On behalf of the Lightspeed and Axios teams, thank you. We love having you here. My name is Guru Chahal, I'm a partner with Lightspeed Venture Partners. We're a global venture capital firm. We manage about $30 billion in assets under management today. We operate across all geos worldwide. Wherever you find technical innovation, you'll hopefully find a Lightspeed partner looking to work with some amazing founders. We invest across sectors, across enterprise, across consumer, across healthcare, across fintech. And 25 years ago when we started the firm, at the time we were very early-stage enterprise-focused, but today that purview has expanded across all of those sectors. And you know, we are one of those very fortunate firms that have seen the evolution of technology across some amazing multiple cycles: across the internet where we got our start, across the mobile cycle, across the cloud wave, and now we've been investing behind AI for over the last decade, much accelerated over the last two years, of course. To give you a sense for the numbers, we've probably deployed about $2.2 billion in AI-native companies across all layers of the stack: those are foundation models, the silicon they run on, the applications that are built on top across various sectors. We support about 100-plus companies in AI-native ecosystems today. But as with all cycles, it's a combination of some amazing innovation and value that's created, but also some hype. And so I'd like to invite our collaborators at Axios and Dan to help us cut through that.
Great. Hello everybody. Thank you so much for coming out today. My name is Dan Primack, business editor at Axios. Unlike Lightspeed, I have invested zero dollars in AI companies, but I have written about all of them. So we're going to do two things quickly. First, just so you all know, we will have another session here tomorrow at the same time. We hope to see you all back here. And I'd like to invite our first guest, which is Lynn Martin, president of the New York Stock Exchange.
Thank you. Thanks for having me, Dan.
Thank you for being here in this cozy little spot in your Swiss chalet.
In my Swiss chalet. We were going to do this in Swiss German, but then we both realized we don't speak it, so we will not do that.
Right. So, Lynn, let me start with this. We're kind of in year three of the AI hype cycle, but we're also in year three of 'IPOs are coming back' this year. You run the New York Stock Exchange. Why is 2025 going to be different than, say, 2014?
2014, I'm sorry, 20... wow, I am clearly... that's okay, that's okay. Yeah, why will it be different than 2014? I don't have my stats on 2014. I don't know if we are in year three of the 'IPOs are coming back' hype cycle, because 2022 was a challenging year and it felt like it was going to be a challenging year, mainly because we saw that volatility start to ramp back up in 2022. And unfortunately, you know, almost two years ago there was the Ukraine—or almost three years ago, gosh—there was the Ukraine invasion by Russia.
Reasons to be optimistic for 2025? This is the year IPOs come back?
Look at 2024. You did see the green shoots of the IPO markets come back. And if we look at the IPOs that happened on the New York Stock Exchange in secondary trading, in aggregate they are 58% above their IPO price. Now that's a darn good stat. It means that there is investor demand, and when the right companies come to market, they see that value accretion. We're so proud that nine of the top 10 IPOs from a performance perspective last year happened on our platform. But you know, a tremendous amount of capital got raised last year. When you consider that we had a US election—a pretty controversial US election—that ebbed and flowed throughout the course of the year, is probably the right way to phrase it, causing volatility in markets. What that meant for the IPO markets is it felt really good up until June, and then the market opened, shut, open, shut, open, shut throughout that election period. We've got some deals on the road. We had an IPO last week that also performed really well, a company called Flow in the LNG and energy space, priced at $24 and closed Friday at $30. So again, green shoots are there.
When I say 'we'—we in the press—we always look for companies that are going public and then say, 'Oh, that's tailwinds, like they did it and thus others are going to follow.' You have a big one on the road right now, Venture Global, which is an LNG company, not an AI company. Could be a billion-dollar IPO. Do you view that as one that, if successful, will have tailwinds for other companies, or is that really more of a one-off because it's not unique but relatively unusual, and it's maybe a little bit like what Arm was a year ago or a year and a half ago, it kind of stands on its own?
It does stand on its own because of the size of that IPO. I mean, they're talking around a $100 billion valuation at IPO, which is...
It's like a small AI company.
Exactly. But importantly, you think about it, you think about what is coming out of the US, some of the initial signs of what President Trump is going to do, the LNG sector is really well positioned.
So you talk about green shoots, and you're right, a lot of the companies that did decide to go public last year priced well, they've traded better in the aftermarket. But David Solomon, good standing member, his company Goldman Sachs, he said, quote, the other day at a Cisco event, quote, 'It is not fun being a public company. Who would want to be a public company?' end quote. Says the CEO of a public company. Your response to David: why would anyone want to be a public company? It's not fun.
Well, to address the 'not fun' part, I think the next administration—this administration that got sworn in yesterday—is going to look at some of the reasons it's not fun. You know, the additional tax, if you will, on public companies, the additional disclosures. Now, no one is questioning the need for accurate financial disclosures. That's good, that's good table stakes. But there's been a variety of other regulations that have been layered on top of public company CEOs over time. So it does appear that this administration is going to look at that. Nothing really beats the liquidity and depth of the US public markets. That's the reason to go public.
Democratization of investor access.
Yeah, right. But what it means is I think companies are probably going to stay private a bit longer, and when they come to market, they're going to be better companies, better, higher-quality companies that actually tap the public markets.
Do you feel you are—when I say 'you,' that the stock exchange and exchanges, maybe even globally, are competing now with private capital in a way they never did? It used to be private capital was almost hand-in-hand. Private capital, you know, helped a company get to you. Now it seems like they're often stopping a company from getting to you.
I don't view them as competition. I still view the private markets as an important partner. A company, when they're seeding an idea, should not be a public company. They're still in startup phase. They're still investing in the next great initiative, the way to make the world better. And when you're in full-on investment mode in your business and trying to get your product to market and get critical mass of adoption for your product, it's probably not the greatest time to tap the public markets. You're very focused on, you know, building still. And I think the private markets play an incredibly important role.
Do you feel—is there any concern? You know, the thinking used to be that one of the reasons that people liked to buy into IPOs was because you were still buying into a big growth story as an investor. If some of these companies get so long in the tooth that at least the rapid growth—all the benefit of that has gone to the private investors, or not all but a lot of it—a lot more of that growth is now staying on the private side than the public side.
Because they are—look at a company like Reddit last year, their IPO, their valuation at IPO day was $7.5 billion, it's now $30. They're a great example of a well-known household name. Viking Cruises, that was another great IPO last year, the more industrial sector. Rubrik, the more tech sector, you know, Rubrik with their cyber defense platform. These are companies that are new, hot startups, still have that great growth trajectory, still are on that hockey-stick journey that the public can now participate in, that hockey-stick journey.
One other thing Solomon said, and then I won't keep talking about him. He was talking about what it took to bring a company public. He said the work of drafting an S-1 used to take a six-person team two weeks to complete for the banks, but now AI can do 95% of that in minutes. Cost—quote-unquote cost—is one reason why companies often say we don't want to go public. Are the costs overall going to come down, including from the exchange, because AI and technology can do a lot of the work that people used to have to do?
I don't know how the cost—the cost will definitely be impacted, but I don't know how really where you see the costs are that regulatory tax that I was talking about. And I still—it's still early days. We're on day one basically of this next administration.
If Scott Bessent, the presumptive Treasury Secretary, came to you and said, 'Lynn, you get to be queen for a day. What's the one thing you want to do? We can get this either through Congress or through regulation. What's the one thing you would want him to do?'
I would want there to be a glide path for companies who are newer startup companies that have a path to the public markets that don't necessarily have the full-on tax of being a public company. Not financials—I think financial disclosure is incredibly important. What's something you think that a company today has—if you look at some of the ESG, which I'm sure that's a regulation that'll be looked at, some of those disclosures, I think that's an area that will probably be looked at.
Let's talk about AI. You obviously personally have a background in neural networks. That's kind of—you're not a financier by trade or training originally. What is a tangible example of how the New York Stock Exchange is using AI internally?
We process three-quarters of a trillion incoming order messages a day—buy, sell, and match trades come into our platform. We can't do that with humans. Can't find enough humans to process that without using good technology. So we've been using AI from our regulatory team to spot nefarious behavior in our markets and then have that as an output to our humans to allow our humans to do higher-value-added work as opposed to crunching, you know, 'here's a message, here's another message.'
Do you think you're catching more than you would have caught three, four years ago?
That's interesting. You hear two arguments when it comes to AI and labor. You hear the one, it's going to replace everybody. The other is it's going to actually create more work. Do you feel, for example, when it comes to—you don't have the people crunching the emails as you said, but do you have more people almost on the enforcement side because you are seeing more?
We have people doing higher-value-added work than they did in the past.
You talk about the importance of having, quote, 'high-quality data.' I think that's the term you use a lot. What practically does that mean? What is high-quality data compared to, I guess, low-quality data?
High-quality data is data that can be trusted. I mean, to me it's just about trust. Is it data that, you know, is the golden source? If you have data that is not the golden source that seeps into the models, you're going to get probably unfortunate outcomes. But you also need the human to make sense of the outcome of the AI.
How do you—or how does New York—how do you determine one from the other? Because I would think that data, if it—how can I put this—if it is bad data, it is disguised to look like good data.
Well, that's where we apply AI selectively to data that we know is trusted in our platforms. We don't just apply AI everywhere because we're very deliberate on the data that we apply it to.
What can you do when you talk about being able to crunch those—I assume you've been able to do that for a little while. Is there something that you've been able to add, say, in the past year? When we think of how quickly AI is advancing, is there something you've been able to add in the past year that you couldn't, say, two years ago, just because technology's gotten advanced?
We're using AI in a couple of other areas at our parent company, which is another area I'm quite passionate about, ICE. So we've been using AI there to add liquidity to our energy markets for quite some time. We have LLMs that are embedded within our proprietary chat platform, which understand if I'm talking to you about the Swiss chalet we're sitting in, or if I'm talking to you about a crude oil trade. And if it detects the crude oil trade, it will pop through some value-added analytics that we produce internally and then allow you to submit that to the trading platforms. We've seen the amount of activity generated—still early days—but the amount of activity generated in our energy markets really do this really nice compound growth.
Do you feel 2025 is going to be the year that we see a true AI—I don't mean AI-native, AI-focused company come to market. I mean, I know there's some that are talking about it. Are we going to see every company is now using AI, apply AI, including legacy companies, but a company that we would kind of all say, 'Oh, that is an AI company,' not just a company that uses it?
I don't know. I think the jury is still out as to who's coming to market. I think the people that are tapping the public markets are the more traditional companies, the ones that you understand. I do think the energy sector, given the demands of AI on power, is very much related to AI, but to your point, they're not a true AI company.
That makes sense. You and I were on stage in New York in May, and it was just days before the US trading systems moved to T+1 settlements, and there were questions of, 'Is this going to work?' And it did, with relatively few hiccups. Does the rest of the world—like Switzerland—does the rest of the world need to move to T+1 before the US can start going to T+0 or begin working toward T+0?
Well, we're pretty happy we got to T+1 as quickly as we did. I don't know what the—I think the US is on a different path from probably the rest of the world. One of the areas we're focused on is the elongation of our trading day. We announced in September that we were moving to 22 hours a day on our electronic trading platform, mainly because our clearing house, DTC, had started to allow trades in the elongated session. Now, there's liquidity that had developed overnight outside of the exchange ecosystem, but importantly, it wasn't guaranteed by a clearing house and it wasn't guaranteed by hardened infrastructure. So we're really focusing on getting that done this year.
Why 22 instead of 24? We match the clearing house. Why are the clearing houses 22? Seems like such an odd, arbitrary number. Like nine makes sense, eight makes sense, but 22?
There is a day—there's a moment—go for two hours—there's a moment in time where their systems had to recycle. So we're working with them too to see if it's possible to elongate further. But we're always going to follow the clearing house because to our...
Do you think the clearing houses will—the systems be able at some point soon be able to reboot a little quicker? That seems like a long time.
Possibly. I mean, we've run futures exchanges that operate 23.5 hours around the globe, and we can recycle our clearing systems pretty quickly.
Give me one macroeconomic signal that you look at and pay attention to that you think the average bear doesn't pay attention to.
Well, everyone's paying attention to inflation, right? Everyone's paying attention to interest rates. I pay attention to volatility. The VIX is a key indicator that I pay attention to because to me that's a barometer of if the market is stable from an IPO perspective.
That makes sense. Last question for you. You've come to Davos before. I'm a newbie. This is my first trip here. I've been here for like 48 hours. Something... having fun?
I am having fun. It's been good so far. I haven't slept nearly enough.
Give—for other people who are like me, give me—do you have a Davos hack?
Drink a lot of water, because here in the mountains you don't realize that as much, right? And you're talking so much. The great thing about Davos is everyone is here and you have the ability to interact with so many world leaders and luminaries and, you know, true icons of business—I don't just mean political leaders—but by the end of the day your voice gets really raspy from talking so much. So make sure you drink a lot of water.
Lynn, thank you very much. We're going to go off and bring up Ina Fried, my colleague.
Hi everyone, I'm Ina Fried, chief technology correspondent at Axios and author of our daily Axios AI Plus newsletter. If you don't get it, go to axios.com/newsletters, I'll be in your inbox, it'll be great. With me is the incomparable Julie Sweet from Accenture. Julie, it's really good to have you here because I think, you know, everyone—I don't think there's any debate, I'm not even going to ask people to raise hands—that AI is big, it's going to change everything. That's what they're spending a lot of time thinking about. It's the how that gets everyone stuck. And last year you all did this great thing, we wrote about it, it was like a boot camp kind of, 'here's what you need to know about generative AI.' We were talking, you were saying companies kind of understand the technology, the big question now is how do they actually incorporate it into what they do. And I really personally—and I don't generally feel for businesses, they do fine, they make lots of money—but I do kind of feel for companies because the technology is changing so fast and businesses are made up of humans and we only change at a different pace and it's not the same. How do businesses possibly keep up? And what are the recommendations, I'm sure you're hearing, from CEOs that come to you and say, 'What on earth do I do?'
Well, Ina, it's a great question. First of all, it's great to be here, thank you. And what's interesting is if you can contrast last year, before we get to kind of the how, is last year we were doing workshops, really trying to make sure everyone understood the power of AI. And last year, agentic architecture, digital agents, was not the buzz it is now, right? So two things happened over the last year. On the one hand, people do really get it now. Like, you know, we don't need to—we're not doing those workshops this year, right? Instead we're showcasing actual examples of AI creating value at scale because that's what everyone's now saying, 'Okay, we get it, but how do you make it actually work?' On the other hand, the technology has gotten more and more complicated in a year, and it's those two things that make it so hard. Because on the one hand the technology is getting better and better, but then, you know, you have to keep up with it and you have to say, you know, 'Am I creating the right platform for like what's coming next?' On the other hand, actually, if you look under the covers as to why the companies who are getting value out of it are getting value, it's not about the technology. It's because those companies have been willing to really change the way they work. And I'll give you some examples of that. They're changing their talent. And in addition to the really cool AI technology, every company that's really scaling and getting value now has been investing for years in the prerequisites of technology like the cloud and the data. And so, you know, all of you came here and had to book a hotel. Like in the hotel industry, there's now some real examples that are going to change your experience. It probably didn't change your experience for—it's going to help you get a Javas—it is not going to help you, right? But so like Radisson Hotels now has always done a great job and done a lot of digital and personalization, but now when you go to book a ski vacation, they are able to show you pictures of the resort in real time, knowing that you actually ski versus someone else who's going on a ski vacation is booking it and they really like the spa. And so what comes up for them are like the pictures of the spa, right? Completely different experience, right? When you're now buying a car, working with a global automobile maker, you know, we've had like—you're able to like build cars for a while, but now when you go to build your car, they know, well, you've bought these kinds of seats and you like this kind of thing, and so you're starting without all of these, you know, names and confusion, massively simplifying. And these companies are getting real results, like numbers that actually are material and matter, right? Both of these companies though had been investing for a long time in digital, and when they looked at this, they completely changed like how their marketing department worked. They're getting 50% more productivity in marketing, which means they completely reorganized it to use the tools differently. And so when you think about the how, right, of course you have to understand the technology and that's hard, and you have to partner and that's difficult, but the biggest how is actually understanding that the value comes in understanding your business processes, how your talent has to change, and being willing to change. And that can be the hardest piece.
Without asking you to totally name and shame, what are some things that aren't working? Give us some examples of some things you've seen, either from clients before they came to you or things that you tried that didn't work out. Because even the best—not everything works out, and you guys are fabulous. What isn't working with AI?
The number one problem is when companies are not willing to actually change the way they work. So what do I mean by that? We sometimes say that it gives you more coffee time because GenAI in particular tends to not take out an entire job, it will change tasks. And most companies are not able to look at their talent and say, 'Well, if these tasks are eliminated, so these 10 people have now saved, how do I rework this department so that maybe I now need eight and I free up two to do something else?' Like sales is a great example. There's so much you can do in field sales in consumer goods because almost 50% of that job is actually administrative. But when you free up that time, you have to then think about like, okay, well I now need—what are those people going to do with that time? Am I going to do less? Am I going to redeploy more? And if they're not doing administrative work, which I've been hiring for, do they have the skills to now build better relationships? That requires a whole different way of thinking about skills, about HR, about recruiting, about the business strategy. None of that has to do with the underlying technology that enabled it.
One of your rivals actually put it really well at DLD, saying the biggest beneficiaries so far have been the dogs of software developers because people are getting home 20 minutes earlier and getting to walk the dog. And it really is—I mean, it's a funny point, but it highlights that, you know, and I think it's also the flip side of that is why companies are hesitant to keep writing giant checks. One of the things we've been reporting is a lot of the software makers, Microsoft and Google, had, you know, initially positioned their business as 'pay us an extra $30 per month, get this business copilot,' and they're now doing pricing things where they're saying, 'Okay, we're going to raise Office or raise Google Workspace by a couple bucks and give it to everyone,' because what they're hearing is it's not that companies are saying, 'Look, 30 bucks a month per worker, if we get a little bit more out of them, you can make that pay off,' but at the same time, it's a big check, and it's a big check when you can't say, 'I know I'm going to get this.' I assume you're also hearing that from CEOs on the cost-conscious side.
Yeah. Well, I mean, there isn't really an industry today that's not cost-conscious. It may be for different reasons, right? So the energy industry is doing better and has to, you know, invest trillions of dollars for the energy transition, right? Versus consumer goods, where gross margins are pressured and they can't get pricing. So every industry has costs and IT budgets are not changing. It's even though the value is there, people are not saying, 'Well, therefore I'm going to double what I'm spending,' right? They're re-evaluating. And that's where I think the technology companies are recognizing that there are constraints to adoption around cost right now until companies are where they need to be. Most companies today are not AI-ready. I mean, we estimate that about 10% of, say, the G2000 actually have the data and the infrastructure that they could, if they wanted to, take AI across the company. So think about all the investments that have to be done just to get ready. I also think though that there's a thoughtfulness as people are starting to say, 'We've got a lot of examples,' but when I think about scaling, they're starting to understand the new capabilities they need to have. Like for example, reinventing HR. So first of all, if GenAI is not going to replace whole roles but tasks, then you have to look at your people from a skills lens. Almost no companies in the world have skills as the primary currency for HR. We do at Accenture. I have almost 800,000 people. We have a skills database. I can tell you how many people have GenAI, how many people know SAP, how many people with this industry and what the skills are, so that we can upskill and change. But that's a completely different way of doing HR. Then there's of course digital agents. Those are digital employees. They actually have to be onboarded. They have to be trained with your data. They'll have job titles. They have to have guardrails like compliance because they're going to be doing tasks that humans did and they have to follow the same policies. That's a totally different way of thinking about technology because they have to follow things that your human employees are following. And then on top of that, change. How do you get comfortable? We just published a blog today about using digital agents in marketing and the journey that is from the lens of our chief marketing officer, from the lens of talent and change, and then from the lens of technology. What we found was that one of the best ways to get people excited about having someone help them in their work was the feedback loop, because we created the digital agent so that the person who's now being augmented by them can provide continuous feedback. And for our employees, when they realized that they could give feedback, the agent would learn and their job was getting easier and easier, so they could focus on things that they want to focus on. Like for marketing, we're now getting campaigns 30 to 40% faster to market, taking out a lot of the stuff that our marketing people didn't want to do, so they can spend more time. But what was key to the change was not me telling them that it was—'Wait a minute, this agent is someone—it's almost like I can tutor them and they're going to get better and better.' And that was really exciting. That took a very different way, that was not taking a playbook on like, 'Let me train you how to do an agent.'
