Running LLMs Locally – Tether CEO Paolo Ardoino Reveals Decentralized AI Platform QVAC

GM, GM. Welcome to what's going to be another fantastic episode of DevNTell. So if you didn't know, DevNTell is a 30-minute podcast held every week, allowing founders, hackers, and anyone in between the opportunity to come showcase what they've built. And today, I'm excited to welcome Paolo Ardoino, who is the CEO of Tether. So if you didn't know, Tether has come up with a new decentralized AI platform called QVAC that they are using to enable an entirely new paradigm where intelligence runs privately, locally, and without permission on any device. So if you stick around for today's episode, you'll get to meet Paolo, learn all about QVAC, learn about a new medical AI advancement that they just announced today, and more. All right, let's get into it.

GM, GM. Welcome to the show, Paolo. I'm so excited to have you on today, man. For those in the crypto space, the Web3 space, probably don't need an introduction, but in case there are some new listeners delving into the world of blockchain, Web3, would you just like to give an introduction about yourself?

Sure. I'm Paolo Ardoino, I'm the CEO at Tether. Tether is mostly known as the company that created the first stablecoin in 2014, USDT. In the last 12 years of history, we participated in all the different cycles of growth of the crypto world, but today USDT and our main product has exceeded all expectations and became the digital dollar for all the emerging markets, developing countries. It is used by 570 million people around the world, has been probably the biggest financial inclusion success story in the history of humanity. So, very excited about that. And definitely USDT taught us a lot in terms of how to use technology to liberate people, to create freedom, to give access to the people that never had access before. In the case of USDT, there are hundreds of millions of people, billions of people that don't have access to basic financial services, and so at Tether, we thought how to use the learnings that we had from USDT and the stablecoins world to also help the billions of people that will not have access to good but basic AI tools in the future. How we can serve them through our network, how we can make sure that there will be an intelligence inclusion, not just financial inclusion. So that's the reason behind QVAC to start with.

Amazing, amazing. And yeah, I feel like us going down the Tether USDT rabbit hole would be a whole episode on its own, but yeah, this one I'm quite excited to learn more about kind of the reasoning why you guys chose to get into the AI space. But kind of just before we get into that, just curious to learn a little bit more about you. Like, how did you find yourself in the world of technology? How did you get into crypto? What was the origin story behind that?

So I've been a geek, nerd, hacker since I was 8. Well, I started coding when I was 8 years old. My father and mother, they were busy sustaining the family and working very hard. I was very lucky because my father brought home a computer and I had a lot of time to play. And then the games were very expensive; I come from a very small town in Italy and we didn't have much access to anything else. So I wanted to create my own games and so I wanted to learn coding, and so I started that. I loved it immediately. For me, many of my friends were good in arts and drawing, I was bad at everything, and so coding was a way for me to create my own universe and invite people inside that universe. And so throughout the following years, I started getting internet and I was playing with... I started using Linux very early, like, well, not early early for me, it was 1995. And then start, you know, learning about hacking. I wrote some articles on E-zines, there was the hacking scene in Italy. I wrote some articles on RSA and large numbers factorization and some math already at a young age, very passionate in cryptography. And then went to the university, then became a researcher at the university. I learned and studied mostly parallel computing and distributed platforms. And loved BitTorrent and what BitTorrent meant for the whole peer-to-peer space. I then worked and I think I led a very important project; I had amazing core researchers and we designed systems, communication systems, highly resilient communication systems for battlefields. When you think about... and I think that thing shaped my mind. Whenever I still today design anything in technology, I always think how technology should be designed to support humans in their worst moments, not in their best moments. I mean, it's easy to design technology that works perfectly when everything is fine, but technology should be the thing that helps us humans in the moments where everything goes to hell basically. And so when we were designing these solutions and the communication networks for battlefields or distress situations, I was thinking about fathers that needed to communicate to their people, to their colleagues, when they were even risking their lives, right? So that makes you think more, that makes you sharper, there is much more at stake. And so we designed amazing protocols, and then I carried that experience throughout all my career. And I keep thinking about how again technology should resist to the wrath of God if it needs to, is truly designed to help humanity and society. And so yeah, I love coding, I love hacking, and learning about things and learning how things are built.

Yeah, I mean, this is the coder way, this is the hacker way. There's no better way to learn than by doing just that. And it sounds like a lot of this early experience and when you mention technology should be built for the worst of times essentially to get us through that. I think this is a nice segue into why Tether decided to go in on AI now. What was the pivoting moment?

Well, I'm a computer science, I'm a sci-fi enthusiast. My preferred author is called Isaac Asimov. He wrote hundreds of books. There are two books that are amazing; one is the cycle of the Foundation, and the other one is a short story called 'The Last Question.' 'The Last Question' is like 14 pages, probably the most beautiful story in the world for me. In 14 pages it includes philosophy, science, physics, AI, religion and a lot more. And it's the story of humanity that from the 1900s and onwards every... from the 1990s kept building better computers and better artificial intelligence. And every hundred of years the technology was better, the AI was better, then every 1,000 years it was becoming even better. And then millions of years pass and billions of years pass and this AI becomes always, always better, it's incredible. At the end, like after billions of years, this AI is part of the fabric of the universe. And throughout the entire history of this story, every single time, like after hundreds of years, after thousands of years, every single time the AI was becoming better, humans were asking one single question to this AI. That was the most complex question of the universe, and it is 'how entropy may be reversed.' How we can basically... it means how we can stop the universe from dying. And then I don't spoil it at the end of the story, but I thought what does it mean for us? And what does it mean for humans that... we are part of the universe but we only know Earth. And so I thought well, how we can transpose that question to Earth means how we can stop society from dying? How we can make society stable? And Tether happened to build a stablecoin, and that is a digital dollar. What we saw and we analyzed in a very scientific way what USDT did and obtained from a sociological point of view. If you look at the data, you see that USDT was able to bring stability to society. Society of all the emerging markets and the developing countries. You know, people that live in Turkey, the Turkish Lira lost 80% of its value against the US dollar in the last five years. The Argentinian Peso 94% of its value against the US dollar in the last five years. So not entering too much in USDT, but those populations living in those emerging markets could not be stable, could not be building a better society, a stronger society, society that was able to innovate and build and grow and study and research more because if your national currency is basically destroyed every single year, the only thing you can try to do is survive. So then there are, not many people know, but there are 4 billion people that are affected by this, and even more, probably around 5, that live in countries with super high inflation. So if we have basically more than half of the population of the world that is kept outside of the traditional financial system or does not have access to basic financial services in a stable currency, then how society can remain stable? So going back to my point, how we can stop society from dying if half of society is actually trying just to survive because of the currency that they hold is dying? And everything is priced in currency, right? So how we can fix that? And so USDT is the first example of something that truly can change the world. But it's not enough, was not enough. So now we know that you have half of the population of the world that is part of the wealthy population, maybe not wealthy but still has access to good financial services, and half does not. And so there is a gap between these two halves of the population. And then you add on top of that AI. So what if AI would increase that gap? Because if half of the population of the world that has access to financial services plus has access to AI because they can pay for Anthropic or OpenAI subscription, this part of the population, this half of the population will become 100 times smarter. And on the other side, you have the half of the population that does not have access to basic financial services and cannot afford to pay for an Anthropic or OpenAI subscription. The gap will become not twice bigger, but will become 100 times bigger because as people need stable currency, they need in my opinion stable intelligence. They need to be able to have access to basic intelligent tools that will help them throughout their life, that support them and they will make them part of this digital revolution. But the problem is that if you have a person that lives in Haiti where the average salary per day is $1.35, in many countries in Africa you have the average salary of a person is $120 per month or $80 per month. How these people can afford to pay for a subscription to get a basic AI service? That is not sustainable. And so we thought that with Tether we have already a huge distribution channel into those populations, into that part of society. What if we were to reuse that distribution channel to build open-source, local-first, on-device AI tools that would not cost a fortune to Tether to run but would be good enough to solve 80 to 90% of the use cases of AI of that part of the population? So many people in Africa or in Central/South America and so on, they will not use AI to try to find the cure of cancer... it's a most amazing thing that AI can do, right? So it's very important the AI usage for solving the very complex problems of the world. That's amazing. But that's not what 99.9% of the people in this world will try to do. Most of the people will need simple things that will be very helpful to their day-to-day lives. Most of the people in the world have very important problems that are very local to themselves, and if they use AI, they will have an help, they will have a lifeline like using USDT will offer them a lifeline. And so why not using the learnings of USDT to build tools that we can offer to these populations that provide education, that provide basic health answers? We know that Tether is working... well today we released a model called MedPSI that is even beating Med-Gemini, that is Google's model, and our 4-billion parameters model is beating 27-billion parameters model from Google. And can run fine on a smartphone, can scale on any small device, even an average smartphone. So that is an incredible testament of what our team could do. But on the other side, we are working on better translation models for African languages. In Africa there are many languages and dialects, so we are trying to put our resources to create the most amazing open-source translation model for African languages so that suddenly we can offer free education to all the villages in Africa that we are already servicing with USDT. We are going to offer them free education so that all the kids in those villages can learn and can study and can be part of an amazing society. And so these are the things that honestly don't cost even that much, but you need to have the right incentive to build. The right incentive to me and for Tether is stability of society.

This is amazing to hear, and yeah, there's a lot to kind of unravel under the hood of QVAC. But as this is a developer-oriented show, perhaps you can speak to some of the tools and the APIs and SDKs that are available for developers watching today who want to start building with this and really unleash the compute power of Edge AI.

Absolutely. So let's start from QVAC SDK, right? So first of all, we didn't want just to build an AI platform on cloud. There's plenty, I don't think it's that exciting to us. So we wanted to offer an SDK that would allow any developer to easily code in, inside their applications, existing applications or new applications, AI functionalities. AI functionalities that would vary from OCR to standard chat LLM to text-to-speech, speech-to-text and so on. And so we wanted to wrap everything, all these functionalities and more into one single SDK with one single interface with one single coding style. So behind the scenes, we took Llama.cpp, that is an open-source inference engine for large language models, and we forked it, still maintaining it open-source, and we called it QVAC Fabric. So Fabric is an enhanced version of Llama.cpp that over time got a lot of additional support, including for example the ability to run the BitNet, Microsoft BitNet models, you know, the 1-bit models that Microsoft pushed. Rather than the version of Microsoft that was only basically supporting Nvidia GPUs and the very high-end GPUs, we made it, we adjusted it so that now the BitNet models can be run on consumer hardware, on the Snapdragon GPUs of you know the ones you find on Samsung phones or on iOS. And so on top of that, we also allowed Fabric LLM to support a general fine-tuning layer. So not only the BitNet models but all the models that are supported by the Fabric LLM, that is basically all the Llama.cpp models, they can now be fine-tuned directly on local devices through one single framework. So we proved that with the Fabric LLM we could fine-tune 1 and 3-billion models directly on an iPhone 16. So this is truly unprecedented and over time... and imagine on iPhone 16 was possible, but then you have your nice Mac or a good laptop, you can truly fine-tune models locally with your own content without losing any privacy. So basically QVAC SDK is a layer on top of Fabric LLM. Fabric LLM is a C++ engine, it's a fork of Llama.cpp but includes much more than Llama.cpp and offers a unified API layer to do all the classic tasks that you want to delegate to AI again: text, chat, text-to-speech, speech-to-text, OCR, and any sort of recognition, image generation... we support now image generation with diffusion. And so there was a huge reception from the community and we started seeing some traction from the open community to develop on top of QVAC SDK. We are going soon to run some hackathons to prove the quality of this tool. And so suddenly now every single... by the way, this QVAC SDK can be used on... has a JavaScript interface so that any web developer can suddenly start integrating... so web developers are the most common type of developers and so they can use basically QVAC SDK and build any logic inside any app in a very simple way. And this is our way to empower every developer to build AI tools by themselves, maintaining the privacy, but also reducing the latency at the minimum, right? So imagine a future where we are surrounded by billions of robots. I want a robot to be able to use a local model and local vision model and understand what is going on around itself and take a decision without waiting for the latency of going to a data center and back. You know, imagine like a smart car; a car should break immediately after seeing what happened directly on the road rather than having to wait for a data center answer. And so all that, the world is moving towards local AI, and we believe that every single developer should be empowered to control their own local AIs and every single user will benefit from having a local, private AI.

Yeah, I also think that this is kind of where the community and this AI scene as a whole is going. Of course, you have the RAM crunch, people trying to figure out what's the best setup to run their local AI. And but like since we're on this topic, I mean, we see every week there's a new model coming out, whether it's from Anthropic, OpenAI, or an open-source model from Qwen, DeepSeek, whomever. From your perspective, being so enshrined in this space now, do you think we'll get to a point anytime soon where the gap will close between the capabilities of these frontier models and what you can run on consumer-level hardware?

First of all, I think the gap will become smaller and smaller, but also I want to... I think it's important to split different segments of population. Of course we are all geeks here, right? So we want always to try the last thing. But the reality is that for probably 95% of the people, even six-month-old models are perfectly fine. They are giving them very good answers, they are very efficient and so on. So again, unless you are a very good coder and you want always to have the last frontier models because you are building very complex things, that's great. But probably you are representing 1% of the population, while 90% of the population, like my mom, would use an application on her phone to scan different grocery bills and have a summary for the end of the month, or trying to translate in multiple languages or transcribe something. Very simple type of tasks that six-month-old models are completely capable to serve those capabilities. And so the more AI will be part of our world, the more even waiting three months, six months will not matter for most of the people. And of course again, if you are doing very complex things and very important things, like trying to find the cure of a new disease, great, you need a lot of GPUs, you need huge data centers, you need to consume a lot of energy, but because the outcome, the result will be important for the entire humanity. But for again, for 95% of the tasks that 90% of the people in this world are doing, normal models and local models on smartphones are more than enough.

Yeah, exactly. Like even the smaller parameterized models, they're becoming more and more smarter. The models are only as smart as the data they're trained on and the context you provide them. So yeah, definitely a lot of the day-to-day tasks of normal people should be satisfied with even today's local model scene. As we're kind of running short on time here, I wanted to throw a couple more questions your way. So obviously you guys have made a name for yourself in the stablecoin space and now you're making strides in the AI space. I guess from your perspective, how do you see these two worlds kind of connecting? How do the dots connect in your kind of North Star vision of where Tether is going with both these verticals?

So I think the future will see AI agents using stablecoins as a payment layer. So even in... if we believe in a future where you have 10 billion humans, 10 billion machines or robots, and a trillion of AI agents, these AI agents will not have a Morgan Stanley account. They will not probably use PayPal. They will need programmable money because agents are all about programmability, right? They are all about executing complex actions. They will want programmable money, and so USDT is the best form of programmable money that is able to scale at the very high transactions per second available in this moment in time and it will continue to grow. I mean agents using USDT can touch already hundreds of millions of people and interact with hundreds of millions of people and hundreds of millions of services. So we'll continue in that direction, I think that will be the merger of the two worlds.

Yeah, I see that same future, and it's just a matter of time before they converge, so we'll play the waiting game so to speak. And for our aspiring founders who are watching the show today, who might be on the fence of perhaps starting their own company or... I mean no better time to start an AI company than now. Is there any sage advice or anything you'd want to tell them to kind of get them to start?

Yeah. You know what Tether made very well, what was the success of Tether? In 2014, Tether was basically a fintech company, we created a new digital dollar. But we didn't try to compete with everyone else. You know, in 2014 if you were building a fintech company everyone was trying to go in the Silicon Valley and fight against each other. Well, we went to every single person that didn't want to, or was not served by that type of establishment. So if you build... the bottom line here is that if you build a new company today, don't try to compete in overcrowded markets. Go to everyone else. And we are living proof that if there are 4 billion people that are not served by the traditional financial system, they are a good enough user base, 4 billion people, imagine that. And you know, today Tether is very successful even not serving the full 4 billion people, we're serving 570 million people. So find your user base, understand their problems, and build something that truly serves them.

Beautifully said, yeah. I totally agree. And one last question for you. How can people keep up-to-date with all the great progress you guys are making with QVAC, Tether, and anything new that might come up?

Go on tether.io or follow @paoloardoino on Twitter or @tether on Twitter, well, X now. All the habits.

All the places. And yes, we will have all those resources for you in the description of the podcast below. And with that, Paolo, I want to thank you so much for taking the time out of your very busy day to come chat with us today. It was amazing to meet you and learn more about QVAC and all the great stuff you guys are doing trying to progress Edge AI.

Thank you very much, Narb. It was a big pleasure. Have a good day.

You too. All right gang, hope everybody has a very fantastic Thursday, Friday, weekend ahead and we will catch you back here next week for another great episode of DevNTell. Till then, have a good one folks. Cheers.