Inside OpenAI’s $3.7B Cash Burn, White House Ban on Anthropic, & Salesforce Enterprise Talent War

23:13Kevin McLaughlin:I mean, this is a real concern for retail investors is you're buying it when the name couldn't get hotter in terms of at least perception, I guess. There's all this energy around the IPO. You're not concerned that retail investors are going to get the short end of this? Well, I think that there's, of course, always a concern there, but that's where PowerLaw, with what we're getting, buying into PowerLaw and having access to, in the case of our flagship fund, the 15 biggest and what we believe most influential private technology companies give some diversification, which obviously protects against any single name.

23:54How do you pick the companies? So in the flagship fund, it's really looking at what we believe are going to be the most influential companies over the next decade and having, and our thesis is all around concentration. So having a concentrated exposure to the best of the best, big private tech. And so in the flagship fund, it's across the board, so it's not as thematic. The next fund we're launching is going to be focused on AI. And so that will be concentrated across infrastructure, horizontal, vertical applications. But the flagship fund is just giving a diversified basket of what we think are the best of the best big tech.

24:31Kevin McLaughlin:And how do you compete against some of these other similar fund products? i think when we had robin hood on the show they launched sort of a similar fund uh other than just the selection of the companies and and you know sort of asserting that we have a better mix what is the the differentiation strategy yeah i mean in our portfolio we have 11 of the 18 names none of our competitors have and so we have a differentiated portfolio to begin with and then on top of that i i you robin had certainly has a great competitor but it's really a brokerage product focused on retail in a small segment of the market.

25:09What we're focused on is being structurally built into U.S. retirement, which is a$25 trillion opportunity in the U.S. alone, and then globally. And so we're the first venture group to do this. Robinhood is certainly a great company, but they are not venture investors. And so we sort of look at that and say it's easy to build a portfolio of what's obvious. But really what differentiates us is we've been around for 16 years, we are venture capitalists. We're in San Francisco. And so what's next? And that's where we're focused. And how do we build in names that perhaps aren't as well known to the market, but will provide long-term value in a way that we think differentiates us in our underwriting discipline and just heritage in venture?

25:53Kevin McLaughlin:I want to ask you, you have largely been a career CFO. You've been CFO of a number of companies, DoorDash, DocuSign, Gusto. CFOs are in an interesting moment here right now with AI. On one hand, they have to prioritize ROI. On the other hand, AI is expensive and you don't want to fall behind and you also don't want to get ahead of your budgets. How are you advising your CFO peers to look at this question right now? I think it's got to be an incredibly interesting time to be a CFO. So I haven't been a CFO for six or seven years. And I can talk about those two companies, specifically DoorDash and Dr.

26:36Stein, which I think are pretty insulated. I think that most of us are just scratching the surface on what AI can do in terms of productivity for all of us. I mean, just the work that I can do myself now is 10 times what it was a year ago. And so it's just got to be as a CFO driving in. It's not around thinking of AI as a SaaS cost line, but it's all around, is it going to help us create a real mode in our business? And do we have a mode even? Can someone else replicate this very quickly? And if we don't, so if it's just informational, that's a mode that will erode very quickly. If it's operational, which is what I think DoorDash and DocuSign both have in their own ways, then it is sustainable.

27:18And AI will just increase the mode versus decrease the mode.

27:22Kevin McLaughlin:operational you mean as far as helping the employees do their jobs no i'm talking more sort of a business model level if you look at say a docusign the value of docusign is absolutely not the e-signature component which is commoditized and can be replicated very quickly it's the operational component being built into workflow and things like this that's embedded in enterprise you know workflow applications which can't be replicated easily or in the case of doordash It's all around the three-sided marketplace and people and operations related to this. And that's where DoorDash has just excelled and was really excelling before AI was as impactful as it is today.

28:01But it's just assessing whether or not your business really can survive. And you do have something that's differentiated or can be replicated quickly.

28:09Kevin McLaughlin:What about the employees? So token maxing is something that has dominated headlines. And now we just wrote a story last couple of weeks that now Meta has sort of embarked on this new era, token minimizing. They haven't said it specifically, but they have said, look, we're going to make it a little more formal in terms of seeking approval to use certain allotments of tokens and stuff like that. That seemed inevitable. But I guess the question that I have for you as a career CFO is how do you balance this? I mean, you need tokens. Like, you know, there's no other way to increase your efficiency other than using the tokens.

28:52Kevin McLaughlin:How do you balance that? I mean, this seems like an incredibly difficult problem just when you think about, obviously, you want to promote the use of these tools to do good things, but you don't want to promote the use of the tools to do busy work. And I think that's going to be the tricky bit, which has to just go back to alignment around what are we trying to accomplish and are we accomplishing the right thing? like are we adding value versus just doing work for the sake of doing work which of course you can chew through tokens just doing that and i think that just goes back then that thinking about making sure that everyone's aligned on the goals of the company and what we're trying to achieve and focused on things that do add value but this is a very trick problem you know we face it even in our small firm at power law making sure that that people are focused on doing the right things and using the tools we have, and we have the same problem.

29:44Yeah.

29:44Kevin McLaughlin:What's the biggest piece of it? What's the biggest, I'm sure you have early CFOs in their careers or maybe established CFOs coming to you for advice and counsel. What's the biggest issue that they're having right now that they want your advice on? You know, I'm actually not as much involved in sort of direct inside of companies. we're really focused on building power law as a capital group and so all of my energy is around fundraising getting ready to launch we're launching our next fund in the coming weeks um and that will be focused on ai and so it's just thinking through any any particular focus within ai how are you going to think about the the makeup of that fund yeah i think like in a lot of ways the the infrastructure companies are a little bit easier because it feels like there's more defensibility when you start to get into horizontal applications to some degree.

30:38But for us, it's trying to make sure we understand which companies won't be assumed into Anthropic and OpenAI, which really are standalone companies and not just features. And it just seems that proprietary data sets or regulatory environments or things like this that just create mode that the big LLMs won't go into, that's where we're focused. which are really companies that will stand the test of time and can turn into large companies versus just -

31:09Kevin McLaughlin:Which companies are you looking at then? Well, I mean, there's things like Implied Intuition, which I think is an amazing company. Vast Data is a great company. And then you get into things that are focused on, say, medical, like open evidence or things like this, that I think do have defensible lanes. So that's where we're thinking, and that's where our heads are at, is how do we build a portfolio of 25 to 30 AI companies that really are focused on and have defensible modes that will be around for 10 years or 20 years. Great. Well, Mike, I want to thank you for coming on. That is Mike Dinsdale, CEO of PowerLaw here on TI-TV.

31:50Kevin McLaughlin:Our next segment is with our partner, Adobe. Digital advertising has been dominated by a handful of tech giants for years now, but AI has changed the game entirely. New players like OpenAI are now in the mix, and meanwhile, marketers are rethinking how to make and measure their campaigns. I recently spoke with Varun Parmar, SVP and GM of Adobe Gen Studio and Firefly for Enterprise about all of this. Here is that conversation. Varun, welcome to TITV. It's great to have you back. Yeah, excited to be here, Akash. Thank you. So I want to start this discussion with AI advertising, because we have talked on this show a lot about how big giants like OpenAI are diving headfirst into AI advertising.

32:37Kevin McLaughlin:And the perspective we have not covered as much on the show is how businesses are thinking about advertising on these platforms. This is a group that you talk to often in your role. How are businesses planning to approach AI advertising? Yeah, no, great question, Akash. So I think if we take a step back and we ask what is the jobs to be done by advertising, there's like really at the highest level two things. The first one is to decode the intent. And the second one is to monetize intent. And what we're seeing in terms of the future of digital advertising is that in decoding the intent, it's becoming hyper contextual.

33:15What do I mean by that? Essentially, an ad is not just a static banner. It's more conversational where the user types in their preference. And just like we have a conversation with an LLM app, you are having a conversation with the ad unit where it's uncovering your preferences and then recommending what the ideal product is. And that's what we call hypercontextual. And then in terms of monetizing intent, what's happening is that there's agentic conversion that's happening. So essentially, you know, an agent is understanding, you know, what your preference is and then making an offer, you know, giving you a discount code, you know, you know, recommending a product that's available right now, you know, that you could actually one click buy.

34:00Right. So all the brands, you know, are looking at these trends and saying, how can they participate in terms of, you know, these macro trends and then see higher conversion rates, you know, having much more efficient CPC and CPM rates as part of the process.

34:19Kevin McLaughlin:So if I'm understanding it correctly, in the old advertising era, the way that we often talked about it was impressions and then click through or conversions. Now it feels like there's a bit of a layered approach where you might have impressions, but then there's sort of a layer which is, well, did the person actually engage with the advertisement maybe through the chat interface like you're talking about? And then at the end, there might be a click through, but in the middle, there's sort of this other layer, which is, well, to what extent, you know, how do you get the agent to give a proper recommendation?

34:57Kevin McLaughlin:So it's kind of a more tiered approach to advertising. Am I understanding you correctly? That's right. That's right. You know, the North Star always with advertising has been around conversions. And conversion is when the user decides to purchase the particular product from the brand. And in the past, you know, there's been something called a scapy, which is the conversion event that is fired off. And then you can attribute, you know, to the specific advertising. Obviously, there's a user journey that needs to go in. So proper attribution needs to happen. But now what we're seeing with all of these LLM apps is that the marketing funnel is getting collapsed.

35:33So all the way from awareness to discovery to consideration to the actual purchase can be done in a single session where the user is going back and forth through the conversational experience. And what we're seeing is that in that experience, you know, the ads are evolving, you know, as well. And the ad unit is actually recommending things in terms of the intent of the user and saying, based on what you need right now, this is the best product that you should get. And while the traditional metrics around CPM and CPC, that's how these conversational ads or these next-generation ad units are also being priced, really what the big disruption or the big opportunity here is dramatically simplifying and accelerating the conversion event on the platform.

36:23Kevin McLaughlin:Right. So the ad unit of the future, it looks more like a chat interface rather than just a static ad. Exactly. And therein lies the opportunity and the challenge is that you want to make sure that when a user comes to your application in LLM app, that it's very clear to them what is the true conversational experience they're having versus what is the ad conversation experience you're having. because trust is really, really paramount. So, you know, the user wants to make sure that there is an objective recommendation that the LLM is making. But at the same time, you need to make sure that when an ad is presented, one, visually, it's very clear, so the user knows this is an ad.

37:04And then once the ad is presented, that they can also interact with it. So that's exactly, you know, where things are headed.

37:11Kevin McLaughlin:Now, what about if we go back to the large chatbot platforms, You know, there is this whole game of answer engine optimization and how I get my brand to show up in the most favorable way in these giant new search platforms, I guess you could call them. So how, what's the answer to that? Is that just really intelligent content management? Is it putting out more content? I mean, how do you ultimately get your brand to show up better in these platforms? Yeah. So I think, again, there are sort of two things that are happening here. One is sort of the structural shift that's happening where a lot of focus was on search engine optimization.

37:55So SEO is becoming SEO and AEO, and AEO is answer engine optimization. And players like Adobe and there are a few others allow you to get better brand visibility. So, you know, understanding what are the sources of information and data that lead to better placement of a brand as part of that conversation experience. You know, we know, you know, these algorithms will pay extra weights, you know, onto sources where, you know, there is a community of folks that are discussing what the pros and cons of a particular product is. There's an influencer, you know, that is doing a review and there's a bunch of comments that are put against the review.

38:40So there's like a bunch of things that these LLMs are doing in terms of better placement of the brand. And, you know, we help you amongst others in the industry to have a better, you know, visibility as part of these experiences. Right. So that's the first part. And then the second part that we are talking about is like, you know, if SEO is becoming SEO and AEO, The second part of this is what happens to SEM, search engine marketing, right? So what's the future of that? And that's where these new ad units are emerging. And our point of view is that it's a combination, you know, both AEO as well as these new ad units, which we are referring to as conversational ads, will lead to the true unlock, both in terms of discovery, but also conversions in these new application experiences.

39:26Right.

39:27Kevin McLaughlin:Now, OpenAI is just one of these chatbot platforms and it's just one publisher of ads. There are also a ton of other smaller publishers, I guess you could say, that will have to deal with this exact same question of how do they change their ad offering in terms of what they sell to brands and stuff like that. Do you have conversations with these publishers about what they want? What can you tell us about how they're thinking about their product? Yeah, no, absolutely. I think there's definitely a very big trend around commerce media networks, retail media networks, financial media networks, travel media networks.

40:09Essentially, all of these publishers that have a treasure trove of first-party data, right? You've been staying at Marriott, and Marriott knows you very well. and one of the big sort of areas of growth for them is around putting endemic and non-endemic ads on their platform, right? In the past, similar to what we've done, you know, with Search and some of the other digital properties is that, you know, there's a banner ad that's introduced, you know, just a heading and, you know, an offer, a CTA and an image and all of them are actually taking a step back and saying what the future looks like.

40:48So I think there are like two things that are happening there. One, there's a tremendous opportunity to go after the mid-market or the long tail, where typically these commerce media networks have relied heavily on large brands that actually had an army of folks who could create the creative, create the offer, create the placements and such. But there's a huge opportunity as part of their ad platforms to allow for self-service. This is where Adobe's product, Adobe Gen Studio for Commerce Media Network, which is this new offering that we released, comes in, where we seamlessly embed inside of the ad platform from these commerce media networks and enabling self-service.

41:28So that's one big vector of growth. And the second one that we just talked about is what does the future of the ad unit look like? So if that's going to evolve in the open web, you know, our belief and that of these publishers is that the ad unit, even on their properties, need to evolve, right? So what does a conversational experience around the ad look like if you go to Costco, if you go to Marriott, if you go to Amex and some of the other folks who have these commerce media operates? Yeah.

41:58Kevin McLaughlin:Well, and it's kind of interesting to me, you know, as you say this, I sort of have come to realize that whereas the old relationship between an advertiser and a brand was the brand gives the advertiser the content, the advertiser now in some ways has a has a widened role in terms of producing the content. because if the ad unit of the future, as you say, is sort of a chatbot-like interface and what I see depends on what I put in, then basically the publisher, they have more of a role now in, like you said, generating the content that inevitably would do the role of attracting a call to action.

42:38Exactly. Yes, yes, indeed. And I think herein lies the opportunity for all of these big advertisers is that more so than ever, I think the role of having a strong brand where the user, when they're going in and if they're searching for a shoe, they actually put in the brand of the shoe while they're asking for a recommendation becomes extremely important. So when they are creating all of the content, it's just not just an ad that needs to do lower funnel conversions. It needs to represent the brand as well. And the question is in the world where all of these systems are being run and optimized using AI algorithms, which are increasingly agentic, how do you maintain, augment, and present your brand in a way that actually drives loyalty and user engagement?

43:28So I think things like Adobe Brand Intelligence essentially allows you to codify not just what's written about your brand, but also how people interact with your brand, You know, building that model that can actually allow agents to understand what your brand is. And then, you know, right brand loyalty becomes extremely important.

43:48Kevin McLaughlin:Let me ask you one last question. So in this world here where we have this new ad unit, a chatbot-esque interface, do you have any early data on how much more effective those ad units are than static ads at driving people down the marketing funnel and ultimately converting them? And the reason I ask this is because I imagine that these ad units must be more expensive than just a static ad, not just because of how advanced they are, but there's also the whole compute underlying the interface to it. I mean, like using these models is not cheap, right? And so are they more effective? Do we have data there at all?

44:29Yeah, I think I would say, Akash, like these are early days. I think these are really early days and we are all trying to experiment and learn. I think one thing is for sure that these units are expensive. If you look at the typical CPM rates, just look at OpenAI, you know, started off at launch with$60 for the CPM or the CPC was between$3 to$5. And we're seeing like what the clearing price is with all of these DSPs, which is lower than that. So I would say they started off high. We are all learning in terms of what the conversion rates are. There is no publicly available report either from OpenAI or Google in terms of Google Gemini conversational ads around AI Max or Pmax.

45:06So none of that performance information is available. But our belief is that if you hyper-optimize the user journey, where in a short amount of time they get a recommendation that is aligned with what they were looking for at a price point that is amenable to them, then the conversion rates would be high. And you actually streamline the entire process of like, you know, awareness all the way to conversion. So, you know, it is a better experience and better experiences we know drive engagement and engagement in this case is actually leading to a conversion. So I would say early days, we're looking forward to more of the data, but really exciting times.

45:44Kevin McLaughlin:Great. Well, Varun, I want to thank you for coming on. That is Varun Parmar from Adobe here on TI TV. Awesome. Thank you, Akash. This is fun. That does it for today's show. A reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. If you can't make it then, episodes are available on theinformation.com, on our YouTube channel, or wherever you get your podcasts. Make sure to follow us on social media on X, on LinkedIn, on Instagram, and on TikTok. I'm already excited for our next show tomorrow. Have a great rest of your Wednesday. Bye-bye for now.