existing retailers.

10:17You know, there's a lot more back-end work that goes into e-commerce no matter what kind of role you play. So, yes, I think you'll certainly see OpenAI invest in staffing up those teams even more going forward. Great. Well, Anne, I want to thank you for coming on. That is Anne Guillen, our e-commerce reporter and our resident tax expert for the day here at The Information. Amazon Web Services is making a bigger bid to cozy up to publishers or the websites with original content like news outlets. The Information published exclusive reporting that AWS is looking to launch a marketplace for publishers to sell their content to AI companies.

11:00Joining me now to break it all down is Katherine Perloff, our Amazon reporter. Katherine, welcome back to the show. It's great to have you here. Yeah, Kosh. What do we know about this content marketplace that you've reported AWS is looking to launch? Yeah, so it would be a marketplace where publishers could sell their content and firms that are either making models, we don't exactly know what the customers would be, but hypothetically be either firms making models or firms that sell products that have AI in them. And they would use the content in, you know, potentially use cases could be fine tuning, you know, training in the actual responses themselves and retrieval.

11:48I mean, these are all a little bit hypothetical because we don't really know a ton of details about exactly, you know, what the contours of the marketplace are. What we do know is that Amazon has talked to publishers about this sort of marketplace, which, yeah, as a means to help monetize publishers' content as it's sort of eaten by AI. So big picture, I mean, this is kind of interesting, right? Because Amazon and other AI companies, they have these bespoke one-off deals with content publishers that we've seen come together with some of these news outlets. And it sort of seems to be the case that that would rely very much on the relationships that are built between the AI companies and the publishers.

12:34Amazon is saying, hey, we're going to get everyone together and it's going to be a little bit more, not automated a little, little easier. You also got a hold of some of the slides that Amazon is looking to present this week at a conference about this. What did those slides say? Yeah. So today actually, um, Amazon is having a publishing event and, um, they talk about a bunch in these sides of like different Amazon products, um, uh, that could help publishers, including like, you know, stuff we know about like Bedrock and Quicksuite, you know, these kind of just like AWS business software tools and AI tools.

13:15And among them was listed a content marketplace. So that was sort of indication that like, they might be discussing this today, you know, not exactly sure. But we do know, apart from those slides, that they have talked about this with people in the industry. And nothing like this exists already? There's no marketplace of this kind? Indeed there is. So, you know, the first, I think the most relevant, you know, example is that Microsoft is launching a similar marketplace where publishers can get paid for their AI. We first sort of heard about it late last year because People, Inc., which owns like People Magazine and a bunch of other household names, said they were in the pilot program on their earnings call late last year.

14:02But actually, it was not until last week that Microsoft formally announced that they were doing this. And apart from Microsoft, Cloudflare, which is a CDN company, basically just helps websites connect with their servers, has thrown their hat in the ring as both helping publishers block AI bots and then charge for access if they actually want to let those AI bots in, as has Akamai, another CDN company. So Amazon is sort of entering this. They're not the first mover here, but they are a big and significant player. So it could be a big deal if this was able to become a scaled product. And how have publishers reacted to those other marketplaces being set up from Microsoft, from Cloudflare?

14:49And what does that reaction tell us about how you think they'll react to Amazon's marketplace? I think they're really excited because, you know, as you noted, there's been some deals, notably with OpenAI and news publishers. But these are kind of one-off deals. A lot of them are flat fee in structure. So it's like you kind of get paid like, you know, X million dollars over a year or something. and they're not really sustainable, like especially when you think about a publisher's traditional source of money, which is advertising. You know, you get more ad revenue if you can get more people to look at your site.

15:23So it's a way to scale. And with these fees, and with these existing deals, you know, you're just getting kind of paid once a year and then you have to renegotiate. The idea of a marketplace can mean the more that an AI company uses the publisher's content, the more money they get, which seems like a much more sustainable business model if people start getting their news and information more from chatbots. So I think there's a lot of excitement. I think, though, the details on a lot of these programs have been pretty scant. Even like Cloudflare, they were the first out of the gate coming out last July.

15:56And we still haven't seen a ton of adoption or publishers talking about making a lot of money from them. So I think how it actually works is still a little bit of a question. And does this help them with their end goal then of getting paid per usage, I guess, to the amount of times your content gets used? I mean, that was the end goal. Is there reason to believe that these marketplaces will help them achieve that goal? Yeah, I think that that is the goal. And I think the idea is like they want to replicate something like the way advertising technology works today, where it's kind of an automated way of, again, yeah, getting paid the more people use your product.

16:33You know, I think there's always a risk of getting technology involved. I think with that technology, we saw. Because then you can track everything. You can argue both sides of it. You can say, well, you're not worth as much as you say, really. That and the more middlemen you have, the less dollars go to the publisher. You know, they're going to take, I mean, we don't know exactly what the business model is, but that was the case with advertising technology. Right. So then, I mean, on that piece here, are the AI companies, are they largely excited about this then? I mean, I'm thinking about this also from a how much they need to pay the publisher's perspective.

17:09In some cases, I imagine it means they might have to pay the publishers more because they are prioritizing this usage-based pricing. On the other hand, I mean, you know, it's just easier for them to get the content. I imagine they're largely pretty happy about this. I guess I think that is the biggest question about all of this is like, you know, I mean, AI companies, you know, like the frontier models, they were for a while just, you know, sucking up all this content and not paying. So it's like, you already get it for free. Are you going to start paying? And I think that, So with the Microsoft example, they said that before they launched it, they tested with Copilot, and they found that having publishers' content made their consumer and enterprise versions of Copilot better, the answers were stronger.

17:55And you do see a lot of need for data, like companies like Scale AI or ultimately data companies, and they're there to help AI models have stronger answers. And what can make one model unique from the other is the data they use. On the other hand, I think there is an open question, like, will AI companies see publishers as a valuable asset in this equation, given that for so long they were not really, you know, paid for it? Right. And last question for you. I mean, I'm curious about the reason why Amazon is doing this from a business perspective. How much extra revenue is this really for Amazon to play a part in all this?

18:38Are they sort of just doing this out of the goodness of their heart and wanting to be a leader in the AI publisher relationship? What's in it for them? Yeah, I mean, it's interesting. I think that like, so they, we don't exactly know the degree to which this is involved, but they have a service with an AWS CloudFront and that competes with CloudFlare and Akamai. And those CDN companies haven't been like, you know, kind of as in the spotlight recently. So it's a way to innovate in these product categories that maybe have been a little bit kind of like older tech. So there is some advantage there and like kind of competing with the market.

19:16I think, you know, also Amazon, like, I mean, it's separate from their ad business, but they've been building a really big ad business and they've been working with publishers on that. So I do think, you know, it doesn't hurt them to have better relationships with publishers when they need them down the line to say, like, you know, work with their ad tech or, you know, place Amazon ads in certain places. So I think that's all the goodwill part could be part of it. But I don't think Amazon, like, is going to fully do anything from the goodness of their hearts. Great. Well, Catherine, I want to thank you for coming on.

19:47That is Catherine Perloff, our Amazon reporter here at The Information. It's been another dramatic week for software stocks. Monday.com had quite the Monday. Shares fell more than 20%. Workday shares also fell after the company said it was replacing its CEO, bringing back co-founder Anil Bushri for the top job. To help us break us all down and break down the state of the sector at large, I want to bring on Rishi Jaluria, software equity research analyst at RBC Capital Markets. Rishi, welcome back to the show. It's great to have you here. Great to be back. Thanks for having me. Let's start with Workday.

20:22What did you make of the CEO shuffle? Yeah, look, I think a couple of things are going on here. Number one is, you know, Carl was brought in with a lot of fanfare. And I think you did a lot of great things in terms of building out the partner ecosystem, some of the go-to-market stuff. But I think there was recognition within the company. Number one, that growth had continued to decelerate. And you've seen that kind of show up in the stock price performance and some of the long-term targets continuing to have to be brought down. But I think there was also this, well, AI is shifting things at an exponential rate.

20:55And there was almost this perspective of we need to bring the founder back in. We need to bring a technologist to do things. And that's increasingly kind of a thesis I think we're still in with is, you know, is this the time where you need technologists to kind of step back in to make sure these companies can invest in the right areas in AI and, you know, cross the chasm in this post-AI world, so to speak. So are you expecting more CEO changes in the software sector? We've seen the carnage, and I want to talk to you about it and what you make of it all. But broadly speaking, you know, I'm sort of trying to think, you know, Snowflake got ahead of this early, right?

21:34They brought in Sridhar, and he was obviously from Neva. He had the AI expertise. That was before all these software stocks started tumbling, although Snowflake had been on a bit of a journey itself. Are you expecting more of these CEO changes to come? I would expect more of this. I mean, you know, we saw that happen with MongoDB. I thought David Tietari was a fantastic CEO. but they brought in CJ Desai to accelerate specific initiatives. So they did that too. And in the middle of all of this, right? I mean, thankfully the stock had already been recovering under Dave and I think CJ can maybe take it to another level.

22:08I wouldn't be surprised to see more of these CEO changes happen with the recognition that this is unprecedented times that we're navigating through. And sometimes you need a new leader to be able to navigate that. Like could Adobe have migrated to the cloud without Shantanu? could Microsoft had undergone their cloud migration without Cetia, maybe not as successfully as they did. Okay, so who are you watching then for these chains? I'm looking at your coverage list here. I mean, you got all the big names under your belt. Who are you watching? Yeah, look, I know this is gonna sound like a cop-out, but I'm watching any company that is not founder led, right?

22:45That's where I think there's probably the biggest opportunities. Because a founder led companies, if anything, I think that, like we saw with Workday, they're kind of stepping back in, managing the roadmap, taking the long-term perspective. So I think it's the companies that maybe aren't founder-led where there may be those opportunities. There's no specific names out there, but I would say non-founder-led companies that have had massive underperformance in their stocks are probably going to be the ones where I think you're most likely going to see a CEO change happen. Okay, so let's talk briefly about monday.com.

23:16There was another name that dropped more than 20 % yesterday. What did you make of the story there? Yeah. Look, I think, and Chandler, I don't cover money.com, but I pay attention to it and cover some of their competitors as well. Look, I think it's showing us, and I would combine that along with ZoomInfo reported yesterday, their stock's down double digits this morning as well. It's telling us that no matter how cheap any of these stocks are, if a company guides below where published consensus is, and the general viewpoint is published consensus is a conservative number and should be beatable, if you are guiding below that number, it's kind of almost a death sentence for the stock.

23:56And I think we saw that play out with monday.com. I think we're seeing that play out with Zoom and Co this morning, that you cannot guide below that. And candidly, even an above market guide may not be enough, as we saw with Microsoft and service now. And candidly, I think public companies are in a very tricky situation where they can't guide so conservatively that their stocks get whacked because it starts to feed the AI bear cases. But if they guide too aggressively, no one's going to believe those numbers. And then if they do end up hitting them, they won't get credit because it's already in the guidance.

24:26So it's a really tricky balancing act that we're, I think, going to deal with this upcoming Q4 earnings season, both on and off cycle. So let's go past the earnings for a minute. Let's talk about the valuations broadly. Look, there's always been a bit of a seesaw here. Are investors valuing growth? Are they valuing profitability? Which one of those is more important at any given moment? You're at a point now where I think the median software company is trading something like three to four times sales. And And that is a number we haven't really seen in a long time. But my question for you is, where are we in that valuing growth versus valuing profitability story?

25:06Is there more weight put on profitability today? Because some of the most profitable companies are still getting traded down. Yeah, it's an interesting question and a very timely one. We actually just published a note on this exact topic. But you are seeing some of these names pull back to the point that you can value them at basically market gap price to earnings ratio. I don't know that I've ever seen this play out in software before. It's very much unprecedented times. And, you know, what's interesting is in contrast to previous SaaS drawdowns that we've seen, like 2016 or the kind of post-COVID cloud rationalization era, numbers in general are not coming down, right?

25:42putting aside cases like Monday and ZoomInfo, but my numbers for Microsoft went off post that quarter in spite of the drawdown in the stock. My colleague who covers ServiceNow, his numbers went up for ServiceNow in spite of the drawdown in that stock. And I think it's telling us this is all terminal value risk. And if you really do the long-term math on some of these names, they're being valued with negative terminal value, which in other words, it's saying that this company is being valued today as if it's going to be declining in perpetuity at 5 % to 10 % a year. That could happen to some companies.

26:17It's not happening to every company. I think that's where it's creating some really big opportunity here. And so what do you think is a good median multiple then for a software company? This is like very hard for anyone to say really, but we're at three to four right now. The question really I'm asking, I guess, is if you think things are oversold and if you expect it to come back up, you know, where do you see them going from here? I do, but I almost think when we get the bounce off the bottom, whatever that is, you know, look, cheap stocks can always get cheaper. But when there is that bounce off the bottom, I almost think it's going to be a little bit of a K-shaped recovery, where some companies might bounce 20%, 30 % off the bottom.

26:58And I think there are other companies that could be a double or triple over the next couple of years from here. And, you know, I mean, I'm going to give it in very broad brushstrokes, but I almost think that the quote right multiple is maybe call it 15 to 20 times cashflow for the companies that are, you know, the incumbents that aren't really going to be AI winners, but also aren't going anywhere. And then that number could be 30 to 40 times cashflow for companies that are going to actually be AI winners that have innovation, that have the ability and right to get AI skews in the hands of customers and actually monetize that over time.