>> from the heart of where innovation, money, and power collide. in silicon valley and beyond, this is "bloomberg technology." with caroline hyde and ed ludlow. caroline: welcome to a special edition of "bloomberg technology ." we are live from the humanx conference in las vegas where we are talking to the biggest a.i. industry leaders, thought leaders, builders of companies,

decision-makers. plenty to come. it is the private sector, but we have to articulate the public sector. another key set of anxiety building among the latest set of tariffs and concerns of energy. vonnie quin has the latest. >> the market has been doing a three point turn several times. we are returning lower. the s&p 500 down about .8%. you can see the chart and the upheaval in the market. partially thanks to a truth social post on part of the president who said he's going to double the already 25% tariffs that were due to go on steel and aluminum imports at midnight. that will be 50% according to the president as of this moment. at that moment we saw the vix spike above 29, the highest it has been since well before december. only spikes past 25 in december. the last two days, the vix has been inching ever higher. top 27 a couple of times, top 29

today. we are below 28 now. in the middle of the panic, if you like. but we do have a good 130 or so stocks higher in the s&p 500, about 392 lower in the s&p 500. about 9% off of its highs. the nasdaq has been in correction and it has not gone out of the correction, down about 12% since the high less than a month ago. bitcoin rallied a little bit, hit 76 and change earlier today. they are hoping the market ever higher as well. if we look at the aluminum stocks, the producing stocks, a massive rally for the likes of century. also a rally for some of the others. that is as we await a response from canada. justin trudeau handing off this week to mark carney as prime

minister. we will see what happens if the tariffs go on tonight at midnight. 50% on steel and aluminum as president trump has promised. >>'s economic expertise going to be key at this moment. we have to think about what the public markets mean for the private markets, but also take a longer-term perspective with a man who brought the first check in into openai and has a long history with how we move from llm to application. vinod khosla, the person to talk to about artificial intelligence, about the broader economy. you are here because we have all been so focused on large language models. you among them with openai. but the application stage, think about how it affects us as humans, our health, mental wellness. you will be on stage with a portfolio company. >> i'm very excited about the

role ai can play in health care. i first wrote about it in 2012, how a.i. can do the role of doctors and therapists and teachers and other roles. so i'm very excited to see it all coming true. probably faster than i had anticipated. >> you been long time thinking about the implications for health care, you also said energy is going to be a key beneficiary from generative ai. what does it mean for ai and how can a.i. help solve it? >> it is a solvable problem. not an easy problem. but in fact, the same time we invest in openai in 2018 we made the decision to invest. we also invested in fusion, and nobody believed fusion was near term. i believe energy will be a soul problem by 2030. and ai will be a problem.

there is fusion that will suspect go on to retrofit every coal and natural gas plant in the world. so we don't have to build new power plants. but the sleeper is also superhot geothermal. geothermal has been -- it is 4%, largely irrelevant. but i think the technologies that deliver right, it can be 20%, 30% of u.s. energy and global energy easily. energy i'm optimistic about, but a.i. is very exciting. a.i. and health even more so. >> is there a limiting factor from an infrastructure perspective on generative ai? innovation we can see here and now? saying the biggest concern is power in many ways, what is limiting the startups? >> today, the big limitation is

not power. the big limitation is can we get enough chaps. building a semiconductor fab is a five-year process. it will take a long time. once we have that, we will have to power it. and we will need the power. planning cycles and power are pretty long. so i think we need to worry about it. but the world is doing a lot. in five years, could you add gigawatts of geothermal energy? yes. in the 20 30's, can you convert most co-plants to clean power and produce many tens of gigawatts with fusion? i believe that is possible if they are smart about it. if we retrofit these plants, upgrade these plants, and their capacity. >> how confident are you we are being smart about it? >> i think we are being pretty good about it. all that will happen if we --

prices will be higher for a.i. technologies and other things. and the markets will adjust to the demand and supply. i think power will be priced higher if you move fast enough. i believe a lot are moving very quickly. organizations like yours tend to listen to the experts. the experts are almost always wrong when it matters. they are right when you are extrapolating the past, but none of them invent the future they want, just extrapolate the past. that is the right approach to these large problems like energy, a.i., almost every large area including health care. >> you said back in 2012 when you are writing thought leadership pieces, 100 page documents about the after missing you have about a.i. and applications of it, when you think about the innovations from a large language model perspective, when you get upending to the public markets,

did that to you build in the optimistic nature? do you like what is happening with innovation at the large language model stage? >> i really love the model and the rate of acceleration. as the naysayers love headlines, clickable headlines, talk about flattening and other things, i have not seen it. in fact, i suspect we will see more acceleration from the next two years of a.i. capability at the model level, not just the application level. then we saw in the last two years, and last two years since december 2022 was stunning in terms of acceleration of capability. we will see even more than two years. i'm fundamentally optimistic. i think we will be in the domain in the next couple of years. whether it is 1, 2, 3 does not matter. >> will china be in the agi domain? >> china is doing well. you mentioned deepseek.

i like to characterize deepseek as one third of the technology was stolen. china is good at that. one third of the technology was clever hacks. one third of the technology was real innovation. give them credit for that and the rest of the world has learned. but by far the largest contribution that deepseek made, and they made a valuable contribution, was convince all kinds of young technologists all over the world to say isn't restricted to the model providers with a billion-dollar data center. anyone can innovate. freeing up people's belief systems of who can innovate is the largest contribution deepseek has made. . and i'm really glad for that. i think we are in a race with china. but by and large, we will see accelerating innovation. not flattening innovation. >> is very risk that innovation is stymied by lack of

consistency from a regulatory perspective? from a tariff perspective? >> i don't care about tariffs. those are short term. i would like to say we have to keep our eyes on 2030. what can we achieve and how much better we can do for humanity by 2030 with all of these technologies. little verbs like stockmarkets up and downs, i know it is your business, but i don't worry about them. i mostly don't look at markets. >> but what you look at is openai innovation, 4.5 innovation that may have fell flat on the market, but ultimately accompany steering to have normal command of the industry. and yet more command of the private market dollars. are you going to keep on reallocating? do you think these companies should stay private as long as possible? >> my view is openai is doing great. whether it is public or private, it will do roughly the same thing.

so if i have to pick openai's trajectory by 2030, which to me feels phenomenal if you can achieve it, i don't really care about short-term or long-term. if it converts to a regular stock, it will be better for the company. but either way, the model innovation will continue. and i believe they are in a very good position to lead the race for a.i.. a.i. will be a comparative area. i don't think it will be covered by one company. anthropic is doing a good job. google is doing a good job in terms of technology. i think we will see innovation. >> how is ai doing and how big of a disruption is it? >> i think elon has put together a good team. they will do a good job. and they have the resources. i think they're doing a good

job. right now i figure they are not only behind openai and anthropic and google, they are also behind meta-in the models. they are doing a good job. they are trying to catch up. we will see how it goes. the more interesting thing elon is doing his bipedal robots. my personal belief, and this is where the model industry has no situational awareness of this, if i can be a little impolite, those humanoid robots will be a larger market in the next 15, 20 years than the other businesses. if i had suggested 5, 7 years ago that tesla would be more valuable than the next 10 auto companies combined, and i know their stock price has been going down, everyone would have laughed with me.

even today if i said the robotics business will be larger than the auto business in the next 20 years. people don't give it much credibility, but i am almost certain that will happen. a prediction no one expects. >> you make predictions, and they often spot on. it was a joy to have you on the show. thank you for coming on. you have to take to the stage alongside the ceo, who is about the artificial intelligence impact on mental health. vinod khosla across physical a.i. and indeed in the application of generative ai. this look at the public markets he ignores, but the market is currently in a volatile state. s&p and nasdaq 100. we are looking were japanese companies are up to. golden dragon up by 2%. a lot of innovations on the model coming up in recent weeks. nasdaq 100 still .30%.

the war selloff since 2022 yesterday. the s&p 500 underwater. we focus on tariffs and an energy state of emergency. coming up, speaking with the key ceo, the snowflake ceo will speak with us. from las vegas, bloomberg technology.

>> from the heart of where innovation, money, and power collide. in silicon valley and beyond, this is bloomberg technology with caroline hyde and ed ludlow.

caroline: welcome back to a special edition of "bloomberg technology." we are live from the humanx conference in las vegas. it is time for stock of the hour. we are focusing in on snowflake. we are seeing a bounce back in shares across the board after yesterday was a pummeling for all parts of the market. whether in technology and will beyond. we are up 2.4% on snowflake. this is reporting market chatter around a potential snowflake ai security deal, potentially with google. we can get into all of that with sridhar ramaswamy. the context is one of market public anxiety. but private market exuberance around generative ai, adoption of generative ai, but also deals to be made. how are the conversations you are having with your customers going at this moment of public market worries?

>> most projects at snowflake are about the medium term and the long term. a specific goal in mind. we want to either reduce expenses or help customers make more money. from that perspective, it is business as usual. we watch the markets, and they are an indicator of how confident we feel about our future. fairly confident that part will settle down. at snowflake it is business as usual, interest in a.i., how people can get value from a.i.. we have thousands of customers now using it. and we are focused on the big value creation items we can create from here on. >> the data is integral when it comes to snowflake. the application of the data. in the return on ai investors you are seeing. >> we just started -- we decided to start going the long and the hard route. you pay us on consumption so there is no upfront -- to some

multimillion dollar deal before people are getting any value. so we are focused on different kinds of chatbots people can re-create, ways to give business users access to data in a quick way without needing to go through painful dashboards or analysts, things like that. and the next generation of applications we are helping customers build built on top of these. the workflows to innovate, how we put workflows in front of people, sales flows that centralize the innovation they need to do their job. those are the things we talk about. every project we do is focused on value creation. we don't recommend investment for the sake of investment. that gives us a pretty rational framework to have this conversation. >> i'm the ceo you're pitching, try to reinvigorate me with the idea of updating my legacy software, ensuring i have access to the data we need. what is pulling me back from

saying snowflake right now? >> every migration takes time. every migration induces business risk. the system has to function until the new system can replace it for the use cases they used to serve. the act of navigating this is a challenge. this is where we have expertise, we've done migrations from every kind of legacy system possible. some of these projects can take a year to get done because of investment on the part of our company to get the data housed. so that process is slow. what ai has done is vastly increase the impetus for getting data to modern platforms so we can create additional value on top of that. maybe you had a supply chain system before and it is working ok. but can you optimize it? can you optimize how you purchase? you can do that more on a platform like snowflake than a legacy platform. how do you get real-time access

data to all of these? that is the kind of thing we do very effectively with partners like blue yonder because they cannot only modernize, but they can let their customers do things on top of snowflake that they would not have been able to do. it is that imperative that drives a lot of our conversation. >> you have partnerships with microsoft and openai, you are thinking about food chain venture and the latest start of to build upon your applications. i'm giving 30 seconds to answer this, but the talent pool. how confident are you that we can have great talent in the u.s.? >> you asked three separate questions to answer in 30 econds. absolutely partnerships are important. and we have them with the best model makers in the world. what we are able to offer customers with partnerships with openai and philanthropic is this guarantee that these models are a part of snowflake.

they don't have to worry about what is happening to their data, we make the same guarantees to our customers, data is used only for our customers. investing in a startup ecosystem, different companies building on top of snowflake because we think access to the customers, to the data, to the technology we provide can go to market. but the final question is really the world of ai five years from now is going to end up with very different skills needed for doing day to jobs of all kinds compared to what it is today. wings like investing in innovation, upscaling people, make sure they have the skills and to be able to build on the talents they have. that is how i think about it step-by-step. >> you managed to do it all in a little over 30 seconds. the snowflake ceo, sridhar ramaswamy. a joy to have you here. let's get a check on the public markets.

the nasdaq 100 has been a volatile trading day amid the barrage of news when it comes to tariffs and energy. state of emergency in the u.s. rocked by .50%. this is "bloomberg technology." ♪

>> let's get a check on the markets now that i have been doing u-turns all day long. closer to lowe's on the s&p 500, point 5% from correction territory. we had a response from canada. willing to shut off electricity exports to the u.s. in response to what caused the markets to be so volatile earlier, president trump decision to double tariffs on canadian steel and aluminum imports from midnight tonight to 50%. we had the resumption of the dip buying rally. we are at the lows. s&p 500 seeing more than 400 companies lower.

nasdaq 100 is also down by more than two thirds of 1%. we are seeing the vix index also higher, much more coming up. this is bloomberg. our xfinity network is built for streaming all the stuff people love. how can it get any better? -i'm just spitballin' here, but, what if we offer people apple tv+, netflix and peacock? for one low monthly price. -yes. so, people could stream the shows they love. and we could call it... xfinity streamsaver! mmmmm. what about something like: streamsaver? ooooooo. -i love that. add streamsaver with apple tv+, netflix and peacock included for only $15 a month... and stream all your favorite entertainment, all in one place.

>> from the world of politics to the world of business. this is "balance of power."

live from washington, d.c. joe: wall street resumes the selloff as president trump announced bigger tariffs on canada. welcome to the fast a show in politics. with the president set to address the business roundtable after the closing bell. as we witness the intersection of wall street together. i'm joe mathieu. kailey leinz in washington. welcome to the tuesday edition of "balance of power." the president reacting to retaliatory tariffs, so the cycle has begun. kailey: the province of ontario putting a surcharge on electricity, exports to three u.s. states, and president trump responding what he views as in-kind what was slated to be effective midnight, 25% tariffs on steel and aluminum imports, still the case for every country that applies to beyond canada will be a 50% rate for our neighbors to t