10 of The Wildest Hot Takes From the All-In Summit 2026

0:00All right, we have a live one. Brad is a very special celebrity because people come up to him with how much he's impacted their life. I have to ask, we had some incredible programming here. Jensen, we had Elon chime in with Gwynne. What is your hottest take from the All In Summit this year? Well, I mean, obviously, we had a lot of controversy at the end of last week that I was quite critical of on Friday, people coming out saying there's a 10 % chance of extinction event from AI. Then we had the Dario essay that I thought was a good path forward around independent monitors on Saturday, a lot of controversy about how much that meant slowing down.

0:38We had Elon come here and say, listen, I tend to agree with Dario. The risk is augmenting, and we need to have peer review, independent peer review. I thought that was a great idea. Jensen said, listen, the labs themselves need to control it. They need to do a better job. They need to get this under control. The president calling in while Jensen's on the couch and chiming in. So everybody thinks there's chaos. I don't think it's chaos. I think what we're doing is we're open sourcing in real time how consensus forms. The sausage making is kind of ugly. But if you really just think about everything we heard, it's boiling down to a pragmatic consensus that yes, there's more power and more risk than we may have thought of.

1:19So we need to act responsibly. It needs to start with the companies themselves, but we probably do need some sort of peer review. We do need to talk to China about whether or not they're willing. Elon said yesterday, maybe they'll participate in open source peer review on this. So those are important conversations, important things to explore. I think sometime in the business of podcasting or X, everybody wants you to join a tribe. You have to be all on this side or all on that side. But the constructive solution around Trump accounts and the constructive solution here is going to be a pragmatic, commonsensical middle way.

1:53And I think you saw that unfold here. I mean, it's incredible. And I just want to say hats off to four of my besties who put this together. I mean, it's far more than a podcast. It's a community that I think is advancing important conversations. You're part of that. But to watch that unfold in real time on stage and to move the conversation forward, something I'm proud to be a part of. I don't think anybody who came here will forget that, Roman. Yeah, it's extraordinary. Seeing history unfold in front of our eyes. Well, congratulations to you on all the success. Thank you so much, Brad. Really appreciate the time.

2:28We are here at the All In Summit with the one and only Naveen Rao, founder and CEO of Unconventional AI. Naveen, welcome. Thanks. I'm happy to be here. So you're taking center stage later on. What are you going to be talking about? Well, I'm going to be talking about unconventional. Made a ton of progress and actually going to be making some public disclosures to the world about real results on real hardware. What are they? Well, let's put it really succinctly. We're actually able to demonstrate many orders of magnitude of efficiency. I'm not going to talk about all of it right now, but we're talking like nanojoules per image generated.

3:05So this is something that no one's ever really been focused on. And when you compare it to existing paradigms, we're thousands or even tens of thousands of times more efficient than existing computer chips. And so what will be the macro impact of a more efficient chip? Well, I think initially it'll enable us to continue scaling. We're going to run out of energy in the next three years or so for scaling AI. So immediately we'll be able to continue scaling. Then after that, you'll start to see these drastic disruptions in price. So that's a good thing. Bringing the price of intelligence down will make its applicability go through the roof.

3:40We'll be able to apply it to your toaster, to everything, because now it's economically viable to do so. So there's a lot of hot topics floating around the ether right now, if you haven't seen. But I'm really curious for your take. What is your hottest take right now? Well, I think I'm the opposite of a doomer. I think AI is the next level of evolution of our species and takes us to places that we simply couldn't go in terms of knowledge and development of technology. So for me, I think it's actually kind of silly. These scenarios have not really been proven out in any way that it's going to end humanity.

4:17I don't think we have any credible evidence other than vibes right now. So that's probably my biggest hot take is there. Show me a scenario. Coordinated vibes. Coordinated Vibes. It's literally Hollywood. I posted on X about this, but go watch Transcendence. It's literally the same thing that everybody's talking about right now. To me, it's like, this is actually kind of silly, and it feels like it is financially motivated. Because the simple thing to me, if everybody is so concerned about this, stop working on it. Don't go IPO. Don't make a lot of money for your investors. If you're really that worried, why don't you do that?

4:52That's not happening. So to me, it's very hard to disambiguate that there is financial gain to be had here. It was interesting. I know you were in the audience for this too, but Satya Nadella took the stage and he was talking and comparing this to when they would have bugs at Microsoft. What would you do when you encounter a bug? Well, you fix it, right? And so I think he was very practical in those statements. It's like, you have a bug, you fix it, you move on, right? You keep building the technology. You make it safe. You put the guardrails you think you need. And I think there has been a ton of work on this.

5:23Frankly, the reason ChatGPT could even happen was because we were able to make it not go in bad places every time you chatted with it. So we have done a lot of work there. And I'm not saying there isn't work to be done. Of course there is. We want to make bots that don't take people into suicidal tendencies, things like this. Of course we want that. I'm not suggesting that there are no problems here. But the reality is, show me a scenario where it goes rogue and takes over. None of that is true. The rhetoric we're hearing right now is models copying themselves, complete fabrication. They don't do any of that.

5:59They need infrastructure. They need tons of hardware to even run. So to me, this is just fear-mongering. Fear-mongering. Also, I've heard that the biggest concern in how people think they're going to die is by humanoid robots. Right. Which don't actually exist. Maybe they will exist at some point, sure. And can't even run in the United States. That's right. I mean, I don't know if you saw the Olympics for the robots, right? They can run in a straight line maybe and then crash into something. So it's going to be a while before we have that. And I think we're going to be developing those safeguards and, you know, sort of moralistic values that robots need to take on to be deployed at scale.

6:36But we're going to get there. I just don't see a scenario where we don't develop those things. All right. So as we close out, for people who don't know what unconventional is, what is unconventional? How did you get here? Yeah, so we're rethinking how a computer actually works from the ground up. And a lot of it's based on concepts that we've uncovered from the brain as to why it's so power efficient. And we want to apply those into an engineered system. And the reason that this really hasn't been done is that computers have always been built to go faster, always faster and faster than the alternative.

7:07And we never really cared about power efficiency because it didn't matter. Today, we're in a strange place where computing is going to be power constrained at the global level. So we need to think about this front and center and really that's what we're doing at unconventional is thinking unconventionally about the problem like remove Assumptions that you've had for many many years and start to think about the problem differently and build something Much much more powerful thousands of times more powerful. All right. We have a second hot take here with Naveen Naveen What is your second hot take?

7:35Well, I think we're converging on some interesting Big blocks of value capture so energy is where it all starts then energy is converted to intelligence with hardware. So energy to hardware to intelligence. And I think in 20 years, there will be very little value to software. Even though it's created so much value in the preceding 50 years, I think at that point it will basically be a de minimis part of it. So every company that's basically selling software as a value add will kind of have to pivot to something where it's like really about delivering intelligence in an efficient way. So the SaaSpocalypse is back?

8:12I think it's absolutely real, right? I mean, we're seeing it today, but it's more of a perception. But I think it's really going to happen later because the user interface starts to become AI. Like, it just becomes an intelligence thing. You think about a human. Like, you get a human who's very good at a particular task, and, you know, they can add value to a company. They don't have a software layer on top. The software layer is, hey, can we please go do this thing, right, or go build this product or whatever. So it's very high-level instructions, and it basically dissolves that kind of interface layer.

8:45What happens to all the NeoClouds? Well, the NeoClouds are part of that hardware delivery, I think. So they're converting energy to hardware and sort of the facilitator of that in some sense. So I think there's going to be value there. There'll be a lot of money that runs through it, for sure. The gross margins are kind of TBD, I would say. I just don't know. But hardware is clearly going to have high gross margins, and so is energy. Awesome. Cool. Thank you. Thank you. All right, we're here with Jonathan Neiman, co-founder and CEO of Sweetgreen. Welcome to Sorcery. We're here at All In. It's great to be here.

9:19Well, right before we started filming, literally everyone in this room ran out, and we just found out why. We got the legend Jensen here. Jensen Wong. He's going to give us a good take on lettuce. With a very special leather jacket he's got on today. I don't think it's vegan. Okay, so I have to ask you your hot take. There's a lot of, okay, people think AI is controversial. We've got some controversy in the food industry, specifically with lettuce. So what's going on with the lettuce? All right, so my hot take is that AI is not going to kill us, but processed food will. If you followed the news this summer, you saw a lot of news about a parasite called cyclospora, which is a real thing, so I don't want to discount the reality of it.

10:06However, I think that it was blown a little bit out of proportion, and what we did is we took something that happens every single summer for as long as we know and used it as a way to scare people away from eating fresh and real foods and moved them towards ultra-processed food, which we know will kill you. And so, again, my hot take is that AI won't kill you, but ultra-processed foods will. We got a lettuce psyop going on. So we can't eat lettuce. The good news is, as of last week, it's been declared that cyclospora is over. Oh. So it is officially safe. But even more so, the whole time Sweetgreen was safe and 99.999 % of places you were eating lettuce were safe.

10:48There was one farm from one supplier that was serving iceberg lettuce to Taco Bell that was unsafe. And unfortunately, they scared a whole industry, you know, the whole country from eating not just lettuce, but blueberries and any kind of food. And I think that was irresponsible because the reality is, is yes, we should have avoided that one one supplier. Maybe don't go eat iceberg lettuce from Taco Bell. I don't know why you're eating Taco Bell. Anyways, you're going to get diarrhea. But sweet green. We know our farmers. We source responsibly. That's the whole what the whole company was built on.

11:23And so lettuce is safe. And one last thing, sweet green lettuce is only a small percentage of what we sell. We serve a lot of hot and protein-forward food as well. What's your favorite meal at sweet green? My favorite meal at sweet green is probably the miso salmon plate or the chicken, classic chicken Caesar wrap. What's the secret about sweet green we should all know when we're ordering? Get two dressings, two different dressings, and mix them up. It kind of like unlocks flavors in crazy ways. It's like fancy sauce? Yeah, exactly. You mix the green goddess with the miso, for example, and you get these amazing flavor bombs.

12:02For those who have not visited every location, what's your favorite location? My favorite location. You're asking me to pick from one of my children. It's hard to do. There's ones that always have a special place in my heart because of when we open them. like our first store ever in Georgetown, our first store in New York at the Nomad, our first store here in West 3rd Street. So those are kind of the ones that mean a lot to me. There's nothing like a daily sweet green addiction. I know everyone in New York can definitely attest to that. We did just launch a 30-day challenge. Really? I don't know if anyone saw.

12:41So we had a customer that decided he was going to create a 30-day challenge for himself and only eat sweet green for a month, lunch and dinner. and he tracked how it made him feel and what it did to his life and the results were phenomenal if you haven't seen the video you can see it on my x account but it went super viral people loved it and so we launched a 30-day challenge that we're now doing so we're announcing it this week we have 30 people who have signed up we have hundreds and hundreds of applications and so if you want to join we would love to have you but if the challenges is 30 days of sweet green we fund it and you tell us how it makes you live your best life and how it makes you feel.

13:19Was that person Brian Johnson? It was not Brian Johnson, but he is a friend and a customer. Amazing. Well, Jonathan, thank you so much. It's so much fun. Thank you, Molly. Cool. Hey, y 'all, welcome to Sorcery. We're here with Toby Rice, founder and CEO of EQT Energy, America's champion energy producer. Of natural gas. Of natural gas. You don't have to do the Texas draw because I'm actually from, grew up in Massachusetts. Oh, okay. So from Boston. I don't know if you notice where we're at today, there's 38 states in this country that produce oil and gas. Massachusetts is not one of them. Really?

13:54Yeah, and if you want to talk energy, the biggest gas field in the world is not in Texas. It's in Pennsylvania, western Ohio, the Marcellus Shale, and that's where EQC operates. So we're going to Pittsburgh. That's it. Okay, we're going to Pittsburgh. All right, Toby, what is your hottest take? My hottest take right now is that there is a tremendous energy demand for natural gas in this country, thanks to the AI revolution that's taking place and the massive amount of power generation that's needed to fuel it. It's going to be powered by natural gas. We see a 20 % increase in natural gas demand to fuel this power generation.

14:29Important thing for people to understand, power generation needs is not just based on the needs for AI. We think that's only about 40 % of this demand. The other part of it is getting back to keep Americans' energy bills low, power manufacturing, and people need to step back and realize that we've shut down over 174 gigawatts of reliable power generation, mostly coal and nuclear. And that is a massive hole that we need to backfill with reliable, affordable energy. Natural gas is going to pick that up and also meet the AI needs. The other dynamic that's taking place in this country right now, and we see what's happening in the Middle East, energy security of the world is crippled.

15:10And you're seeing this play out with what's happening in Iran. Global energy security is national security. Every country is felt by the geopolitical impacts. It highlights the second big driver for natural gas demand in this country, and that's providing energy security to the world. And we express that through US LNG, liquefied natural gas. We went from producing hardly any natural gas in 2016 to now we are the world's largest LNG exporter as a country, something to be very proud of. And going forward, we see natural gas demand for exports to increase another 10 to 20 BCF a day. So you step back and you look at, for a market that's around 110 BCF a day, we see a 20 to 40 % increase in natural gas demand.

15:56It's a lot of energy, but people need to understand we've got the resource in this country to make it happen and Companies like EQT are gonna make sure America gets all the energy they need so they could run full throttle damn Well, okay. What is the biggest misconception in energy? I'd say there's a couple things that we're seeing this play out right now in the AI Actually one of the biggest challenges we're seeing with AI It's been a big focus of this conversation is the pushback on on artificial intelligence, the pushback on data centers. We're in a world right now where the good news is that everybody's talking about artificial intelligence, everybody's talking about energy, that's the good news.

16:35The problem is that the conversation starts with fear. And people are concerned about what they're going to lose. People need to be concerned with what we can create. And with abundant energy coupled with intelligence, we have the opportunity to create abundance. One thing that people don't understand People do not understand how much energy is needed in this world. It's a big thing to get your head wrapped around. But understand this, 3 billion people in this world live on less electricity than it takes to run a fridge. Think about that. Really? Billions in this world are living in poverty. And when we look at what are the ways that we can help eliminate poverty, giving people access to energy is one of the fastest, most efficient ways to lift people up.

17:19When people get more energy, the quality of life improves. The longer they live, the more energy people have, the wealthier they get. And this is the opportunity we have for the world. In a world where people are saying we need to constrict and restrain our energy production and consumption, we actually need to be expanding that. Because to give the billions in this world, in addition to providing security for the developed world, energy security, we would need to triple the amount of energy that's produced and consumed in this world. All forms. Triple it. Solar, wind, natural gas, nuclear. We need to triple all forms of energy.

17:50And if we can do that, the amount of wealth that we're going to create, we're going to increase our global GDP five times over, in addition to lifting billions out of poverty and the moral benefits we get of doing that. So whether you care about energy, you care about business, or you care about making the world a better place, you should care about energy. Double damn. Where's that Texas dam? Damn. There we go. What else you got? I mean, listen, that's a pretty big one. But this is one of the reasons why we're here, because it's what we've seen in the political debate with industry. You know, the oil and gas industry, really the ESG movement that took place, swept in this country, really starting in 2018 up until this new administration took over, was a very dark cloud that made it very difficult for market forces to get the cheapest, most reliable, cleanest energy to the marketplace.

18:44And that's caused, had some very negative impacts. Americans' energy bills are up over 40 % because of what happened when political force overrides market forces. And unfortunately, what we're seeing happen with AI and the pushback that's coming to place, and this pushback is real. I mean, 80 % of the public saying that they don't support data centers. I mean, ICE is more popular in this country, the immigration. ICE is more popular than data centers right now. We're seeing a lot of the techniques that were used to help shut down energy development in this country are now being used to shut down intelligence.

19:20And it's something that we're very concerned about. But I'm optimistic because with all that force that was put on the energy industry, we were able to get through that. And we were able to get through it. And now we are the world's largest energy producer. America has gone from being energy dependent to an energy powerhouse on the world stage. We produce more energy and natural gas than any other country in the world. We've got some amazing cuts like EQT. If we were a country, we'd be the 12th largest natural gas producing country in the world. So we can persevere, but we really got to start talking about the pros that we have.

19:50And one of the biggest opportunities that I see that people aren't talking about, and this is why we have to get it right, because if we can bring energy into this world, coupled with artificial intelligence, that is the recipe so that we can create abundance. Abundance for Americans, abundance for allies, abundance for the billions that are living impoverished lives. That's why it's so important we get it right. and how does artificial intelligence help create that abundance? Energy is not going to be the only thing that helps make people's lives better. We need to make sure that energy translates to wealth.

20:20And intelligence in the form of education, in the form of technology, in the form of adoption, in the form of running businesses and creating smart policies, these are all the things where intelligence is going to help make sure we have a very high correlation and a very successful translation. We're not just playing for data centers. We're playing for abundance in the world. This is a much bigger playing field and this is one of the most exciting times to be a part of this debate And we're excited in our role to lead a business as important as EQT to help make sure great things happen with the application of industry.

20:49Okay, Toby Rice for president. Thank you very much, Toby. This is awesome. Thank you, Molly. We are here with Wen from GenSpark AI. Wen, welcome to the All-In Summit. Thank you. Thank you, Molly, for having me. So for people who don't know what GenSpark is, what is it? So, Genspark is a Palo Alto based technology company. We built AI agents for knowledge workers, the one billion plus global knowledge workers. So in the past, we human beings, we have to do so much grunt work to get the real work done. We have to craft every document word by word, presentations pixel by pixel. With Genspark, you no longer have to do that.

21:25You command your Genspark super agents, they would run around and then get the busy work done for you. You focus on being strategic and creative so everybody could work like a Jensen Juan, right? Like he says I love working that's because he doesn't have to do the groundwork And so goes with Jamie Dimon. He has a fleet of associates running around to do the work for him So yeah, that's what we do. We put a fleet of Goldman Sachs analysts in your pocket so you can be strategic and creative Jen spark work like Jensen Okay, so what is your hottest take right now? My hottest take based off this morning is we won't slow down So look, our view at Genspark is this.

22:07To Gens' point, over the weekend on X, AGIs are out of here. But it's not everywhere. Not only because it is domain-focused in certain areas, you see a glimpse of hope that, oh wow, it's so good. But also because if you look at today's world, the vast majority of the people have not touched AI. The heavy users are the people who can maneuver the coding agents. That's in a single million. People who are paying 20 bucks to do a genetic AI is in order of magnitude 10s of millions. People who have free chats are in order of magnitude of hundreds of millions. The vast majority of the world has not touched AI just yet.

22:42And why? It's not because it's not there. It's because it's hard to maneuver. I'm a mechanical engineer by training from MIT. I got my PhD from MIT. I don't know how to code. It's been decades. It's been years, not decades. I'm not out. But the thing is, I don't even know how to draw the full power out of Fable 5 with my bare hands. I don't know. So imagine for the 1 billion plus global knowledge workers. I mean, if you don't code, it's like if you use ClockCodeWork, if you ask the code work to build a presentation for you, it asks you to download 10 things from the GitHub. I don't even know what those 10 things are.

23:18So we build a product that is so easy to use that you don't feel like you need to drive a Formula 1 racing car to get the full power out of those intelligence engines. We build an FSD vehicle. You just tell us your destination. You don't even need to touch the steering wheel. We get you to the destination. It should be as easy as how you use a Google. So that is kind of the bridge we're trying to build, and that is the gap that we believe that is currently the biggest problem for AI. If people don't have easy access to something, they don't know about it. If they don't know about it, there's fear.

23:53Step zero is you got to create opportunities for real human beings, for normal people to get access to your tech and benefit from it. Turn tokens into values. And then there's a lot less fear. There's a lot more. I'd say, so basically don't leave. We don't want to leave anyone behind. Don't leave anyone behind. We are in this together. Let's do this together. That is my hardest take. Yeah, it's not about slowdown. It's not about slowdown. Safety is absolutely important. When NVIDIA launched the Open Secure AI Alliance, we were one of the very initial members. We're now contributing to the entire community on how to map out the telemetrics of agentic AI, especially in the productivity world.