Fireside with Brad Gerstner, CEO, Altimeter | Global Alts Miami 2025

all right welcome back altimer Brad gersner he's an amazing investor and leader in Silicon Valley today you got 22 billion this in my hand here that now famous letter to Mark Zuckerberg I had vors call me up telling me what a fool I was he believed he said flatter is faster leaner is better 40% of your Revenue today is INF it's about to go up by a billion times billion right Nvidia is not just in the business of generative AI it is our biggest holding today going to have to replace the trillion dollars of CP driven data centers there is no intrinsic value he has a vision technology enables value that consumers are willing to pay for and he's not going to be knocked off the vision are you leveraging AI to increase productivity reduce cost increase value reduce waste right that's the [Music] practice I mean it ain't Diplo but we're going to have fun so you really don't need much of an introduction but for those of you who don't know Brad is one of the most successful VCS of this era uh for sure returned how much have you returned in the last three years four years more than we've ever raced yeah it's a big number um and Brad is also known as the chart guy I remember actually the first time I saw you it was probably the all-in Pod and the engineer in me just loved the charts I was like I love how this guy goes through this so we decided that the charts were a big hit last year so we're going to let Brad kind of tee up where the market is right now and run through some charts for you I actually run an investment firm early stage Venture late stage Venture I've been in Silicon Valley for 25 years and so I actually like when we have a conversation this week about deep seek I don't like to hyperventilate I like my team to actually do the work build the model like so we can have a real conversation now is a time that you can generate a lot of alpha the last 2 years we're up 169% % versus the S&P up 53 the point of that slide is in the moment of big phase shifts like we're in MH you know internet 2000 to 2004 cloud and mobile 2010 to 2014 and now in AI it's the fog of War mhm people don't know which way is up I mean look at the trillions wiped out of the public market yesterday because a butterfly flapped its wings in China literally really and so that is a huge Advantage for those of us who do this you know for a living you know one of you know something that you and I have talked about over the last couple years we got well known for was this letter in the fall of 22 that we wrote to Mark Zuckerberg called time to get fit and it was this open letter about the opportunity for incredible Market margin expansion and what most people didn't understand at the time it became our biggest holding almost a billion dollar position $90 a share today it's $670 a share and the point was there was amazing amazing unta potential at at meta right number one they had 87,000 employees he let 25,000 go got down to 62,000 employees and this company's still 2 and a2x Its Top Line how leveraging AI right and so at the time this was pre-chat GPT we said you got to cut your spend on reality labs and double down on AI so as investors we walk the walk right and Mark sat down and talked with us cuz he knows we're doing the work in that period of time but if you really just think about like why we invest in technology if you look at the last 10 years last 15 years last 20 years Tech has been the biggest free lunch in investing Tech earnings have compounded at 14% versus non-tech earnings compounding at 7% so it's not just that you know these are high multiple stocks no it's these companies are twice as profitable and growing twice as fast as non-technology companies as we say you know over the last 10 or last 15 years Tech has gone from 5% of global GDP to 15% MH and if I asked anybody in this room do you think Tech is going to be a bigger percentage of global GDP in five or 10 years you would all say yes and if I said do you think it's going to continue to out earn non-tech you would all say yes our job is to find the best of tech and avoid the worst of tech because the one downside of tech is disruption so companies that you think are great today Nokia in 2009 can be a eviscerated right AOL eviscerated Yahoo alav Vista eviscerated so I that's the puzzle that's the game that we get to play so as we look at this year you know what is the market pricing and what are they not pricing I'm going to give you a little bit of about about the macro backdrop you know I think the big thing to keep your eye on this year is the tenure okay we're at 4 and a half% rates are too high for the average economy the housing Market's in a recession the car the auto market is in a recession we have a role in recession through the economy because rates are really high why are rates so high not because the fed's not doing its job but because the market is stimulatory right we're going to extend the tax cuts under Trump and the market doesn't think we're going to do anything about the debt that's why the bond market went from 35 to 48 on Trump getting elected what it was saying is you're going to cut taxes and you're not going to do about the debt and we're calling your bluff and we know all the greats we're out there shorting the bond market so the question is at 48 or 5 the equity markets will not work um inflation expectations have gone up because of this everybody's saying the deficit is going to go up we're in a we're in a debt spiral the bond market Vigilantes are coming out I happen to be helping out on Doge with with with our good buddy Elon I actually think I take the over on this we are going to do something this is the first Administration in my lifetime that has the ability to do something about this because you have both houses of Congress you have a president determined you have Scott besset Kevin hasset and you have this phenomenon called Elon Musk um and so this is a little bit I'm not going to spend time on this but here's what I think is possible I think you can balance the budget in 2029 in This president's term at $6 trillion what that does is return us to the 2019 Baseline if you take 2019 you grow at 3% to 2029 you would spend $6 trillion okay in 2029 we'll have 6 trillion in federal revenue it's actually not that hard don't buy what they say right you can you can make these Cuts if you do this now all of a sudden the bond market rolls over the tenure comes back to 35 and it's the Golden Age We're Off to the Races but we need the president and Doge and Congress to actually make the cuts to get us back on this path um mentioned rates are still restrictive so I gave you the setup basically the macro backdrop well expensive or is it not expensive and what you know what are the earnings expectations this is I think really interesting last year Tech earnings grew 44% this is mag five grew 44% the consensus expectations this year 20% nonte only grew earnings last year 2% wow look at the huge acceleration they expect this year 11% okay so everybody's like oh well Tech's expensive I'm going to go hang out in this nonch Tech non Tech has to have a huge earnings acceleration where do they think it's going to come from let's break it down a little bit what categories healthc care last year grew 4% this year consensus thinks Healthcare earnings are going to grow 20% last year Industrials negative four this year they think 16 that's a freaking High bar yeah I'm not sure where those earnings are coming from and if you look at the the multiples against the 5year average nonch is actually trading higher relative to its 5year average average than Tech is so not only is their earnings bar higher but it's actually trading at a higher relative multiple to the 5year average so a couple things and then I'm going to yield back my time Ron because I'm filibustering because I know you're going to ask me hard questions you had a lot of charts the biggest thing and this will get to deep seek and everything else our most non-consensus opinion is the world is underestimating the amount of compute that's going to be demanded for the next three to five years right Nvidia is the poster child for this but it's a lot of things it's everything required to be consumed from memory to power in order uh for the world to make a transition from an x86 intel-based architecture to an accelerated computer architecture largely driven by Nvidia um and you see this was even before deep seek this was the setup people were like scaling laws we're hitting a limit right that small models and open models and all this stuff so this I put this together in December right right so yesterday was a continuation of um of this question mark But if you think about this new Super Cycle open AI raised money at $150 billion it's the fastest growing company in the history of capitalism open AI is doing far more Revenue than Google was at the time of its IPO has far more users than Google had at the time of its IPO is just a private IPO right we're investors in open AI if you look at capex everybody talks about the $250 billion all the big companies are spending it's one of the greatest assets in the United States of America right compare this to the R&D that's going on in China R&D in China is 100% the state it's the government here it's all these companies with literally printing presses in their back rooms plowing all this money $250 billion a year to keep the United States on the front the front edge of tech and notice people are saying well we got deep seek you're not going to need to spend this money 2022 before we even talked about models before there there was no model training we still spend $150 billion on gpus gpus run all of Technology now every single Google search you do requires accelerated compute every single time you do an inference transaction on Instagram requires accelerated compute it's not just about training a model it's about running all of these businesses that's how their revenues have accelerated because accelerated computers a lot smarter at targeting consumers and doing all the things that these companies do I'm going to I'm gon to I'm going to leave it there there are a couple more that we could get into um but let's let's let's go have a little back and forth all right so can we start with deep seek because this was like an atom bomb basically in in the AI World yesterday which I mean I'm far from an expert in AI but it felt like it certainly caught everyone by surprise in a major way could you just start out by explaining what it is because I suspect a lot of people in the audience don't don't fully I mean deep seek is an open source model the has been around for a decade one of nvidia's closest Partners incubated not surprisingly by a Quant Fund in China remember Quant funds have been doing AI for a decade right like you get a you get a millimeter advantage in a Quant fund it's a license to print money they have huge incentives to be really good at this and so they've been working on this before the export bands they were buying tons of Nvidia gpus even after the export ban they were buying a bunch of deprecated gpus and what was disruptive yesterday was not the model that they produced the model they produced is basically 01 and 01 preview quality that's where open AI was a year ago to put it in perspective internally open AI is already on 04 four generations ahead of the model that was released yesterday so it wasn't about the Benchmark it was about the cost they said it cost us $6 million to train this model and all the Talking Heads went on CNBC and said oh my God open AI spending 500 billion I just saw Stargate announc ment and these guys did it for only 6 million but nobody asked the question open AI hasn't publicly talked about what they spent to train 01 right I was just going to ask you so isn't that the logical question you would want to know the answer to what did open AI spend to train 01 I'm an investor in the company noan Brown who is the inventor of 01 is an adviser to the firm would you be surprised if open AI spent less than 20 million to train 01 that's my speculation I think they spent close to 12 or 13 million he had a tiny team and Tiny compute okay and if you think about how compute depreciates so gpus get two to three x better every year which means the cost goes down by 50% a year so if you spent 12 million and these guys do it for 6 million it's in the center of the bullseye of the expected amount that they would spend I I think that's highly informed speculation by me in terms of what open AI spent and so the point being that when you're in the fog of War when you're early in a phase shift that's when Partners like altimer can offer great Advantage whether we're partnering with Sovereign wealth funds endowments like we do or or families like we do because we live and breathe this stuff every day we're not only investors in open AI but we've looked at every model company that's come through our offices and we're investors in Nvidia and Microsoft Etc so it's about putting together those data points and on a day that the market is panicking we were buying we were pushing risk on the table when everybody else is capitulating because they don't really have conviction they're they're passengers they're tourists they have derivative conviction right and derivative conviction when the market is moving against you will get you run over and that usually happens I mean I remember I got into this business on the public side 04 and 05 and I would own Google did go down 10% for the day and my partner would come to my office what are we going to do about it I'd say buy more and go to sleep yeah you know because I know next year Google's going to do more searches than they did the year before and I know the revenue per search is going to be higher than it was the year before boil it down to its Essence right and so I think when you look at DC it was not disruptive in terms of training cost it's not disruptive in terms of inference time reasoning costs and again we have very good understanding of exactly how many gpus they have exactly the cost of running the GPU but it plays into narratives that in the fog of War you see how the mob rule can move that's MrMarket Buffett tells us all the time MrMarket could put on sale and panic for any number of triggering reasons yeah right and so I think today I think yesterday if your conclusion was that all of a sudden we can have all this compute for free right you couldn't be you're 180 off the Target right we're going to De we will have a shortage of compute in the United States of America for the next three years almost guaranteed so you're expecting then the market the market is just behind where you are they just haven't really fully appreciated yet the numbers you just shared with us what open AI spent on it where does it go from you see the market I mean like our portfolio is between the two days and this is welcome to my world you know because I have big LPS Sovereign funds they're like oh my God I can't believe we own so much of this stuff this is terrible and then I'm buying a bunch of it and it's up just as much today and they're like oh my God this is great you were totally right you know like it's hard during these moments but that's why you have to have Partners who've been through three of these super cycles and live through the volatility you know of these moments here's the thing I'm on Team America I want us to win the race in AI we have the best we we have the most R&D in the world we have the best chips in the world we're doing all these things but we shouldn't be naive MH China will have great AI we will do some things not to make it particularly easy on China we can do anti-d distillation at open AI we can have you know chip bands so they can't have our most bleeding edged chips but trust me none of that stuff is going to prevent China from having really good AI what we need to focus on is running the race as fast as we can not how fast my competitor is going to run it we need to get the regulatory out of the way so that American companies are not hamstrung to win the race and it's fantastic that we have Elon competing against Sam Alman I love their jawboning on on X like let the let the let the competition ensue that's why we're so good okay and by the way nobody's talking about Google yesterday but Gemini with deep reasoning is far better right than deep seek and so we're early in the battle final comment on this the model layer we passed on every almost every one of these models anthropic Mistral go through the list it reminded me of 1997 98 and search alav Vista lyos Ash Che AOL like you know the fact of the matter is there's zero intrinsic value to the model just like there was Zero intrinsic value to the web crawlers and page indexers at Al to Vista and Leos what's valuable is when a consumer says I'm going to use you and I'm either going to look at your advertising or pay you a subscription fee that's why Google won in '98 we all thought we were Geniuses we all knew search was going to be the dominant winner on the internet but all of those companies went to zero and you could have waited until 2004 and bought the Google IPO and you would have got 98% of all the profits ever generated in search okay why are we investors in open AI because when I look at it the indicator I was looking for is who breaks out who wins it tends to be win or take most in 2004 when I bought the Google IPO they had 250 million weekly average users when we invested in open aai they had 350 weekly average users there's a critical mass and if you look at the numbers since then they've continued to Skyrocket they captured the Zeitgeist they were the first mover they produced a really good po you know product and so like it's doesn't mean they can't be disrupted but the market leaders who get that sort of critical mass rarely are and there's not a company on the planet that has a billion C weekly average users it's not worth a trillion dollars you had a great run in 21 and a very tough 22 I it's hard to argue with frankly most of what you say because you're so researched on this and you know the numers cold but how nervous are you especially after yesterday that maybe we are at a peak and this whole thing is going to burst let's compare relative valuations okay in the fall of 21 because the government had gotten full full Zer on us literally paid us to take risk negative interest rates and then 0% interest rates had the greatest distortion on the market in my career 20 you know 25 years of doing this in the public markets software multiples went to 18 times forward Revenue okay the 10-year average is 6. 2 okay and we're at 18 times right huge standard deviation above where what you would have expected and there was only one reason for that the government got exactly what they wanted when you have a 0% interest rate the hurdle rate for the alternative for your money is zero so you push you get into the equity markets that's what they wanted to do to prevent the economy from crashing we made a huge mistake and we kept zero rates for far too long as we all know now which ignited you know basically hyperinflation for a moment today that same software company is trading below the 10year average at 5. 7 times forward earnings okay so I hear a lot of bubble talk and trust me 25 years in Silicon Valley having lived through a couple bubbles like I'm always on the lookout for the bubble when you're talking about the bubble there rarely is right and that it you know I did CNBC from milk and I think down here in October of 21 and this is what I said CO's over rates are going back to where they were before Co started and multiples will go back to where they were before Co started that's the reset in 22 that's what you had to get right nothing else 22 was not about stock picking 22 22 was recognizing a macro shift from zero rate bound to a much higher rate bound and it went way higher than anybody expected right the Fed was caught way offside and we went from the easiest money the country has ever seen to the most restrictive policy we had seen in 20 years within 10 months right so I think like don't use that for any pattern recognition except the next Global pandemic talk to us about 2025 what what are you what investment themes are you focused on for this year yeah so I mean again it reminds me of 2002 you know everybody got really because we you know we had some days like yesterday in the internet everybody's like oh the internet's a fad Amazon's never going to sell all this stuff it's like you go through these these moments of Despair and I hear the same things about AI right are you ever going to earn a return on these Investments yeah talk to anybody in business today your company will not exist if you're not leveraging AI in five years right in the same way that no company can exist without the internet right and so it's just the time series that we're really talking about so for me the Primitive to all Ai and I'm talking whether we're using chat GPT whether we're using Google search whether you're using Instagram whether you're using Microsoft teams whether you're using any Enterprise product because it's all going to rely on the exact same accelerated compute and inference and AI um that requires compute what I mean like AI compute Nvidia compute and that requires SK hyx and Micron it requires Ires you know the the whole custom uh as6 movement requires companies like broadcom and estera Labs um it requires power right like vistra or you see you know we did a pod from Diablo Canyon the nuclear facility in in California the reason we did that is because we said we need to set nuclear free in this country so now we see Amazon and others who are negotiating directly with nuclear facilities to set up AI data centers right next to them because they're massively power consumptive um so I think that whole complex is underappreciated yesterday took even more gas out of that complex and so I think it's one of the easier trades Nvidia trades at 24 times next 12 months earnings which by the way is one of the lowest multiples it's ever traded at because we have a wall of worry um so that's one number two Market leaders and margin expansion this was the meta trade right when I bought that when I wrote that letter and you and I talked and and you know backed up the truck on it in the fall of of of 22 every headline meta's dead it's over for meta Zuckerberg's driven the company off a off a cliff gersner is an idiot for this letter he Mark will never do what you know he's suggesting like you have to have the conviction that can withstand that okay but what did he do he radically expanded the margins at the business because he grew the top line and didn't grow the bottom line because of AI every market leader every market leader not just in technology has an opportunity to do that right I just spoke with one of our investments who's here today A great company and he said you know we've twox our Top Line over the last two years and our Opex is flat exactly build that into your DNA get fit leverage AI like this is the opportunity the companies that do that are going to have great runs and the companies that don't are going to be on the wrong side of history and there are going to be some seminal examples over the last 20 years our entire career for most of the people in this room if you invest in technology the architecture of Technology was the internet and Google was the apex predator okay why they gave us these 10 Blue Links that allowed us to have access to the world and for the first time in 20 years that is at existential risk why because my sons don't they're like why would I look through all these Blue Links I just want the answer chat GPT gives me the answer they talk to chat GPT at night it's their specialized tutor in my son's honors chemistry class or AP History class or whatever you name it this is the first generation that will be chat GPT natives right in the same way we we felt about the internet during that period of time and like those who grew up with the internet would say what you used to use a card catalog you went to a library what's a library and you pulled this thing out and you had to pull the card out and go find a book like that's how you got the answer right and so this is highly disruptive to the entire economic architecture so you have to look at companies like Google and say will it continue you have to even if they're great at AI they have the greatest business in the history of the world that's getting disrupted booking.

com Expedia Zill all the companies that lived in the underbelly of that search ecosystem disrupted right and so I get excited about that because we're moving the world positively in a great Direction but yes technology faces you know these waves of disruption and it can be very dangerous in Venture because Venture often takes 10 years you may make a great bet this year 3 years from now could be totally eviscerated so from my perspective we've done very little in early stage Venture in AI the discount rate has gone up what do I mean by that it means my ability to forecast the future in three years given the pace of change is very low so I have to be humble in the face of that if there's a startup company with 10 million in ARR that's growing 50% and everybody says oh you know a slap of 10 multiple on that and do the series b or do the series a I said I don't even know if the world's going to need that company in three or four years okay let me just give you one example I know you're going to kick me off stage but they took more time so let's us take more time exactly okay so I could I had a bunch of meetings the last three days and I'm sure somebody on my team who was in the meeting with me left the meeting went back to our Salesforce instantiation typed in these little boxes who we met with and what the conversation was about we have this CRM that keeps track of it and then when I want the information I'm like how do I get this information what I want to do is just put my phone on the table chpt listens to the conversation it summarizes the conversation it knows who I'm talking to and the next time I'm going to have a conversation with Ron I just say hey chat what was the last conversation I had with Ron yeah it just tells me that is a potential disruptive Force to application software that has never existed so I had a slide up there that went kind of viral on our pod when Sacha said you know application software is just a thin layer that sits on top of a crud database that's what he meant mhm right he's like that's all dead right okay so you're hanging out think about your Diversified portfolio I own salesforce. com I own Google I own booking. com I own all the market leaders and by the way all my friends and all great companies and they will all make the pivots necessary to continue to be great companies I I'm sure at least they're sleepless nights trying there are times in Tech where you can set it and forget it just let it compound you're early in a super cycle and you're on the right side of the way but there are periods of time when the things that you saw compound for the last 15 years in a straight line you know I had I had dinner with my 88-year-old mother on Saturday night and that this was a convers she's like how should I be using chat GPT Oh you mean I'm not going to use Google so much I I should sell that in my portfolio good thought Mom I know but this is is you know like this is where the rubber meets the road and so for us I get the great Fortune to get up every morning talk to the brightest people in the world and get to play anthropologists I have an incredible team and we get to try to answer these questions and we put together these charts and we try to put together a forecast for the way the world is going to evolve and we have incredible Partners who think we're okay at it you know and it's led to good results by investing across four super Cycles over the last 20 years I this is the most exciting point in my career so the thing I love about having you here is like there's literally no one more passionate about this business and just about the future of the world than you really just try to do this in two minutes but you got to tell us about your invest America project because what I love about it is it's a way for you to take all the things that you love about the country about technology and try and put it in the hands of way more Americans than it is today yeah well thank you and I remember one uh a great moment for me last year was you had Jared Kushner on stage and he came up just after me and you said to Jared hey Jared you hear what Brad just said about invest America and you know what do you think about it and he said he's crazy enough to make it happen yes he's like 99% of people come through Washington have a great idea throw it over the wall and that's it and that's that's not how startup it's not how companies get built and that's not how ideas happen yeah right I put thousands of hours into this and I'm happy to report so for those who don't know the idea is very simple that we have the only threat to democracy the only threat to the greatness and the competitive Advantage we have as a nation is that we attack ourselves and we have an unstable situation because 70% of the people do not enjoy the upside of compounding right they feel like the system is rigged against them that the oligarchs and the Rich folks get all the money because they they're participating in the markets but if you look at you know inflation adjusted real wages over the course of the last eight years they're flat people don't feel there's a reason people are angry there's a reason people don't feel like they're winning because they're not yeah right we got to get everybody in the game everybody on Team America everybody aligned with capitalism and we can do it very inexpensively everybody who has there are 3.

7 million children born in this country and when when you have a child in the hospital you fill out a form cuz that's how you get your Social Security number every kid has to do it that's how you know you prove citizenship at the time we do that we ought to we ought to give them an invest America account so the treasury Department will cause to be created 3. 7 million private accounts a year seated with $1,000 in the S&P 500 that they can't touch or trade but they open their phone they see they own a little slice of apple of Berkshire hathway Etc they the they call it the money page and it's managed by private Market participa uh participants on private label to the federal government what I would say is this we have bipartisan legislation moving through the Senate we have a house and a speaker who very much want this to happen and I'm happy to report that now we have an Administration that we have a tremendous amount of support and progress with um but you know everything in Washington is a long putt um I think we're going to get it done I think we can get it done soon I'm hopeful that it will be part of this reconciliation package led by Scott bessent and Kevin pass it out of the White House who are both big fans of it um you know there's the right thing to do for the country yeah right it will make us stronger because it gets everybody into the game and just the total cost because I'm helping out on Doge we're we're targeting reducing the the discretionary spend in the country by $200 billion a year the total cost of this is 3.