The Exchanges

Q Lauren, you've been inside a fab in Arizona. What's it like?

A Yeah, well, like, this was the Intel fab 52, which is their most advanced fab. It's a two nanometer fab, and it is on the level of what TSMC does in terms of two nanometer, but of course, at a fraction of the scale of what TSMC does. And to Max's point about, well, what is our redundancy or our resiliency plan in the event of an invasion of Taiwan? It seems like right now the US government is very interested in what Intel can do and its fabs, but, um, it's It's pretty small. Uh, and so, yeah, I went to Chandler, Arizona, um, saw Fab-fifty-two, um, you know, got the bunny bunny suit on, and, um, like Max said, it's, it's interesting to see, uh, everything is roboticized. It's like, if you had this, um, utopian fever dream of what the future of manufacturing looked like, and everything was white, and everyone was dressed in all white, and everything's, like, being, you know, moving around, they call them foops, That actually carry the silicon around above head, and then they just sort of automatically go down to get like etched and carved and stamped and everything. Um, it look, It looks like it feels and looks like science fiction, but from a reporting perspective, oftentimes the companies will cover up the names of the vendors on the different machines, and you may be able to get some insight. Um, for example, there, there were ASML, UEV machines in these giant bays the size of…

AI assessment note: “got the bunny bunny suit on, and... everything is roboticized. It's like... science fiction”

Q on a variety of different chips. And once people, once these labs train their models and they're happy with their models, most of the computing is going to go to these inference chips. And therefore, even if Nvidia has a bet there, there's not going to be able to sustain their dominance. Is that what do you think about that? Is that a potential flaw in the armor for Nvidia?

A I think it is a flaw, and I think there is, if anyone's going to sort of attack NVIDIA's dominance, they're going to do it on inference, like you said, and there is sort of a massive effort underway right now to make all inference chips under the sun work well, and so every single company that buys NVIDIA chips and is spending a lot of money on NVIDIA chips is trying really hard to make these other, other chips work, whether it's in-house chips from Google or Amazon or even in OpenAI's case and potentially Anthropik's case. You know, do we make our own inference chip? Um, so I'd have a hard time, you know, believing that none of those chips are going to pan out. But, you know, they might not tackle the bulk of the inference workload even. But then again, even if they do 10% of your inference, maybe you're saving enough money that you think the effort is worth it. And it gives you negotiating leverage with Nvidia. If they know that you have an in-house chip team, you know, Jensen's going to be a little bit worried when negotiating with you when you're playing hardball with him. So I think everyone's going to have to have an inference chip Answer to Nvidia. But when I talk to data center companies about what they're seeing, you know, some data center companies don't really care what chip goes inside of their data center, but they try to get hints from the companies that they're w…

AI assessment note: “I think it is a flaw, and I think there is, if anyone's going”

Q Like I think for For us, you know, we're deep into the tech world. It, it sort of makes sense to be wearing an AI around, but there is an argument, you know, looking at history, that it's, there have been attempts to do it, and it hasn't worked. Thinking about the Umane pin, for instance, or the Google Glass. So, what do you think is different about this time?

A I mean, I think the biggest thing is that, at the core, it's a good-looking, comfortable, stylish pair of glasses. I mean, it's the reason we're partnering closely with SLR Luxottica. They're our partner we've been working with for years. They're the makers of Ray-Ban, Oakley, they have many other brands, um, both owned and licensed. They have a number of stores. They are experts at lenses, experts at manufacturing, and working with them It's because they deeply understand what it means to have a good looking, stylish, comfortable pair of glasses that works well for people. And that is the core of what we're trying to do with the glasses. And so we've, we've launched the Ray-Ban Wayfarer was the first. Then we had other Ray-Ban styles. We've launched the Oakley Meta Vanguard, which are sport, the Oakley Meta Houston's, which are lifestyle Oakley's. That's just the beginning of what we want to be a really broad selection of brands and styles and SKUs. We want them, everyone's face is shaped differently. Uh, everyone has a different sense of style, and we need to make sure that they're comfortable, and people like how they look at them.

AI assessment note: “at the core, it's a good-looking, comfortable, stylish pair of glasses.”

Q out. You can just run them on device. So, you know, Meta is an interesting company because it's in the middle of doing that big build out, but it also has devices that I would imagine someday would want to run An AI model on device. Can you talk a little bit about that balance and sort of does one eventually cancel each other out or do you need both?

A Well, the, so the, um, servers and data centers that companies are building out, Meta included, and there's two kind of core functions that those serve. One is the training of the models, and so that certainly happens in data centers, and that's where it's gonna continue to happen, and that's where a lot of the investment is going right now. And the other, you know, as you say, it's running the inference. So it's when you're using the model, where are those models running to Answer the prompts or to agentically help you with tasks. Um, the, the former, I think the model training, I think is going to consistently run in data centers on where inference is served. Um, there's a number of reasons to try to move that to devices that are not in data centers. Um, one is the latency and reliability will be better. You know, you've used apps on your phone and the more things are local. The less network interruptions or low bandwidth or, you know, packet loss can cause issues. And so that's one reason to do it. Second is cost to the user, you know, and you're paying the fees to whether you're, whatever assistant you're using right now, you know, a lot of those fees are because it costs real money to be, um, serving you.

AI assessment note: “One is the training of the models... And the other... running the inference.”

Q Um, can I just ask, like, why not just distribute A book with all the rules and regulations. Is that so stupid? I mean, like, talk to me a little bit about why this needs to be so sophisticated from a technology standpoint.

A Because that book would be 10,000 pages. I mean, because our, our, well, because especially in HR, like, the rules and regulations and what applies to you is so unique to you. Um, the California laws, from an employment perspective, vary sometimes by county, by city, and so, um, we can put that into a system and that With ServiceNow, and it can tell that you live in this city, in this state, so this policy applies to you. So when you're asking it a question of how much PTO do I have, it knows specifically, okay, this all applies to you as an individual, where if you put that in a handbook, That's a piece of paper. That is, it's not a dynamic response, right? It's a one-size-fits-all, stagnant, it's gonna die, you know, it's so hard to maintain.

AI assessment note: “Because that book would be 10,000 pages.”

Q So now some in the field believe that superintelligence is close, and you've already said this is moving faster than you expected. Do you believe that?

A Um, I don't know how close it is. I think unless we blow ourselves up, um, I think it's going to come. Nearly all the experts believe we will get super intelligence. They just differ on how long it will be. So not that long ago, Demis Hassabis thought it might be 10 years. Um, Jan LeCun thinks, um, unless you do it his way, it'll be much longer than that. But if you do it his way, I think he thinks we might get it in some reasonable length of time. I think we'll probably get it within 20 years. That's all I'm happy to say at present. Dario Amodi thinks it might come in a few years. Elon Musk thinks it might come maybe next year, I think. Um, so there's a big variety of opinions on when it'll come, but not much disagreement on that, that it will come.

AI assessment note: “I think we'll probably get it within 20 years.”

Q I've always wondered, because then you came out and recently, like we talked about, 20, 23 and said, you're concerned about where this is going. And I've always wondered, after seeing you make those statements, what do you think it is that you didn't anticipate in the beginning that you ended up where you are today? You know, isn't this kind of what you wanted?

A It was a combination of two things that made me realize how dangerous this stuff is. One was seeing the chatbots, particularly ones produced by Google before OpenAI. That could understand why a joke was funny. Um, that had always been a criterion for me of, do they really understand? If you can understand why joke's funny, you have to understand quite a lot. And they were very good at understanding why joke was funny. For example, um, in 20, 23, when I went public, I got lots of requests from Fox news. And I started off just replying Fox news is an oxymoron. Um, but then I left a gap between oxymoron. And so then I asked, um, I think it was GPT-IV why that was funny. Might have been 3.5, but I asked it why that was funny, and it understood why it was funny. Initially, it thought the gap between oxymoron was just a typo. So it explains that Fox News is an oxymoron, is saying it's not real news, it's just a drug. Um, sorry, it's just, um, nonsense, it's not real news. But then when I told it, what about the gap between oxymoron? It said, ah, that's an extra layer of humour. Um, it allows you to use the word moron, um, and also the oxy implies that Fox uses a drug. So it understood all that. Right. And that was, um, it's that level of understanding that worried me. Um, the other thing that worried me was up until the beginning of 20, 23, I'd always believed that making, um, these …

AI assessment note: “It was a combination of two things that made me realize how dangerous this stuff is.”

Q from open AI and Apple has not participated, uh, in this moment. They said Microsoft and Apple come from real innovations and use cases, executing LLMs Is expensive, risky trash for grifters, and they know it. There is no paradigm shift without Apple and Microsoft on board, so why should they take the risks of building these models driven by stolen data? This stuff is toxic, and they know it.

A Could you find someone who has a stronger opinion and is less equivocal in their commentary? Um, let's see. Uh, it's unclear whether that was a choice Apple made or a, a, Inability to, um, do whatever they needed to be doing. Um, I've found it It's, it's surprising how Siri for, Siri or iMessage, for example, have been, been so, um, impossibly slow to keep pace with what appear to be things that, um, AI startups with 10 or 12 people have been extraordinarily successful at, like text to speech or speech to text. Um, Granola, for example, does a 100 times better job of interpreting what I'm saying than Siri does, which seems, I would, I would imagine has Um, orders of magnitude, more people and resources focused on it. So I don't think it's that Apple has seen that what OpenAI and Anthropoc are doing is, you know, grifting. I don't think that's the case at all. And I think, I think Microsoft pulling back from the OpenAI relationship probably has more to do with Satya's, um, understanding of the dynamics that are going on Inside OpenAI and with the relationship between him and OpenAI and maybe his perception of what he believed to be true and what is actually true and his understanding of where he wants to take Microsoft.

AI assessment note: “I don't think it's that Apple has seen that what OpenAI and Anthropoc are doing is”

Q But if you think about the brand names, people say Claude Code, Cowork, um, they don't say a combination or a medley of your favorite models working together, so is that a specific instance with ServiceNow or is that just the way these things are being built?

A Um, well, you're just interfacing with one UI when you're dealing with a cloud code or an open work. You actually don't know the orchestration that's happening behind the scenes, and I'm not going to speculate on, on what they're doing, but I could talk about some of the stuff we're doing between us. Uh, we build, you know, today they were talking about blueprints, right? How to, how to, the blueprint for Agenda KI. One blueprint that we have is, we call IQ, that does deep research, right? And so, Sometimes, you know, it's not when you just need to ask her a quick question. You want to go deep to understand what does it take to resolve this issue? What is the context behind it? What are different data sources I got to look at, synthesize, and understand? Um, that deep research blueprint or, or agent is actually made up of, of no less than seven agents, and bear with me. There's an orchestrator or agent that is kind of like the team leader that's there, right? Um, we've seen some of the best results with using either Opus from Anthropic or GPT for OpenAI for the orchestrator. Next to the orchestrator is also a planner. Um, so this is almost like an assistant that sits next to the team leader whose sole job Is to maintain tasks and to-do lists and to cross those off, right, as those tasks are accomplished. Underneath those layers, you have a whole bunch of sub-research, sub-resea…

AI assessment note: “you're just interfacing with one UI... You actually don't know the orchestration”

Q What are some of the use cases that you're seeing? I mean, I, my, my email, I thought, I think is pretty fun. I let that go. Um, but I also see taxes.

A Yeah. I mean, it's, it's any, uh, so we actually are launching this week, 36 different workflows that go on top of computer. Um, so this is everything from building a financial model, uh, of a company, uh, to filing your taxes, uh, if you're a wealth manager, uh, prepping for a meeting with a client, um, and again, this takes advantage of connecting to, you know, your internal data systems, your, your, you know, your Snowflake, your Databricks. Uh, just last night, Uh, I ran a, uh, analysis of, you know, what are the models that are being used inside of perplexity right now? Like, what's the distribution of between, you know, Opus four seven and, uh, GPT and Gemini, and it got a very elaborate, uh, result back, and I know zero SQL. I don't, I can't code if my life depended on it, and I didn't bug a single data scientist at perplexity. Um, and I was able to do this because we connected perplexity computer, uh, to our snowflake, and I was able to, you know, pull in that, that analysis, you know, within a few minutes that in a previous world, you know, that would have been 10 emails, and I certainly would not have been able to get it at midnight, um, as, as I wanted to kind of dive into that, right? Um, what we're seeing people do is, is be able to operate with much greater velocity, uh, whether they're accomplishing marketing objectives, analytical objectives, like building produ…

AI assessment note: “everything from building a financial model, uh, of a company, uh, to filing your taxes”

Q Let me bring up one more headline. Um, there's a story this week that also says that you had some XAI stock. Do you have any SpaceX? Is that a conflict?

A No. I sold all my SpaceX. Okay. And I, so what happens when you take one of these jobs, you, you show your whole, whole sort of list to the Office of Government Ethics, nonpartisan. They go through it and they say, we think these things are, these things are red lines. You sell your defense company stocks, which you didn't have much. SpaceX was on the list, so you have to sell that. Um, and then depending on your role, here's the things you have to sell that might be specific to your role. And then based on the kind of connection to that, you could recuse yourself from dealing with the company. So I just recused myself from dealing with XAI until I could sell. And I was pretty, you know, active about it because I didn't have the AI portfolio until the fall.

AI assessment note: “No. I sold all my SpaceX.”

Q If I wanted to build, let's say you hadn't built creative director, I wanted to build creative director. How do I sort of onboard that type of experience in Asana?

A Absolutely. So the way you would do that is, uh, you would like type in a little prompt that says, I want to build an AI teammate that is, uh, that is a creative director. Uh, uh, the builder, uh, chat AI agent will come back to you and be like, Hey, so what kind of, um, you know, projects and portfolios do you want this creative director to have access to? Because let's say you'd hired a real human creative director onto your team, you'd probably have an onboarding plan for them, right? You'd probably tell them to, uh, read these documents about the company's mission and values and, uh, look at these projects and tasks for how the creative team has worked in the past. So you provided with that context.

AI assessment note: “you would like type in a little prompt that says, I want to build”

Q executives who've appeared as guests include meta platforms, Mark Zuckerberg, Microsoft Satya Nadella, and OpenAI's Sam Altman. All right, I have some thoughts about whether this is a good purchase or not. Um, but Liz, I'm curious to hear your perspective. Um, can you explain what TBPN is, because there are some people that still don't know, and whether you think OpenAI is making the right move in buying them?

A Yeah, it's a really popular tech talk show. I mean, it's on air live for like a lot of hours every day. Um, and has, has actually has a real following in Silicon Valley has real executives come on. Um, I don't know whether, you know, they would call themselves journalists site and be an interesting question to ask them. Um, they are clearly, there's no doubt about what side they're on. They, they want The people on their show to succeed and, and it's generally a fairly light interview. Um, but they've, they've like captured a real portion of the sort of tech media brain space. Um, do I think it's a good move? I don't know. I mean, Sam Altman said he bought it cause he likes it. And, uh, and you know, one way to look at it is that, you know, billionaires buy media companies because they're influential and they like having a megaphone, you know, Rupert Murak owns a lot of media, uh, Mark Benioff bought time magazine, obviously Jeff Bezos bought the The Post. Jamie Dimon gave an interview recently where he was saying maybe what he'll do after he, uh, after he retires is start a media company. Um, you know, it is a sexy, fun thing to own. Um, you know, but, but also I think tech and particularly AI is sort of looking around and realizes that the country is not with them, uh, and that they have some questions that they need to Answer and some real pushback is starting to emerge. I d…

AI assessment note: “do I think it's a good move? I don't know.”

Q jobs, I think is a little bit overblown, but you had a very interesting conversation with Larry Fink, CEO of BlackRock, about a few things, and first is AI jobs that he talks about. He says, AI is going to create many jobs, and we're not prepared as a society to fulfill those jobs. That's a crisis. That's an interesting perspective. Can you talk a little bit more about that?

A He was really talking about blue collar jobs, um, that, you know, which were already really in, in short supply of the, the people to fill them talking about welders and electricians and, um, you know, which is in some ways like the last jobs that the robots will take that, you know, HVAC plumbing, that stuff seems hard for AI to do in the physical world. Um, and it's like acutely needed now to build all the data centers that BlackRock and others are, are financing and, and throwing up to, to service, you know, open AI's Compute needs and all of that. Um, you know, but on the broader jobs front, I guess it is a lot easier to see, and maybe this was true of past, you know, industrial revolutions. People talk about this as the fourth industrial revolution. It's a lot easier to see the jobs that AI will kill than the jobs that it will create. Um, and particularly in the knowledge economy, these are, you know, I've spent the last 15 years covering Wall Street, and these are big pyramid businesses where a lot of, you know, they hoover up these kind of young graduates from, You know, they're from the IVs and from business schools, and they spent a bunch of years doing, like, kind of, dredge work, like, consulting, investment banking, that kind of thing. A lot of PowerPoint presentations, a lot of spreadsheets, and I can actually, today, do almost all of that. Um, and so, you know, th…

AI assessment note: “He was really talking about blue collar jobs, um, that, you know, which were already”

Q Definitely. Have you been out there? Have you spent time out there before?

A You know, it's funny, I, I have, uh, I was in Saudi Arabia last year, I was in, in Abu Dhabi the end of, um, in December, and I remember flying in, and I don't know why it hadn't occurred to me, but you're flying into Abu Dhabi, and out the right side of the plane is, is the Arabian Peninsula and, and the Emirates, and then the left side, right across the, across the Gulf, is Iran. I mean, it is, it is, when you go, it's, it's easy to forget, because you're in Abu Dhabi, and there's a Galleria, and there's a Four Seasons, and like, it It feels, um, very Western and there's a lot of rich people in suits walking around, but it's, it's like, I, I heard about this a couple of weeks ago. It is a nice house in a bad neighborhood.

AI assessment note: “I have, uh, I was in Saudi Arabia last year, I was in, in Abu Dhabi”

Q And why are people making this run? Is it just behavior or talk about that?

A It is partly behavior. Um, it's partly look like for the first, you know, six years it was around blue algae raised money from sophisticated institutional investors. And it's not just Blue Owl. It's Blackstone and KKR and Apollo and all of these HBS big firms. And then they said, well, we're kind of tapped out there, but there's a whole bunch of individual people. What if we sold it to them too? And because people live their lives differently than institutions, you have to give them this quarterly liquidity. So this is sort of like they were, this is a consequence of their own actions and their own desire to keep growing and to find new, new places to raise Money. Um, the question of whether the loans themselves suck, that is a fair question. We don't actually know. A lot of them are, certainly there's this vintage of lending from, call it, twenty-twenty-one and twenty-twenty-two. Probably some real bad loans there. They raised too much money. This was the, the sort of peak software. They were making these loans, I don't know if you've talked about them on the show yet, called, um, ARR loans. Usually banks underwrite a company's ability to borrow based on its profits. But tech companies, especially fast growing ones don't have profits, they have revenue. And so they based it on annual recurring revenues said, well, this company's making a lot of top line money. Let's lend them.…

AI assessment note: “It is partly behavior. Um, it's partly look like for the first, you know, six years”

Q Yeah. That's the, that's kind of one of the hypotheses we've been playing with on the show now is did Apple just do it right where Apple becomes the infrastructure for AI or it happens on your phone? And on the Mac mini and the models become so efficient or purpose built that you don't need the data centers.

A It may very well be that that's where this lands. Having said that, I would imagine that this should be the greatest opportunity for Google and Alphabet Effect or Alphabet to truly take share because Gemini, if built the way I think it should be built, should be able to move you around their phone in shocking ways, right? It should be able to move into any app, control any app, do everything throughout the phone. I can't imagine that Apple, which really has made its name around privacy and controls and sort of a walled garden, Even under this new deal that they're going to have with Google and using Gemini as sort of their Siri are going to let whatever that is go super deep throughout the entire phone. And so I would imagine that's why I've always, I've always thought this should be Google's time.

AI assessment note: “It may very well be that that's where this lands. Having said that”

Q being actually held up by Tom Tillis, who's not happy with the fact that the executive is, is, uh, or the white house is, you know, influencing the fed in this way. How important is fed independent? How important is fed independence, um, from politicians who obviously will want that money to be spent no matter what. Um, if we're gonna have a system that doesn't get into another crash.

A So I'm generally of the view that Fed independence is hugely important, especially because typically if you do get into a crisis situation, you often have to make hugely politically unpopular decisions. Um, if you just were playing politics, you might not do some of these things, and that's a problem. The question that I don't think we will know until a year or two or maybe even three from now is Kevin Warsh, who I've known, by the way, for many years, um, who is very, very bright, clearly has gotten this job in part because the president believes that he's gonna follow what the, either the president wants or that currently their thinking is aligned. To lower rates. To lower rates. It'll be very interesting to see whether, A, Kevin Warsh stays in line with the president. I, by the way, think he could be an independent actor, ultimately, because interestingly, once you get this job, the job is yours. It's a little bit like becoming a Supreme Court justice. There are people that the president's appointed, uh, that have then, um, you know, decided, uh, decided against him. So I think it's possible. The other thing is, The Fed is a little bit like the Supreme Court also, in that just because you are, um, the, the chair in this case, Kevin isn't the one who, isn't the only vote. He has to convince all of these other people to do what he wants, and I would imagine in the first six mo…

AI assessment note: “I'm generally of the view that Fed independence is hugely important”

Q But then What is the difference between that and like a Claude code that will take over your computer and code up applications for you?

A Yeah. I mean, I think, I think the difference right now is somewhat minimal and it's going to narrow. I don't know if you've seen this new trend of companies like Pulsia, which are basically like a wrapper on open Claude and Claude code where, uh, you say, here is my business idea. And it says, okay, I'm gonna go and use like a Claude code to code up a product for you. But then also it uses an open clause style architecture to say, I'm gonna go set up a marketing campaign. I'm gonna buy meta ad dollars, those things. And so it's more of a, you could do that all on Claude code, but if you're a non-technical person, it's very, very hard to kind of like bridge the gap there. Um, that being said, I think Claude code is gonna continue to get better and better, and so we'll probably see some more compression. Palsy, I think the founder tweeted they were at, like, three million ARR in, like, a week and a half. So the idea of people being able to bring a business to life with just a prompt is, like, very compelling, and I think we'll see a bunch of, of companies doing this.

AI assessment note: “I think the difference right now is somewhat minimal and it's going to narrow.”

Q At the top of the US App Store and reached one million downloads faster than ChatGPT. Since then, downloads have decreased. I think that's sort of putting it likely. It's falling off the face of the earth. What's going on there?

A Yeah. The, the Sora data is really interesting. Um, There's, there's, like, a ton of lessons, I think, embedded in that one experiment. So the first thing I would probably say is the model's actually very good. I think, uh, it's close to something like a VO-III in terms of, like, realism on both the audio and the video. Their big unlock, which was super smart, was the Cameo feature. So the fact, that was why, like, every, every other Sora was like Jake Paul, because he granted them the right to, like, use the Cameo. And that's what made it go viral, because people were making memes of their friends. Um, but because the videos were exportable, what would happen is that the best Sora videos would get uploaded onto TikTok or Reels, and then they would compete against the best human-made videos, and so the overall feed experience was just kind of strictly better on one of those platforms than on Sora alone. Um, downloads are way down to that point. I think it hasn't become the social network that they maybe hoped it has. Where it is succeeding is as a creative tool, because the model is quite good. So they still have Three million DAOs, and it's actually, you know, slightly climbing over time. Um, so active users. Yes. Daily active users. Yes. So people are still really using it as a creative model, but they're not using it as like a social graph product. We haven't seen anyone cra…

AI assessment note: “I think it hasn't become the social network that they maybe hoped it has.”

Q and it's just an ego blow up. The second is that potentially, is it the Anthropics CEO, Dario Amadei, valiantly standing up against mass surveillance and the potential of mass surveillance through AI? Or third, is this what's really happening if the Department of War valiantly pushing back against a private company dictating it how to run wars? What do you think is closest to the truth in this scenario?

A I mean, there's probably like a little column A, little column B, little column C going on, uh, like fundamentally, but to me, this is about personalities and politics masquerading as a policy dispute, although it raises really important policy issues. And let me, let me tell you what I mean by that. If you look at the relationship between Anthropic and the Pentagon, Anthropic was the first frontier AI lab willing to do classified work. To support American national security. So starting right there, like Anthropic was ready to be behind the scenes with the Pentagon in a way that other frontier AI labs weren't ready to do yet. And Anthropic was also, there was no dispute between Anthropic and the Pentagon about any current projects that Anthropic was doing. It wasn't like the Pentagon asked Anthropic to do something and Anthropic said no or had hesitations. It also seems as though there were not any upcoming projects. That the Pentagon was going to ask Anthropic to do that Anthropic had, uh, had questions or concerns about. It seems like this kind of started when after the Maduro operation, when the United States, uh, uh, plucked, uh, the leader of Venezuela, uh, from, uh, from that nation and brought him back to the United States, that, uh, somebody from Anthropic basically called somebody from Palantir and said, like, hey, was our tech involved there? And that's because the wa…

AI assessment note: “to me, this is about personalities and politics masquerading as a policy dispute”

Q Right, and we'll get into that. I think one of the interesting things is that the questions about consciousness is, you know, if let's say the goal was survival, you know, we could do that all sort of mechanically, but For some reason, we don't just do it mechanically, right? We do it in a way that we have awareness of it as we go along.

A Yeah, and that's kind of, you know, that's one of the hard problems. I mean, this idea, you know, most of what the brain does, it does without our awareness, right? It's monitoring the body 24 seven. It's adjusting your heart rate, your blood pressure, your, you know, glucose levels. I mean, amazing amount of things to keep you at the proper homeostatic set point. And we're not aware of this. Yet some at the tip of that ice iceberg of mind is this area of stuff we are aware of. So why isn't all automated? Why wouldn't that have made more sense? Why did some of it have to come into our awareness? And there are various theories about that. Um, you know, one, and I find it persuasive is that, um, there are certain things that go on for a creature that need to be addressed In a reflective way. In other words, um, that if you, let's say you have needs that are incommensurate, um, and, and are in conflict, you're hungry and you're tired, which should you, which should you address first? That kind of stuff would, would come into consciousness. Um, I also think consciousness is really helpful in a social situation. When you're dealing with Um, a world that is fundamentally unpredictable. That is to say what other people are going to do at any given time, what other people are going to say at any given time. And you have to be able to imagine yourself into their heads. We call it theory…

AI assessment note: “we live in this intricate social world, being conscious is a huge boon”

Q opposed AI mode. But the story does say that there was a group of people within OpenAI who have Within the company, uh, stated their opposition seemingly loudly to the fact that it's going to roll out, uh, this adult mode, and by the way, adult mode is going to be coming out, seems like, in the coming weeks, coming months at most. Stephen, what should we make of this?

A It's, it's hard to weigh in on any one personnel incident. And also, uh, this would not be the first time that OpenAI seems to have done a pretextual firing where they got a person out of the organization who had safety concerns that the company, um, either didn't like or didn't like how they had expressed them. And, and notably those are different, right? Like you can have concerns about how the company is operating and that doesn't mean you have license to Say anything in any forum. Um, but Leopold Ashenbrenner, who wrote this huge essay, situational awareness in the past, um, maintains that open AI said things to him when he was fired that implied it was, it was basically because he had contacted the board about security concerns that opening eyes models were not actually secure. And so I think the way to interpret all of this, right? These, these aren't an apocalypse in the sense of Something super, super substantively scary happening right now, but I think they are early warning signs that people within the companies are raising flags of sorts, and they are not being permitted to speak freely. They are paying consequences for it. And so the question is, as we get to more and more dire issues at some point, hopefully we don't, but we might, you know, will, will we have people who are still sounding the alarm who are willing to have the, the courage of their convictions in t…

AI assessment note: “I think they are early warning signs that people within the companies are raising flags”

Q I think this is notable because, uh, it just shows that there's so little room for error in this world that if you get, uh, you know, if you, if you get anything wrong, the market will interpret that they are going, that you're gonna, um, have some serious problems when it comes to AI. What do you think?

A Yeah. I mean, it's, it's every quarter of the earnings that come out from these big tech companies, it's sort of a lottery Like you look at the numbers initially and you're like, wow, those are big beats on revenue, big beats on profit. Like we're seeing a 30, a 39% growth in Azure and a huge increase in bookings. And yet the shares plunge. Microsoft gets absolutely killed in the market. And there are always games played by Wall Street, right? People don't just want to see you beat the analyst estimates. They want to see you beat the beat, right? And so Microsoft has found itself in this very strange situation. Where a one percentage point miss in its Azure revenue momentum causes it wipes, you know, hundreds of billions off its share price. And then people are also, there are also several other things going on at Microsoft. You know, there are some very interesting things disclosed, uh, about open AI. Um, but also it's capex. Microsoft is on track to spend a hundred and forty billion this fiscal year, which their fiscal year is a bit weird. It runs from to June instead of to December. Uh, and that's pretty much with on track with what they spent in the last quarter of last year, which was, you know, thirty seven billion. But people are questioning this. This is 66% more than they spent in the same period the year before. These are vast jumps. And then the other thing that caug…

AI assessment note: “Where a one percentage point miss in its Azure revenue momentum causes it wipes”

Q of open AI. Um, interesting thing you said before the break, Brett, uh, that Cigna, I think rolled out in what a couple of months with you guys. Um, there is a whole industry of consultants that integrate systems and are, you know, basically connecting business processes and they take a year and they usually do it like. Uh, effective, uh, what's going to happen to consultants in this world?

A Consultants is a broad category. You know, you have strategic management consultants, you have systems integrators, you have, uh, you know, outsourcing firms, you know, that, um, essentially can, like, build software on your behalf, and I think all of them will have different impacts on this technology. The first principles view I have is, Software engineering agents are bringing down the cost of developing software. It's not going to go to zero, but it's, uh, you know, it's going from extremely expensive to relatively inexpensive, and that's a huge change. And so, where you had a dynamic where a company was outsourcing software development to save costs, the cost of the AI agent might be less than the cost of the outsourced, uh, I'll say resource even, so it's somewhat a dehumanizing term, but Often the term used in the industry, and that'll disrupt that part of the industry. But behind every technology project is a business transformation. Um, you know, if you're, uh, digitizing part of your business, The software is a means to the end, not the end to itself. And I think most consultants would say the change management that they participate in, the actual advice on how to do that and how to be competitive has always been the more valuable part of what they do. Um, so I think a lot of these consulting firms will probably need to transition away from, um, you know, uh, the bill…

AI assessment note: “consulting firms will probably need to transition away from, um, you know, uh, the billable hours”

Q methods like machine learning can't because with machine learning work with the predictive analytics and the machine and the, um, optimization side of AI, that's been something that we've seen in market for a long time. It's done a pretty good job. So when it comes to the advantage that quantum might have over some of those traditional processes, what do you see and what do you, your customers see?

A So let me start by giving you a concrete example or two of, uh, customer applications that, uh, we have in the market today. Uh, and then maybe I'll come back and share some thoughts on, uh, how AI and quantum relate to one another and potentially work together synergistically. So let's take BASF, one of the world's largest chemical companies. Um, they need to fill orders For customers, for a variety of different types of chemicals that frankly are created by, um, a manufacturing process in their facilities, starting with a variety of raw materials that come together to form those chemicals. Okay. The challenge is to, uh, optimize how the plant floor operates to fill those orders as quickly and efficiently as possible. Um, leveraging our quantum computers, they have been able to reduce the time required to do the production scheduling from what was 10 hours down to seconds. Okay. So a very significant reduction in the time to do the scheduling. Well, if the computer is running for 10 seconds versus 10 hours, that's consuming a lot less electricity. So there you have an example in the real world of the use of quantum computing to deliver business value and reduce electricity consumption. Another example is a large mobile carrier in Japan, NTT Docomo, has used our quantum computers to optimize cell tower resources. Basically, how the cell towers interact with the mobile phones to…

AI assessment note: “leveraging our quantum computers, they have been able to reduce the time”

Q already some new, new data to me, so good to hear that. Um, but I wanted to, again, like, just get into what it, what it takes to build these things, uh, these data centers. Um, you're, you're assembling them with incredible speed, uh, so I just want to hear a little bit about, like, On the ground. Uh, what does it take to put together, uh, these data centers?

A So the, um, historically, uh, You know, let's say two years ago, we were able to go out and buy capacity or lease capacity that was much further through the development cycle, right? They were basically, the shell already existed. It was a fit out construction process, which means going in and installing like the last, last pieces of the cooling infrastructure, cabinets, conveyance for all the cabling, all the hundreds of miles of cabling we have in these things. Um, but it's shifted over the past year is that now we're doing much more, uh, bespoke in-house design, right? To make sure that we're meeting the needs of what our customer's deployment is going to be. Right? So it's everything now from, okay, how is the cooling and electrical distribution designed? Um, how are we ensuring electrical redundancy and reliability? Uh, you know, how are we cooling the air-cooled side of these things? Because you have liquid cooling, there's still a component of it that has to be cooled with air.

AI assessment note: “everything now from, okay, how is the cooling and electrical distribution designed?”

Q Right. That is, that's expensive work, by the way. I mean, it, what, what this, what I think this is doing is it's compressing. I mean, you talked about weeks, you could take weeks of work and, and basically condense it into a prompt. Like what are the economic implications of that right away?

A Sure. And I'm going to give you a little bit of backstory. So I've been covering legal tech for about seven months now, and It came about because my editor-in-chief at Business Insider, long before she was a journalist, was a Yale Law School graduate and a law clerk, and she ran in social circles with a lot of lawyers, and what she kept hearing was this, um, fear and paranoia around the death of the billable hour. Um, lawyers get paid by the hour. There are definitely exceptions to that, um, where the work is more flat fee based, but if you're working with Enterprise clients you, you typically charge by the hour and software that promises to save them time. Um, that math doesn't matter for them, uh, when the risk is that it cuts into their, uh, revenue at the end of the year. What I found talking to lawyers is that there's still paranoia that this changes the business model for them. It will nudge them to look at alternative business models like switching to flat fee. Another one I hear a lot is that, um, they're still going to charge by the hour But the kinds of work they charge a billable hour for changes. You, a, you know, top partner at Quinn Emanuel might be billing 10,000 dollars an hour because there's a lot of rote, repetitive work that your class of associates used to do that you're, you're doing with a generative AI tool now. So they can, um, push aside Some of the wo…

AI assessment note: “fear and paranoia around the death of the billable hour.”

Q these are the people who are having the interactions with the end customer. And you're right. Like if you can get the sort of change password questions out of the way, you now take this, this, um, division, which ends up like, which is, has been dealing with problems and you make them sort of the owners of the relationship. With the, with the end customer. Does that sound right?

A Yeah. If you think about it, um, you know, you have a long relationship. We talk about lifetime value in marketing, right? If I'm not mistaken in terms of how much is this dollar of email spend getting you? How much is this dollar advertising getting you? After that initial conversion, the only people who really talk to the customer are the customer service agents. You know, if we look at our own company data, 54% of our customers contacted us for support in 2024, but they represented 95% of revenue. Right. So the most important customers are speaking to you. They're speaking to you in what we would call the moments that matter where they need some help. And I think, you know, AI is, because it gives you this incredible potential to raise the level of every agent up to the level of the best and longest serving customer service agent you have with co-pilot technology and assistance, right? It also means that you can kind of like have an immediate response and kind of deal with things. To a certain extent, customer service is this human powered factory, right? The metrics are ones of throughput and effort. It's like tickets today, per day, time to first response. You know, all of these things where you're measuring productivity, average handle time. In an AI world, you know, especially when you're automating tickets, tickets per day is infinite. You know, I'll just, there will be…

AI assessment note: “After that initial conversion, the only people who really talk to the customer are the customer service agents.”

Q hand window and they get a chance to have a conversation, uh, with that bot. Um, I wonder, you know, because you, you're the chief technology officer of Zendesk. So you'll have a insight into this. Do you think that customer service is going to happen, um, on, let's say, client websites in the future? Or do you think it might migrate into, like, the big, broader bots like ChatGPT?

A I think that if the bots represent, um, our agents, it's clearly, clearly there's going to be some kind of migration, right? Already for a given brand, you know, if you want to know about some companies, you begin with a search in Google, and turns out Google's results are usually pretty good. I think the same is happening with ChatGPT. But when we get into like real service flows, where will I be saying to ChatGPT, I bought these shoes last week. You know, go, go figure out that order, return them, and generate me a, uh, generate me a packing label. Yeah, I think that's probably going to happen as brand short circuit. Now, is that the customer service flow so much as the action flow where we're actually moving towards systems of action that do that? I think it's all becoming integrated, and the nature is changing. On the back of it, though, everyone is still probably going to need to have those moments where Uh, a place to go to actually get contact and talk to a brand. We, um, you know, recently we kind of looked at, we took a sample of. Fifteen million or more, um, customer service conversations across send us customers. And we, we actually use chat GPT to classify the contact reasons or the intents and just kind of group them into cohorts. And you get, uh, something along the lines of 47% are basically, there was some kind of failure in the business. Right. There was someth…

AI assessment note: “clearly there's going to be some kind of migration, right?”