Box CEO on Layoffs & AI, Cerebras’ IPO Price Bump, Robinhood Co-founder’s New Company

31:27So we're seeing some companies just really have to be able to go fund the token expenses or then take those dollars and reallocate them to other higher growth areas, I think, as we've seen in maybe the Cloudflare example. But that's how I would sort of think about and characterize it. That's not what we're seeing or doing in our business. and I have lots of friends that are - I was going to ask you, but tell us about how you're managing it at Box. Yeah, I mean, on our end, there's definitely areas where on an incremental basis, the dollars are going more into one org versus another because we might be getting more leverage from AI in one spot, taking that leverage that we're now getting and then reallocating to another function.

32:07As an example, we are hiring meaningfully in areas like go-to-market. So if you think about sales reps or customer support or customer success managers or consulting, as AI rolls through the economy, our customers need a tremendous amount of help and support of how do you actually adopt these tools? How do you enable them and embed them in your business processes? So there's actually going to be more people needed than ever before for that side of a software company when it's actually getting rolled out into an organization. Right. And then there are other areas where we've decided, okay, maybe like the incremental dollar, again, has to go more into some of those functions than a different one.

32:50But I think on an absolute basis, I can't think of many orgs that will be sort of smaller in total size in two years from now, simply because there's just still so much to do within our organization and the things that we have to go work on. I mean, you mentioned consulting, and this is one of the things that – among the many things that you've written on X, your points of view on software and the sector. You had a post last month actually that caught my attention about – I think it was OpenAI partnering with consulting firms. And you made the point saying, look, anyone who thinks that companies are just going to do away with people and know how to implement AI.

33:31I mean, take this as an example that that's not going to be the case. You know, I do want to ask you about another post you put out there. I'm going to read this part of one of your expos. As agents become the biggest users of software, then all software has to be available in a headless fashion. Agents won't be using your AUI. They'll be talking to your APIs. Can you expand a little bit on what you meant there by a headless fashion? Yeah, I mean, this is the, you know, one of the, I mean, there's like multiple trillion dollar questions right now in technology. This is one of the trillion dollar questions, which is in a world where you have AI agents that you either deploy in some kind of completely background fashion where you or I never see them as end users.

34:14They're kind of just always on and running. Or where you or I go into codex or co-work and we ask a question and the agent goes off and fans out across systems and does some work for us. In both of those modes that I just mentioned, we as users of the agent or the underlying tools never see a user interface for the software that that agent is consuming. So we're no longer pressing any buttons. We're not clicking on any parts of the interface. And the agent certainly doesn't need to either because the agent is able to sort of say, oh, I understand the API of this company. I understand the structure of how the data models work.

34:51And I can go and interact with this in a kind of an API interface. So in that world, then the sort of typical way that an enterprise software company thinks about its monetization, the way that we go and build our very technology starts to subtly shift. because you might have an agent that could do the equivalent of what maybe a thousand people would have done within a single seat now. So imagine if I have an agent that's sort of just constantly sort of combing through data or processing information or executing tasks, and I let that agent run in parallel and 24-7, and it can work at kind of hyperspeed.

35:31Well, then obviously that wouldn't make sense as sort of a single seat of that software anymore. You'd have to have a new way to monetize that that represented all that utilization of the software. Which is sort of like this whole usage-based pricing movement. Yeah. So I think the way that this plays out and kind of what I've laid out is – and it's not particularly like that crazy or counterintuitive. It's kind of the natural evolution that one might expect, which is for you or I, we'll probably continue to have seats inside of a lot of the software that we use. like we will still have seats inside of Slack and inside of Box and inside of Salesforce, because there's a lot of reasons we have to go into those tools and do work ourselves as well.

36:12When agents are kind of working on our behalf and they're doing, let's just say, a kind of a normal, you know, amount of work for us, maybe it's like three times what we currently do, but it's not a thousand. I think I would expect that most software vendors have to let those agents use our seats in a completely kind of included fashion. So we can go and have an agent go and do work for us again in Codex or Cowork, and it will execute some tasks. And I think most software companies will sort of say, okay, that's actually something included in your seat price going forward. But then you're gonna have a lot of agents that either do an insane amount of volume of tasks, well beyond what you or I would be able to deploy just a few agents for, or you'll have scenarios where it doesn't make sense to have a seat because the task is not sort of owned by one individual.

36:57It's sort of owned by the corporation. It's a background process that is sort of always on or happening that we're executing. And so if you think about that kind of work, that should almost all be consumption. So you should just basically say, hey, if I have an agent do 5 million things inside of Salesforce, I should pay for the equivalent of 5 million API calls or agentic API calls or whatever that looks like. And so I think you're going to see a new component of the business model of enterprise software, B, what is sort of the headless API usage of these systems? Now, some software companies are already sort of built for this, by the way, because we've been doing PaaS business models for a while, which is platform as a service where you can already use our APIs in the background inside of other software.

37:41Agents just are sort of a force multiplier because the scale of this could be far larger than any amount of kind of PaaS usage we've ever seen. Well, so let me ask you a question So in this usage-based model, which, look, this is nothing new. I mean, usage-based pricing has been around for a while. I mean, that's one way that enterprise software companies will be able to keep people paying for their products based on how much they use. The flip side to this question, though, is, I mean, you know, in terms of pricing right now, pricing right now in a lot of ways, certainly for newer companies, it's subsidized.

38:15It's lower than the actual price will be. And we, you know, we've reported here at the information of about companies raising their prices, not the least of which is Anthropic. You mentioned Cowork, you know, something you guys use. So I'll talk about your own experience in a minute, which I'm curious about. But just broadly speaking, I mean, when prices do go up, you know, what evidence do we have that people will still keep paying for them? um well uh i would say this is sort of a unrelated point to the headless api side would you uh just so i'm clarifying right yeah yeah i'm talking i'm talking now we're talking i mean consumption based pricing is one thing i'm i'm now talking about just pricing as a whole as a separate yeah well well because i i think and this is going to be these are going to get lumped together a while for a while consumption pricing can be can be uh there's two categories of consumption pricing There's consumption pricing on just the API calls of our underlying services.

39:12The API call in the box, the API call into Workday, the API call into Salesforce. And that's a sort of a predictable pricing model. It's mostly hitting CPUs or storage environments. So we kind of know how to model the cost structure of those types of API calls. There's a different part, which is the subsidization of tokens, which is much more of a constrained resource right now. Now that's obviously a GPU capacity and the kind of surrounding components. And there has been some business models that have been subsidized where the underlying kind of tokens are being paid for by the vendor, obviously, to try and get market share or kind of prove out use cases.

39:55Then now, ultimately, probably prices have to rise or they're rising as a result of scarcity. And so then that's just market forces playing out, which is if you have a scarce resource, so you can charge more for it. And then you kind of find like, what is the market clearing price for where you start to, you know, lose customers at a rate, you know, higher than the sort of revenue you get by raising those prices. So I actually think we're just sort of seeing microeconomics play out, you know, at a big scale. I'm not, I'm not, I don't think there's any kind of meaningful, you know, thing that's happening other than that, which is prices will rise up to the point where volume sort of drops.

40:33And at least the rumors I've seen, the gross margin on inference of the major frontier labs is actually like very, very high quality gross margin from just an absolute number standpoint. So obviously you have to pay for things like the training costs. You have to pay for things like the infrastructure build out. But I don't think this is, I mean, you can see in Anthropics top line numbers and OpenAI top line numbers, clearly clearly we keep paying for the AI, which is effectively revealed preference that it's working and we are adopting these tools. And if you go to most engineering organizations, you would never be able to rip out AI from them at this point, simply because the productivity gains are too high.

41:17I realize that asking this next question is sort of asking to pick favorites in some way, but Co-work, Codex, Cloud Code, Codex. Which suite of products are you using more right now? The battle of our lifetimes. I have two icons right now on my Mac trade for each of those. And I like to be pretty ambidextrous, partly for my own use cases, but also because it's really important that we at Box understand the ways that the people will be working in the future. and we want to make sure that our platform works extremely well with both of those as well as others. I mean, I have perplexity computer running in a tab oftentimes.

41:57So there's three, five, ten tools that I'm personally using. I unfortunately will not be able to pick a favorite amongst those at the moment, but I think we're in an incredible moment because of the pace of innovation and the sort of pace of product development that's happening, where ultimately us as customers and consumers and prosumers are just winning because of how much innovation is happening. Well, okay. So, and so maybe let's go back to sort of your own usage and the levels of usage then of these products and all sorts of products. I mean, there is a question around how much budget you can allocate as a company to using these tools.

42:34And we've heard, I think it was the CTO of Uber told us that we've kind of hit our budget now for using tools in some cases. And then you contrast that with other companies that token maxing is the alternative. So I am curious around, did you go about setting a defined budget at Box? And have you blown through that budget yet? You know, we're only a quarter into our financial year, so I probably can't talk about where we are in our budget. Just for AI tools. Just for AI. I very much get the question. So here's what I'd say. First of all, probably only a few companies on the planet can afford to sort of token max.

43:21So I think that might be like a great fringe benefit at Meta and maybe a couple others. Mere mortals obviously have to balance the overall annual budgeting process and how we allocate spend across the organization. And that's the kind of stuff that the rest of us have to do. We are, I think maybe the bigger picture thing that will be super interesting is right now, AI probably started first as an IT expenditure. And so that meant that AI sort of had to fit in within three to seven percent of your corporate sort of of of spend as a percentage of revenue. You know, IT kind of runs at three to seven percent of revenue in across the economy.

44:10Some some businesses less than more that that's probably too constrained of a of a line item to be able to really get fully augmented productivity across your organization. So what's going to happen, and the interesting thing to watch over the next couple of years is the jump from when AI moves from an IT expenditure to an OPEX across the general sort of budget planning cycle in an organization. And when you as a department head have to decide, do I want to add one or three or 5 % of my budget to AI compute or maybe even more? And what that looks like from a resource allocation standpoint. So this year, we got ahead of that a little bit in especially engineering.

44:49but certainly as you have things like codex or coworker others that are pretty, you know, AI, you know, compute intensive, you'll have to do that in other areas of knowledge work. You'll have to decide, you know, what am I going to do in my marketing budget when, when I want agents running and producing a lot more marketing collateral? What am I going to do in, in legal or financial, financial operations when again, we have, we have way more of this augmented workforce. So I think, I think this is a trend that's going to start to shift. We're probably even going to need new software just for that problem because in normal resource allocation, you have these sort of one-time fixed expenses, which is like onboarding the new person and then their new salary that is sort of sustained.

45:31In AI compute, you kind of go up and you go down. You can kind of quickly tell somebody to spend less. You can tell a team to spend more. So there's a much more dynamic sort of form of how you do budget processing with AI. And I don't think a lot of our kind of existing tools probably help with that yet. And then who owns this? Does the CFO own it? Does each organization own it? Does the IT leader own it? Many fun questions that we're going to be confronting with over the coming years on this front. Right. I just want to go back, Aaron, before I let you go back to that pricing discussion, because I just want to make sure I understood what you were saying correctly.

46:05I mean, look, my sense was, as we've heard about companies starting to raise prices for AI-related products, I mean, my sense was that was a reflection of them, you know, really trying to make the price more reflective of what it is costing them to produce the service. And I mean, it's not just Anthropic. You know, we've seen other companies also suggest that they will raise prices. And so, I mean, if I heard you correctly, I mean, were you suggesting that basically the margin is healthy enough and they won't have to increase it as much? Or the question I was trying to get at is what evidence do we have that people will still keep using these products if the prices keep going up?

46:47Well, I would just say as of May 8th or whatever. We're recording this on Friday, right? Yeah. Yeah. So as of early May, it's very clear that we keep paying for these tools. And I don't get the sense that prices are going to double or triple from here. You know, it's important to also have the backdrop of how compute constrained as a factor that this all is driven by. So if you imagine a world where there's vastly more compute, if you imagine that you get more efficiency in the actual sort of AI at the software layer of AI and you get hardware efficiencies in next generation kind of NVIDIA or TPU or Tranium kind of computer approaches.

47:37I think you have multiple factors that can bring down the price of AI. Not to mention, there's this really great kind of counter pressure that will always be out there, or at least for the foreseeable future, which is open source. And so open source kind of is this nice kind of, you know, lever on the whole on the whole space, which says, which says at any moment, I can peel off and get, you know, sort of frontier as of three to six months ago, and be able to run that on the lowest cost stack that I can go find. And so as long as open source is around and kind of out there, you do have this nice, you know, again, ripcord that the customer can pull.

48:15Now, not all customers are sort of savvy. Are you using open source at Box? We are, but not for the frontier type capabilities. So we use open source and things like our embeddings models or like things that are really kind of background components of our AI stack. for the frontier, which is I want to be able to go generate a PowerPoint presentation, or I want to be able to process a loan document and extract the critical data. We do need models like GPT 5.5 and Gemini 3 Plus or Opus 4.7. These are the model families that we work with for those types of use cases. But there's lots of stuff that over time, you will be able to peel off and put it into an open source or a much cheaper model.

48:59And we even stratify within our use cases where something will go to Gemini Flash, as an example, because we have, you know, obviously much lower cost tokens and the use case can be sort of adequately or very successfully served by those models. So, you know, in five years from now, you will have, when you look at the sort of token mix of a company, it will not just be that all the dollars are going into the frontier at that particular moment, you'll see a full, you know, kind of pie graph of where the different use cases sort of land, just as actually knowledge. What does your pie graph look like if you were to say open source versus frontier?

49:36I mean, you know, just give us a view of what Box's operations look like right now. Today, it's probably 80 % frontier. But I think what would happen is if I had to guess in three to five years from now, whether it's open source or from one of the frontier labs, as a lower cost model is sort of immaterial. It's all that matters is just like, what is the cost of the token? I think you would imagine that in five years from now, it would be much more like, you know, 30 % of our token spend goes to the truly frontier type model. And then sort of from there, you're like, okay, you know, another 20 to 30 % goes into, you know, some just behind frontier type class.

50:18And then another 30%, 40 % goes into these models that are just doing easier compute operations for us in the background. Now, again, what's amazing about AI is that lower tier model that I just laid out as an example is probably three times better than the best frontier models that we have today. So like what? Right, because everything will be, yeah. You're not saying 20 % open source. You're saying 80 % frontier, and in that 20 % is something that is maybe a little sub-frontier, and then open source percent is probably even smaller than that. Well, today I was sort of giving you the rough allocation.

50:53In the future, I just think it's probably – just for lack of any better way to think about it, I'd probably cut it in thirds. And probably the second and third thirds could be solved by open source. or again, if you're a lab and you wanna stay very competitive, you probably would have an equivalent model that you're running at just a lower margin. And that's your way of sort of making sure you don't get flanked by some other provider. So I think that's sort of how I expect to play out. Now, a great example, again, of a company already roughly delivering the shape of strategy is Google, where Gemini Flash is meaningfully cheaper than Gemini Pro.

51:30And you use it for totally different use cases. You're gonna do your orchestration and your planning with something like a pro model. But if I wanted to do really, really high volume data classification where I have to like look at, you know, 10 million documents that are being piped through a system very quickly, a flash model or an instant model or a thinking low model, you know, each vendor, each frontier lab has a different way of sort of framing this. That'll be totally more than adequate for that type of process. And so by volume of token spend, you'll just allocate the tokens differently based on the type of workload.

52:04And again, it's sort of, not to anthropomorphize this too much, it's kind of how a company already is structured. You have some things that you say, okay, this is the very rare, highly complicated task, and we need to put as much of our kind of capital from a human capital standpoint into that area. And then there's other areas where you're like, okay, there's a little bit more high volume. It's not as competitive. It's price competitive for the buyer of that type of task, and then the costs are a little bit lower. Right, great. Great. Well, Aaron, I want to thank you for coming on. As always, that is Aaron Levy, the CEO of Box here on TITV.

52:41Our next segment is with our sponsor, PwC. Tech companies left, right, and center are rethinking how they staff teams in the AI era. PwC has a front seat to that work as it advises some of the biggest corporations in the world. I want to bring on Dallas Dolan, a leader at the firm's tech, media, and telco practice, to share with us what he is seeing. Dallas, welcome back to the show. It's great to have you here. It's great to be here as always. Good to see you. Well, so look, one of the big news stories of the last couple of weeks have been all of these headcount reductions that tech companies have embarked on.

53:15And they've given a variety of reasons. AI is central to a lot of the messaging around it. My question for you is, I mean, how much of these restructurings do you think are because of AI and how much do you think it's because people just overhired?

53:32Dallas Dolen:Yeah, it's a great question. And it's certainly something a lot of people are asking, Akash. I think it's a convenient, we'll call it perhaps excuse to go blame AI immediately for everything going on. At the same time, there's definitely some truth to it. And it just depends on where the reductions actually are. If you're going through and clear cutting a forest saying, you know, oh, we had, you know, a small problem over here probably isn't the reason to go do it. And I think that's the place that a lot of CFOs and candidly a lot of the folks in the technical arenas, whether it's in the product development or the IT, you know, in related, you know, core engineering teams are probably at the most, you know, loggerhead, so to say, around the what and the when things should, you know, be progressed, right, as it relates to the use of technology and the, you know, the resulting headcount.

54:18Dallas Dolen:dynamic as it relates to the same. I think what you have is, in particular, the CIO and the CTOs who are under a ton of pressure, especially around cost, just in general, because the cost of everything has gone up. I mean, especially tokens in the last couple of weeks, as we've seen, and you guys have reported really well. That in and of itself is putting a lot of pressure on budgets, coupled with the possible areas of noting, hey, there's a lot of engineering tasks that can be completed by AI. It's a really good area to focus on. It's also a really good quote justification, you know, to do some, it's springtime, I guess, technically, to do some spring cleaning around it.

54:55Dallas Dolen:And I don't make light of it. I mean, it's a real significant issue and a real serious issue for a lot of these companies who have probably delayed, you know, some of the housekeeping on just overall, you know, we'll call it human capital. And, you know, what is the structure of the organization as they still, you know, look to get fit coming out of, you know, probably five or six years of uncertainty, that started in February, March of 2020. I am curious, you work with a number of companies at the scale at which we're talking about. And look, managing headcount is something that is, it's a constant endeavor, something they're evaluating every quarter, every year.

55:33I mean, I wonder just broadly speaking, you have an executive that comes to you say, hey, I'm at that time of year again, or we're thinking about this matter of managing headcount, what are the key questions that you ask them to answer for themselves insofar as in deciding whether or not it's the right time to do it? Because, of course, there's the other flip side of it, which is that you cut too deep and suddenly you don't have capacity to service your customers.

56:00Dallas Dolen:Yeah, well, and that story is a real personal one for us at PwC, too, because we've actually had a couple of experiences over the last 25 years where in anticipation of slowdowns or reaction to slowdowns, we cut and then you end up not having enough people, especially in the middle part of the business, you know, to actually help run things. And that's a real concern that you might have. I, you know, we've had this conversation with a number of companies. I've personally participated in various conversations, including last week, did a really wonderful roundtable with a number of CIOs and CTOs here in the Valley, you know, on that topic.

56:31Dallas Dolen:I think, you know, they're going to be responsive, right, on a personal level and professional and we'll call it fiduciary level to the asks that are coming to them. So the question is, can they also push back in an adequate and forceful fashion in order to say, hey, listen, like, I can cut as deep as you want, but there is risk to that and explaining that if I were to get rid of, you know, the middle part of my team, where I think the most maybe efficiencies can come from as it relates to technology. And of course, you know, the whole 10 times concept, we haven't seen that, you know, talked about a lot recently, but the ability to make a lower quality engineer or a higher quality engineer, you know, and it might cost a quote, a little bit less, again, depending on the cost of technology.

57:12Dallas Dolen:You know, what happens if you lose all the 10 time people, right? I think those are the two areas and manners of pushback that you really need, because history does tell us that when you cut too deep, there's always a reverberation from that and you always pay for it in the end. And that's the real concern that we have to guard against. And the real considerations that I think, especially on the technical side and on the engineering side, the most push is going to come. But we're going to see that, I think, in the sales and marketing organizations as well. And certainly in mine, too, in the services businesses.

57:44Dallas Dolen:I mean, we are thinking long and hard about this as we go into planning for our fiscal 27. And it's a real consideration as to where do we need talent? Not just now and not just in fiscal 27. So like looking at 18 months, but also where do we need talent in fiscal 30, right? Because the decisions we make today as far as inboarding and onboarding, you know, people who are sophomores and juniors in college has a really dramatic effect on the way that we can run our business, you know, looking at up to five years. I want to ask you about a partnership that PwC signed last week with OpenAI. Tell us a little bit about the ambition there.

58:19Dallas Dolen:Yeah, sure. I mean, I think we look at all these frontier model companies and really all the tech companies, who are building in this space is a really great opportunity to lean in and both figure out two things. One, how do we deliver our services better and with technology? And so the one coming here with OpenAI is a good example of that as far as the future of finance. But also if you crawl back into time, looking at different industries, the opportunity to do that with Anthropic and looking at cyber and things that we've done recently with Google, and then even looking back a little further, you know, some of the more automation and engineering space and things that we've done with Microsoft and others, all of these are opportunities for us to look at our, again, our ability to deliver and how we're delivering right from a capacity and competency perspective.

59:04Dallas Dolen:But we're also looking at our clients who are saying, hey, if you show up with, you know, a team of 10 people and say, you're going to do something, we're going to view that as not evolving and not embracing the business model of the future. And so we look at that relationship, you know, with OpenAI and the future of finance, you know, being a really exciting one to build out the what does it look like? How do you run a finance and accounting team in the future? Same thing goes for all those other organizations as well. And that's, I'll just maybe tease a little bit here, Akash, that that's probably the first of a few that are coming.

59:34Dallas Dolen:And it's a really exciting next couple of months for PwC and our journey, especially with the model companies and the hyperscalers. Some really great stuff is coming. I do want to ask you one last question on this topic. So all these AI companies, not just the model companies, but application layer companies, I mean, everyone is now creating teams of forward deployed engineers, which are meant to go in and help their customers implement this software. And that is a role that traditionally consulting firms have held and continue to hold. I mean, you know, we've seen there's a lot of data out there about how consulting firms are getting heavily utilized right now.

1:00:16But my question for you is forward-deployed engineers. I mean, do you see them as competition, you know, as a consulting firm? If, you know, if a lab has their own team of consultants that goes in and helps people implement all this stuff?

1:00:30Dallas Dolen:Yeah, I think you have to look at all of it as a competition. The landscape is evolving in a way that, you know, no one's really anticipated. Certainly, if you look out, you know, looking backwards a couple of years, I don't think anyone saw this as being a likely, you know, source of competition. Technology, though, Akash, has always been a source of competition for, you know, the consulting businesses in particular. I think what will be unique about this one is, you know, whether or not it'll be something that is a direct as in, you know, it's us versus a technology company who has a professional services org, you know, with the forward deployed engineer, or will we replicate and have the ability also to compete?

1:01:07Dallas Dolen:It's one that I don't think the story has really been written yet, and it's going to be how fast can, you know, say all of us in that space adapt. The one differentiation, I guess, is that I think there's still the regulatory dynamic that's important to consider here as it relates to some of the things that, you know, we'll call it the big four, you know, do maybe in comparison to your traditional consulting firm. And then I think if you subdivide within consulting, that between, we'll call it strategy, the transformation space, and then the pure maybe engineering and that type of work, it's going to be differentiated there.

1:01:41Dallas Dolen:The stratas and impact are going to be really different. But probably the biggest thing - What is the regulatory? I mean, when you say regulatory here, do you mean the accounting work? Yeah, that's right. So like whether it's the IRS or HMRC on the tax side or whether it's the SEC, the PCOB, you know, and others on the accounting side and the mandate that they still have around what we call the human in the loop, the human intervention piece. You know, what can, for example, you know, the IRS's view on whether or not a taxpayer can rely upon advice that they've been given by a technology tool versus, you know, being given by a human.

1:02:17Dallas Dolen:Same thing goes for the rules that are out there around how do you document an audit and whether or not individuals, you know, have to be involved as, again, human in loop in the same. And, of course, there's also the issues of, you know, are you allowed to practice in this space? Yeah, that's right. Right, right. Well, Dallas, I want to thank you for coming on. That is Dallas Dolan from PwC here on TITV. 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, our YouTube channel, or wherever you get your podcasts.

1:02:54Make sure to follow us on social media, on X, Instagram, and TikTok. I'm already excited for our next show tomorrow. Have a great rest of your Monday. Bye-bye for now.