AI Leaders Reveal the Next Wave of AI Breakthroughs (At FII Miami 2025) | EP #150
31:13 Share

Welcome, welcome. We're about to have a conversation and I want you to listen up. This is the technology that's going to reshape your families, your lives, your businesses, your industries, your nation states. And the question I'm going to be asking at the end here is, are you ready? So let's dive in.

I'm going to be going through two rounds of questions. I'm embarrassed that we should, in fact, have a three-hour session for our panel, but we've got 30 minutes, so you'll excuse me as we run through this. Prem, I love what your company has done, and you're an example of a CEO who takes a company and, like, doesn't 10x it, you 100x it. Stability, AI, what are you doing, and how are you going to impact the world?

Thank you for that. So Stability is the creator of Stable Diffusion, which launched August 22, which changed everything in image-based AI generation. It was the chat GPT moment for image.

There's been over 270 million downloads of stable diffusion to date. Give you a sense of scale. The number two most popular model has been downloaded 9 million times. So it is by far the market leader. What we're using it now for is my background is in professional film and television, and we're now fine-tuning...

what I call ultra-narrow AI, fine-tuning our model to work in professional content creation. So film, TV, gaming, and marketing and advertising. So you brought Jim Cameron onto your board. You have investors. I'm an investor, full disclosure. You brought in Eric Schmidt, an incredible group of individuals. How far are we from creating reality given the technology that exists right now?

We're already there with certain workflows. Now, what we're doing with... To make this a full reality, we're doing exactly as the artist creates a film. So, what we've seen in other text-to-video models are one text and one video.

That's not how professional content is created. Professional content is created in shot elements and then they're composited together to make the shot. So what we're doing is going step by step in each one of those processes, whether that be rig removal or paint rotoscope or camera match, plate reconstruction and doing hyper narrow AI models around each and every step.

We're probably about two dozen models in and about 50 to 60 models. How far are we from me starring in my favorite episode, Star Trek? You should be in that now. I know. If it was up to me. But in all seriousness, in terms of video generation on the fly, where I have a request to create something extraordinary that looks real.

I would say we're probably, that's going to happen this year, I think within six to 12 months. So how does your reality change when you're not sure if you've created it or if someone else has created it? What's possible for you in your businesses, in your lives? All right, we'll come back to you in a moment. Alexander.

Ramin, again, an investor in your company, so full disclosures over here, Liquid AI. You came out of the gate, zeroed a $2 billion valuation in just about two years. At the end of the day, you're enabling private AI capability with your Liquid models.

So, I know a number of companies that are fearful, they don't allow their employees to use ChatGPT because they're concerned that OpenAI has access to all the data. So, what's possible using Liquid AI? Absolutely. So, yeah, we are a foundation model company. We are building AI system, generative AI systems for enterprises.

And we're providing these systems, we're powering these systems by a new technology, not the transformer architecture that enabled the new wave of AI. We have created something on top of something that we invented at MIT, liquid neural networks at Daniela Russo's lab. These are warm-inspired AI that...

We evolve these kind of systems into something that is more tangible and now we can create value off of this new type of AI. The very special thing about this technology is that the amount of compute that is needed for packing a lot of intelligence into a device is very, very minimal. So as opposed to other type of AI, you can get chat GPT experience on a phone, locally on a phone, on a laptop. And...

in places where privacy matters. From a product perspective, what we did today with enterprises, wherever there is sensitivity on data or security issues that you cannot actually use a cloud kind of solution, or you don't have access to GPU-based infrastructures,

This is the places where liquid can immediately come in and expand at large, like putting our type of enabling enterprises to use generative AI. I mean, you just raised a killer round with G42 as one of the leads, a quarter of a billion dollars. Congratulations on that. Thank you very much. You're tracking on revenues this year, which are spiking, which is fantastic. I remember you used your liquid neural networks to fly fighter jets. Can you take one second about that?

Yeah, well, I mean, this was one of the only neural network architectures that enabled safe applications of AI on device. So they trusted, United States Air Force trusted our technology to be the first version of a neural network that can navigate autonomously a flyer jet. And you can imagine, like, private AI is not just having a chat GPT experience on the phone. It can

can power cars. It can go on a satellite. It can go on a jet. And the applications of these things is phenomenal. Like recently I talked to some CEOs of an education company that they're providing tablets to students and they want to have an experience in this kind of sector. This is your educator, your physician, your pilot, your everything. I mean it's innate intelligence on your device

that you own. - 100%. - Amazing. We'll come back. A dear friend, Jack Hittery. First of all, I have to point out, Jack is dressed in Miami. - We're in Miami, Peter. Welcome everyone to our home in Miami. Richard, thank you for bringing us to Miami. And I have a request for Richard, Peter, if I could. - Yeah, please, of course. - For next year, for FII Miami,

could we request that the dress code be Miami business? Does everyone agree? Yes? Okay.

And by the way, Miami business means wearing avocado dress socks. We have avocado dress socks. There's many others to choose from, though. I do want to do a commercial for Jack's U.S. book, AI or Die. It's very subtle, Peter. But it is something that every person should read. It is literally what you need to understand as a leader about AI, and it's written in a very readable fashion. So, Jack...

You are the CEO of Sandbox AQ. A is for AI, Q is for quantum. You've got an incredible chairman of your board, Eric Schmidt. You spun out of MIT with... Google. We're very involved with MIT also. Out of Google with an incredible seed round, I think, of how many...

We raised $850 million now. Wow, amazing. All right, so what do people need to know about Sandbox AQ and the quantum liquid networks, the quantum networks that you're producing? Language models.

Peter, it's a very exciting moment. First of all, it's great to see so many friends on the panel here and in the audience. And it's a beginning moment. This is the incipient moment for AI. Everyone's excited. Lots of businesses looking at it. But I think we're past, hopefully, the shiny object phase of AI. And now it's getting serious. Everyone on this panel has very serious offerings that really impact business.

and impact how Hollywood works, how many, many parts of the major sectors of the world work. At Sandbox AQ, what we realize is that language models are fundamental. Everyone, its table stakes should be using language models to cut costs. If you have customer service because you're Delta, you're Hertz, you're Hilton, any company with thousands of customers must be using large language models to cut those costs and actually deliver better customer service.

I think we all know customer service cannot get worse, you know, as it's delivered now. So it's only going to get better. But we at SandboxAQ decided, Peter, let's actually go for a different part of the economy. Let's go for the quantitative AI, not the language AI. And what do we mean by that?

If you're Sanofi, if you're a drug company, if you want to create a new medicine for cancer, for Alzheimer's, for dementia, each of our families here in this room, unfortunately, will all be impacted at some point in our lives by these diseases. Language models can help initially when they scour and look at all the summaries of scientific literature. Very, very helpful to give you some ideas about what's been done before.

But ultimately, Peter, if we're talking about building a molecule, we need an AI that is not trained on social media and cat pictures, but is trained on molecules and atoms. That's fundamental. And that is the AI that Sandbox AQ is the global pacesetter in. Very importantly, you're not talking about using quantum computers to run these quantitative...

That's correct. You're using, you know, current... Quantum equations on the GPUs. So what kind of quantum equations are you using on the GPUs? Right, so this is something that everyone, of course, we all know from elementary school, Schrodinger's equation, all the equations that everyone is familiar with, but what we realized, the breakthrough that we had is that GPUs were getting so much better, and we have a hardware person on the panel representing Token, hardware person, thank you, on the panel,

Fundamental to GPUs is the ability to run in parallel matrix algebra. Imagine a spreadsheet like Excel times another big spreadsheet, a million rows by a million columns, and a million rows here by a million columns. That magnitude of matrix algebra

we can actually convert the quantum equations, the equations of drugs, of treatments, of new energy, of battery storage, all that we can convert to the language of that GPU. And that's the breakthrough that we had. Now, when quantum computers come in scale, we just had a great announcement from Microsoft, people may have seen that yesterday, another announcement from Google just a few weeks ago. These announcements you're gonna see come in a great cadence, culminating in a crescendo, Peter, in about five to seven years.

of having great quantum computers, we'll add those to the arsenal. We'll have GPU, QPU, quantum processing unit, in one mesh cloud hybrid. But today, Peter, we use the GPUs to get the work done with drug companies, with Aramco, Aramco I see is a sponsor, our newest announced customer is Aramco in Saudi. Why Aramco? Because they want to take the hydrocarbons coming out of the ground and convert them to higher order

chemicals using carbon and hydrogen, not low-grade fuels, but carbon composites, as an example, that could be used to make a car lighter, that could be used to make a space rocket lighter, that could be used to make an airplane lighter for Airbus or Boeing. This is the kind of transformation that we focus on with LQM's large quantitative models versus the very necessary large language models. Amazing, amazing.

Yeah, Jack is a nuclear power plant behind that man. All right, Jim Keller, Ten Storen, you're a hardware manufacturer. You're our sole hardware manufacturer against all of these software geeks. Congratulations on your recent round. Thank you. Pretty good, $700 million of a Series D round. So you've got a lot of capital to build hardware. Yep.

I'll take that. So what kind of hardware are you building? And when someone says, no, I only do software, how important is hardware versus software today? Yeah, so...

So GPUs got a real solid head start on building AI because they had parallel computing. But they're still relatively complicated to program, and the way they do, like, handle tensors and stuff actually wasn't native to GPUs. Now, GPUs have evolved to add tensor processors. TenseTorrent builds a native tensor processor that's simpler and easier to program. Also, we build it so that the tensor processors natively talk to each other really nicely.

And then we, last year, we open sourced our software stack. And it's really interesting, the fundamental math AI is simple. A equals B times C plus D. Like it couldn't be simpler at some level. But the scale of it is amazing. When I started building computers 40 years ago, we were doing millions of instructions a second. Now we're doing trillions of trillions of instructions a second.

And to scale that takes a special collaboration between the hardware and the software. So when your machines are up and operating, I guess the question is, what are they enabling for people in the room here? Well,

Well, right now there's a really large family of models. So our mission is to run all the models with really simple transparent code. So big LLM, people say, oh, the software is huge. Actually, it's 600 lines of code. It's not very complicated at the program level. But when you go down in the software stack, it can really explode.

And that's where by building a native software stack that is tensor-based, it's communication-based, and it's open source, people can see exactly how that works and how it runs. And I think that's going to unlock a lot of AI applications that are currently hard to program with GPUs. So there's been a lot of debate on open source versus closed source AI models.

And we just saw the R1 model being open source. We're seeing a lot of conversation where the leaders are saying, you know, open source will win. I think there's been an extraordinary velocity in open sourcing. How important is open source as far as you see?

Yeah, so I have personal experience working with GPUs where we're trying to solve a hard software problem. We couldn't because the math library was encrypted or part of the software stack was proprietary. And because we couldn't look all the way down the stack, we couldn't figure out the problem and solve it.

Now, open source AI is really wild because most of the high-end research is published. Many of the models are open source. Some of the weights. Not that much of the infrastructure. Not that much of the foundation library. So it turns out to be kind of a mixed bag. Like one thing we're going to do is we're going to open source a whole software stack. And it's for our hardware, but I encourage people, if you have your own hardware and you want software stack that works, you know, steal our software. It's a beautiful thing. And then...

I want to make it so many of the foundation models, the environment, the framework to build and train your own models are also open source and available.

And I think it's really important to democratize the hardware stack and the software stack so it's not just a few very large players that control the AI world. So a lot of people from around the world here, are the machines that you're building likely to be used in the global south more than in North America? No, we're going to sell to everybody. We license a small AI configuration to go on a television chip, and we're building machines that can train large language models and everything in between.

and the other part of our business model. And again, I think innovation comes from lots and lots of input. So the software's open. It's been our best hiring strategy, by the way. This is great. Our programmers look at our software stack. They like it. They send a resume, or worse, or funnier, they don't like it, and they send me a resume because they want to come fix it. And I think that's really great.

And then we've licensed our AI and our RISC-V CPU technology to people, and they like it and they use it and they send us feedback. So we're going to license our AI technology, but also build and sell systems. Fantastic. How many folks here have heard of McKinsey? Everybody, right? How many folks here have heard of Quantum Black? Could you raise your hand?

You need some publicity, Alexander. So Alexander runs Quantum Black, which is a 5,000-person software and engineering team inside of McKinsey on their AI-focused area. Your story has been incredible. Tell us about Quantum Black, please.

Thanks, Peter. I think there are basically two worlds. There is a beautiful and shiny world that is on this stage, and we all enjoy the age of AI, enjoy the evaluations, right, and having a great life. And I think we have quite a confused audience that kind of hears it, but doesn't see any impact in their lives, be it bottom lines, but be it also as human beings, right? And kind of

So the question is how do you reconcile these two? And if you look at the numbers and we take the technology companies aside, the said number in the last five years is 11%. Meaning only one out of ten use cases ever saw the light of production. Everything else is kind of entertainment, we play with it, but it's irrelevant. Now while we're inspired by Gen.AI, the success rate is maybe 7% at the best.

So what we're trying to do in Quantum Black is basically bring these two worlds together, where we're trying to move from 11% into 100%. And what we got today, exactly like you alluded to, is 5,000 people working in 50 countries, trying to transform nations and company. We have five R&D centers that are working on the most precious products for humanity, and roughly 43 products that we deploy globally.

And we clearly work with a lot of colleagues to kind of try to bring their innovation into day-to-day of enterprises around the globe. And kind of indeed, McKinsey is known for producing slides and many other funny things. But also to be fair, it's reinvented the consulting profession already twice. And this is the third attempt and humbled to be here and try to reinvent it. You know, one of the things I say is that, you know, by the end of this decade, there are going to be two kinds of companies.

those that are fully utilizing AI and those that are out of business. Do you agree with that? I agree with it, but I think the problem is slightly different. That today, most of this transformation, and I think everybody here tried to do a digital transformation, right? Or at least declared it to their boards and shareholders. Now, the success rate is not high because what we're trying to do is to infuse technology into by definition broken process, into the old process.

And what I truly believe we are, instead of kind of learning technology, as all the colleagues here on the stage, and stepping back and trying to reinvent something and understand completely different reality that operates on very different cost structure, different social rules, we try to force it, and therefore it fails. So if anything, I truly believe that we are at the end of age of mediocrity.

whatever mediocre, whatever standard, could be kind of played by machine, but actually beginning of the age of creativity, because the notion of how do you create something with technology that is a commodity becomes much more interesting. Prem, back to you. Most important thing that the audience here needs to take away from the work Stability is doing, from your perspective on AI as an enterprise creativity tool,

Speak to us about that. Sure. So probably everybody's seen so much controversy around AI in the entertainment industry. And even the industry went on strike for over a year. And then they've settled and all the guilds have cut deals with the studios. And this is really no different than what really happened in 1927 when the first

when movies went from silent to talkies, what they called before. There was a great controversy at that point. They thought everybody in the Broadway... They thought talking was for Broadway and movies needed to be silent. And obviously that was proven wrong. Color took forever. Color took decades to actually catch on. And then finally in the 60s it did. And now it's unthinkable. It's like an artistic choice, obviously, to be in black and white. And then, of course...

In digital transformation, everybody kind of fought it at first, and they're confusing headwind with tailwind, I think is probably the best way I can summarize it. And of course, when digital kicked off in 2000 instead of film, by 2017, I think like 98% of all films were made on digital. So the lasting statement for the film industry is don't look at AI as headwind, look at it as tailwind.

And do we see stability becoming a creative agent so that every individual can become a creator? Absolutely. And we're going to see it in Avatar 3, 4 and 5?

Well, Avatar 3, Jim's editing now, so I think it's done. And hopefully it'll come out in December. But definitely, I think, in the later Avatars and others, I think you're going to hopefully see a lot of our tools in there. All right. Well, congratulations on the success. Prem came in as a CEO how long ago?

About eight months ago. Eight months ago, and it just revolutionized the company. It had a huge legacy of models and capabilities, but it's really driven it in an extraordinary way. Thank you. Ramin, the world needs another LLM. Why? No. So what the world needs is, as I mean, detailed out very nicely, it's the phase right now that we have to make AI useful. You know, it doesn't matter what runs AI. I mean,

And then we are in this amazing period of time where at every scale AI can bring value. And you put it nicely, like I can see a future where everything is going to be integrated in our society. It's not that we need a different type of LLM. We need to do it right. We have a motto in our offices that every engineer at Liquid AI, we are designing AI

We call it ML done right. Machine learning done right. So that means you don't need to consume a lot of energy to build AI systems. We don't need to use a lot of energy to host AI systems. So what we do, we are basically democratizing kind of access to AI. So if you think about it,

in the cheapest possible way. So hosting a foundation model on a phone or on a device today with Liquid AI costs zero dollars because it doesn't run on a GPU anymore. Everything is on a device. Is every device in my home, in my car, in my office going to be AI enabled? Correct, yes. So what's your world look like when everything is intelligent? Every single device you're touching, talking to, thinking about is intelligent.

I think it's going to be amazing. And the humanoid robots that runs in our homes in the future, they're not going to be connected to the cloud. They're going to have their own kind of local AI banks when they have privacy issues. That'll make it safe so Elon can't actually sort of start the robot revolution. Yes. Jack.

You see the future and you're leading it and it's just, it's 100xing, it's not 10xing. As we start to see, you know, QLMs and as we see quantum computers coming online, is the world ready for how much is going to change in the next five years? What?

I think it's going to be a fascinating next five years. But, Peter, if I can just give maybe two ideas that can share with the audience to help absorb what's about to happen. First is the power of small teams. What I recommend to every one of us and what we're practicing is that armed with LLMs and LQMs, armed with these new AI tools that are from the panelists or from others at this conference and elsewhere...

Small teams can change the world. Small teams. If you're a big company, portion off a team of 10 people and say, you're going to be in this new area, you have this mission, go. So you're moonshot teams. Yeah. And so at Sandbox, thank you. Just to give you one practical example...

There's a big issue when you try to fly a plane now. There's no more GPS. If you try to go to parts of Europe, no more GPS. If you go to anywhere in the Gulf region, Saudi, you're landing in Riyadh or Dammam. If you're landing in Abu Dhabi or Dubai, there is no more GPS. It's being jammed. It's being spoofed. It's out.

In the Indo-Pacific, again, PRC China is blocking GPS. They are an 11-person team, Peter, that we armed with this kind of AI and some quantum sensors, solved the problem. It's flying now on the United States Air Force. And so small teams now are the order of the day. As managers, as leaders, this is what we're doing. This is what I think more people will start to realize. And the second thing I would leave you with, Peter, is that

I know a lot of people are still concerned about AI, concerned about what are the implications of AI. Let me also say that what we're concerned about is people not embracing AI fast enough to solve the big problems of our current society. Let's hit the

the big diseases that have plagued us and challenged us for 40 plus years. Let's bring battery storage to a new level, going beyond lithium ion, going beyond the current chemistries. This is where we need to really focus, more embracing, yes, responsibly, of course,

but making sure we lean in. And I was pleased to see at the Paris AI Summit that we all just came from, that there was a lean-in attitude rather than two years ago when people were like, oh no, should we even touch this stuff? So small teams and let's lean in and let's solve the big problems in society now. Amazing. Let's cheer for that. Nice job, Jack. Thank you. Jim, what do you want people to take away from the work that you're doing? What should they remember? How should they utilize the tech that you're building?

Yeah, so AI doesn't have to be unbelievably expensive, unbelievably big, unbelievably proprietary. Like, that's not required. Like, the computational hardware is fairly straightforward. And we want to make that available to lots of people so they can use it.

I think there is going to be a big up-leveling on how we build and write software and build machines. It shouldn't take two years to build a computer. We want to pull that down. It shouldn't take $10,000 to buy a single chip. We're going to take that down drastically. How much cheaper is the systems you're building compared to the old times? Our target is 5 to 10x cheaper. I'm sorry? 5 to 10x cheaper. 5 to 10x cheaper. Than the current systems. And then we have a roadmap to continue to make that better.

And then the other piece is, like, you have to, I like the big swing approach on using AI, but start small, right? Like, I'm asking my software team to double their productivity this year, and everybody's starting to use the code generators, the code helpers. We're building our own tools to go check the quality, and just start working on it and get used to it.

Because, like you're right, if your system is broken, like patching up the broken system isn't quite right. But getting a real feel for it and using it and then starting to iterate on how your system works is really important.

And I think everybody should, you know, dive in and embrace it. But we don't have to solve world peace first. I would like to make my code have a few less bugs. Alexander, last words from you. Who typically comes as a customer to Quantum Black? And what is your value proposition that you offer them? Is it we're going to understand your problems and solve your problems? No, basically, first of all, and it goes back to your question of what needs to be done, the customer is the chairman or the CEO.

Unless number one in or head of the state is really interested in this problem and really going to invest her or his time, don't waste your time. It's not going to work. We're never going to do it. It's so important, right? Unless you've got buy-in from the very top of your organization to be prepared to make yourself an AI first. One of the biggest challenges a lot of companies have here, you're not competing against your other competitors.

typical companies. You're competing against the startup that is AI native from the beginning. And it starts with your own literacy. Because I think this room, and we grew up under the paradigm that unless I cannot explain you something in two minutes, I'm probably incompetent. And it's all right, but we need to speak the same language. And the first thing is kind of go and study language. And while we could claim that AI failed in many things, what you could clearly see the drop of AI during summer

And you ask yourself why? Because all the kids are out of school or university, right? Because AI was the best tutor to this world. So first of all, use AI to educate yourself. That's number one. Number two, why I'm saying it's a leadership challenge, because what you need to do is really to go fully in to transform the enterprise. You need to get the data right. That never is right.

You need to change your architecture. That essentially means changing politics within the organization. And we don't like to change politics. Then hopefully you need to hire good people, but then teach the rest of the body of the church what the hell is it and how to ensure you embrace it. And then all of a sudden you have a team with human beings and certain agents and kind of how do you operate it yourself. Now you multiply the likelihood of these things and likelihood to success. Unless you believe and go fully in,

It's risky because you put your career, you put your company future on the line, and you cannot go small. You need to go big to succeed. And that's what we're trying to do. Just use AI as the way to make the world into a better place. Ladies and gentlemen, let's give it up for this incredible panel. Thank you all so much. Thank you.