Hello everyone and welcome to this edition of talking to experts. Today, we have a special guest, a friend, and a colleague we have known for 20 years in the States, France and Europe. So, it is my great pleasure to welcome Jack Hidary, CEO of SandboxAQ Hi Jack.
Hi Francois. Thank you for your invitation. Jack and I we met we've met I think 18 years ago? Oh it's been a while! And and Jack and I we were friends and colleagues and he did a fantastic job uh for us in the US bringing new ideas lot of energy so it's my great pleasure to welcome my friend and colleague Jack the CEO of SandboxAQ so before we go into artificial intelligent Quantum tell us a little bit about yourself well first of all it's great to be here and uh to be here in Paris we're doing so much now in France uh across the whole country I think we'll get a chance to talk about all the activities we have here but for myself I started very interested in how the world works I was interested first in how the brain works I studied neuroscience and at the same time I was also interested in how the underlying architecture of the universe works and that's physics and uh these passions of physics Neuroscience came together and as that was happening the computer Revolution was emerging uh before even we knew about the web we started programming we started coding and I went to computer Camp as a kid and so I was a very big geek as a kid learning how to code and it was very exciting where you could see your own work active on the screen and so these three different strains of my life coding and Neuroscience AI and physics my dream was always to bring these together and to solve some of the world's biggest problems the problems of drug discovery of bringing a new drug to Market uh the problems of new clean tech Innovation uh the problems of new materials for our world uh the problems of using AI in terms of analyzing you know what trends would be in the future these are all issues that we need all these three different kinds of disciplines to come together so I was always interested, Francois to be a cross-disciplinary individual I was never one wanting to pigeon hole in one place you could be French because you know in France we are famous to have horizontal thinkings aggregating VAR things so tell us about the phone call you received seven years ago from two famous uh entrepreneur well seven years ago I had created a few companies I was focused on clean energy I was focused on some philanthropy and uh I got word from Google that uh they were starting something new something called alphabet and they were looking for individuals to advise and start companies and start new divisions within this new entity and the idea of alphabet of course is that while Google focuses on search and organizing the world's information there's a lot of other things that Computing can do uh self-driving cars already was an idea inside of Google and they decided to launch alphabet as a way to house these new ideas uh self-driving cars ended up being the first spin that's wh and of course one of the next ideas was Healthcare information uh my friend Andy Conrad started that uh inside of X initially and then spun out as verily so a lot of exciting ideas and when I joined about seven years ago uh the idea really that we started to talk about with uh Sergey myself Astro many others started to join in the conversation is how could Advanced Computing be it AI be it quantum how could that change the game when it comes to society's biggest challenges so Jack tell us a little bit how the spin-off occurred so we were building and building uh different kinds of Technologies and realizing that uh the world was going to change the GPU in particular this Graphics Processing Unit GPU alphabet was launching its own gpus called the TPU tensor Processing Unit Nvidia of course a very strong company now one of the leaders in the world creating gpus and many other companies as well Amazon and now Microsoft and many others ALS also creating these specialized chips these chips that would accelerate certain kinds of operations initially for G for graphics then people realized of course that this was the same kind of operations that we needed for deep learning for advanced kind of AI today we all hear about large language models different kinds of deep learning the graphics of uh Dolly 2 Dolly 3 now just came out so it's a very exciting moment and gpus are core of that but what we realize in SandboxAQ is that these gpus can do even more they can do something very very special and that's called simulation and so AI very very important when you have a lot of data you could train a deep Learning Network a network that's inspired by Neuroscience by the brains distributed architecture and so you can train that in the way we train Toddlers and children they look at many examples they learn from examples but in addition there's another very powerful tool a tool franois simulation that would become equal in the global lexicon very soon in the next few years to the word llm or gender of AI and that word is simulation simulation means that we don't have a lot of data we want to create a new drug a new molecule a new kind of battery chemistry we don't have any past data because we're creating something novel and so to do that we have to use the first principles of physics itself we have to say will this electron from this molecule bind to this electron of this molecule in the body will this ion and this ionic structure in a battery chemistry be something that has good energy density so these are questions that are quantum questions and now the gpu's got fast enough we can actually do some of the simulation on the GPU itself even in advance of the Advent of scaled quantum computer so with this realization we decided it was time to spin out and about 18 months ago we did spin out as sandbox AQ with Eric Schmidt as chairman myself as CEO and a Wonder group a cross disciplinary group fris of physicists and chemists and AI Engineers really a wonderful group and the company is really grown since then particularly because in December of last year we closed on a round of $500 million that's from tro price from gugenheim from wonderful leaders like Eric and others and that Capital allowed us to become I don't say Global I say multilocal a company that has its roots not only in the US but also Al in Canada and here in France and Spain and England and many other countries as well serving those customers and those communities to with Advanced Computing so it's been a very exciting year and a half fris so Jack four months after the spin-off you wor a company in France tell us about that well as we spun off and raised a large amount of capital we started to hire a lot of people and wonderful people we also look though for key groups of talent around the world and again we took a multi local view of the world where we want to have deep roots in many parts of the world so that we can serve those customers and also engage with the wonderful talent in those countries we heard about crypto sense from several customers of ours multiple Banks who are customers of our told us hey SandboxAQ we're going to use your software it's great for encryption management but also we're going to use crypto sets and so it was a wonderful team that spun out of inria the research uh organization here in France uh a group mainly of phds uh and based in Paris but with people around France and we started having conversations with them but maybe some kind of cooperation partnership and quickly within a few weeks we all realized this would be a great acquisition for SandboxAQ and so we did acquire the company uh we had to go through some uh engagement with the French government because of course it is National Security and we successfully passed this uh level of Engagement and so now we have a full subsidiary and a full here in France we're very proud to say yes that we have very good footprint in fact more than 15% of our employees worldwide are here in France and we're growing that that population France has an additional attraction of the universities in France um the poly technique uh NS many many uh wonderful universities across France that we draw wonderful Talent from and we keep that Talent right here in France uh the idea is really people should work where they Thrive and they're thriving here they should work here as well so it's a very exciting Time For Us in France so you're talking about deep Tech I'm very interested in this world because we talk about this more and more how would you uh Define uh deep Tech versus tra traditional VC based VC backed companies it's a good question fris so when we look at the history of Silicon Valley particularly the last say 20 years we saw a lot of investment in productivity tools where you can look at slack all right or jro or different kinds of group productivity tools you can look at other VCB companies in social media as an example but deep Tech says something different deep Tech says we could also call it Frontier Tech Frontier Tech deep Tech says let's fundamentally have a Leap Forward in our society in the cars that we drive in perhaps even flying cars uh in how we go to space how can we reduce the uh cost it takes to get to space Not only to bring Crews up there and humans up there but also for satellites and we see the the big success of course of starlink right now that's a deep Tech uh Innovation when we look for example at the world of medicine that's a key area for deep Tech Innovation both in terms of new Pharmaceuticals but also of course Diagnostics themselves and I mean just Diagnostics in a hospital setting part of the deep Tech Revolution is to bring Diagnostics to the home and bring it to a ubiquitous setting as well so when we look at now deep Tech as an asset class what we're finding with investors investors like tro price like Guggenheim investors like BR Jim Brier of Brier Capital Mark benof the founder of Salesforce Thomas Tull all these are investors of ours in sandbox AQ and we're proud to have them because they recognize what we recognize which is that deep Tech actually is less risky in investment because when you make these Investments you're creating a deep moat around this technology a deep moat of intellectual property of expertise of connection with customer you're making a very very deep differentiation that's very hard to commoditize and it's also lends itself to platform type companies what do I mean by platform companies I mean companies that have a core technology that you can build many different products on that's a platform company and that company is going to be far more valuable and more impactful than just a single feature company that might be a good acquisition for some of the big tech companies but uh is very different than Frontier technology so while Frontier technology maybe 20 years ago Fran may have seemed like uh you know wow that's way out there now the largest asset managers in the world tro and many others the largest family offices in the world Mark benof Eric Schmidt Jim Brer many others are recognizing this as its own asset class with a particularly favorable risk reward ratio because of the connectivity across these different parts of the platform and the ability to defend that platform and build on it for impact in society most of our investors franois are here in our company not just for financial return but for positive impact and that positive impact is very much there with Frontier technology so Jack we're talking a lot about AI consum not a lot of people knows you know what those acronym means but I think AI starts with a touring thing uh in 1953 aristot talk about Quantum you know easy own wordss and with Einstein and the quantum at the beginning of the 20th century uh could you give us in your view a simple definition of quantum a simple definition of the ey why sandbox aggregating Ai and Quantum Quantum is ahead of the game yeah it's a great question so you know we there's a lot of buzzwords out there and we throw them all around and so let's let's define some of these a great great idea AI let's start with AI artificial intelligence you know that's a name that came up and was coined in 1955 because uh a number of people John McCarthy and others were applying for a grant from the Ford Foundation of all places and they said we want to have a summer uh up at Dartmouth and we want to really focus on this new kind of Technology uh at that time actually it was not called AI it was called cybernetics and Norbert weiner was the one who coined that particular term uh but John wanted a new term so they called it artificial intelligence we're kind of a little stuck with that now we could have called it augmented intelligence we could have called it synthetic intelligence uh there's many other possible names and so maybe we should still think about some new names for it but today we focused on the term AI so what is exactly we're referring to well in today's context in the context of now an understanding of how the brain works when we looked at the brain and we opened up the brain over the last 50 60 years what people found is not a CPU not a central process a unit and not like distinct memory areas what they found actually is a distributed set of neurons the average human brain has between 86 and 100 billion neurons and 100 trillion to a thousand trillion connections or synapses we call them that architecture inspired the idea of these kinds of artificial neuron networks these biological neuron networks translated to artificial neurons and artificial networks and we train them similarly to how again a child might might learn we present many examples and we have the network learn those examples initially it's not going to be very good at doing good prediction and classification but over time with feedback you can make it very good and we had the move to deep learning the idea that we can have not just an input layer an output layer and one layer in between that's called The Hidden layer but actually we can have thousands and maybe even millions of internal layers and that's the Advent of deep learning that came with the GPU this chip that we talked about so AI basically is neuroscience inspired architectures it's a different way of coding it's not coding with if this then that if something has four wheels then maybe it's a car we actually just present the cars in the buses to the neural network and then it learns the nature of what it means to be a car what it means to be a bus we don't actually specify the rules in there and so that became a very powerful architecture that now we have have the llms but llms are just the beginning friend well just the tip of the iceberg in terms of where we're going with AI but now let's turn to Quantum Quantum says that the world that we know about the world of billiard balls of sending Rockets to space this is the world of Isaac Newton this is the world that 300 years ago Isaac Newton gave us the sense of how things will move how things will row down Hills if we drop you know a feather and a big bowling ball what will happen this is all the world of ISAC n but when we get to the microscopic when we get to the world of atoms and electrons um we have to have a new kind of way of describing these in fact in 1899 a crisis occurred because the electron was discovered JJ Thompson got the Nobel Prize because they discover he discovered the electron and obviously then the atom was not the smallest thing and it turns out electrons do not follow the laws of Isaac Newton and we need a new set of laws and so Mox pling and Albert Einstein and Neils Bor and Schrodinger and dck and many others came up with a set of laws and rules that says this is actually the way the world Works fundamentally and that's quantum physics and so Quantum is a language it's the language that molecules use when they speak to each other when they connect to each other that's the world of quantum now you asked about putting these two together so when we think about putting these two together that's a very powerful combination of computing on the one we have ai that's very good when you have a lot of data and we could train on that data and so we give a lot of the data if we have for example a million patient records and we say what happens if somebody enters the hospital with chest pain what are the likely outcomes we can train an AI on that and we can make predictions because we have a million points of data but what if we want to have a new molecule a new device uh new material we don't have any data it's novel and that's when we have to use physics itself to figure out and make new data we call that synthetic data data that we create ourselves and that data we can then apply AI to so the next wave of AI is beyond llms beyond the graphics tools and Image Creators and video creators is this fundamental deep Tech revolution of the combination of the best of AI and the best of quantum and these two together are very powerful to help design a a new world that meets many of our challenges in society that's fascinating coming back to soundbox where do you see the application of this technology and what are the lwh hanging fruit you are going to focus on right now well good question so I think there's a few areas of lwh hanging fruit immediately first is simulation optimization so here again we we have the opportunity to go after new drugs again we're not a drug maker ourselves at sandbox AQ but uh we can help drug makers we can help pharmaceutical companies and Drive their productivity and drive their acceleration right now it it takes on average 13 years to develop One Drug it takes on average two and a half billion Euro to develop one drug and even after all that time and money 80% failure rate in the clinical trials fris 80% failure that's a lot of money loss and a lot of patients who go without a medicine so this is now a fundamental change in how we can get things done Ai and Quantum coming together can say okay biotech company okay pharmaceutical company you have an idea for a New Drug we can run it through millions of simulations way before it ever gets to the first human and that is a very very exciting so the time to Market time to Market is is shrunk money goes down the capital goes down and that means we can go after more conditions that means that rare diseases that means that yeah right extending life and and Longevity is another great area so that's one lwh hanging fruit area uh clean tech is another wonderful area where we want to have new kinds of solar panels solar panels have come down in price but there's another way to bring it down even another 90% And that's called parav skites a chemistry that is really gaining good momentum in terms of its efficiency its level of converting sunlight to electricity but it's not stable on a roof for more than about a year or two and so using Ai and Quantum we can change that game we can actually make it last hopefully for many more years so be it clean tech be it biotech there's a lot of different things that we can do there so that's one area of SN but now call it simulation optimization but now let's talk about cyber security as much as we love Ai and Quantum and it gives us many positive things let's also recognize fris that there's attacks on cyber security both from Ai and Quantum and what do we mean by that well if I have password files in a large company and the adversary the hacker has AI they can do analysis of those password files and find patterns to maybe find a default password that was issued by the it organization that gives them an in into the company that's using AI as an attack tool Quantum also could be used over the next number of years and it is a tack tool because the fundamental nature of cyber security of how we communicate over the Internet of how we exchange credit cards How We Do Bank transactions the global economy is built on public key encryption public key cryptography the way that we can exchange information over large distances over the public internet Quantum as it scales into larger computers that are fault tolerant breaks this cryptography and so both Ai and Quantum very positive Technologies but also means that we must upgrade our cyber security and so that's the second area that we must address with AI and Quantum we must have better cybercity tools particularly in encryption tell us about the third pillar which is sensing you know and how Quantum will uh change dramatically the way we sense we acquire informations well sensing the world around us is critical and so be it medical Diagnostics or be it other kinds of sensing this is fundamental to modern technology and so when we look for example at the challenge of GPS jamming so all of us now are addicted to GPS imagine getting into a car and not GPS a map available Google Maps Apple Maps any of the maps out there um most people would literally be lost literally be lost but in fact actually we now see that more and more we see that adversaries around the world franois are jamming GPS it's actually not that hard to jam GPS it's a known publish frequency so of course we can jam it people can jam it uh and so be it in airplanes uh recently uh two passenger Jets full of passengers from Quant Airlines on March 14th of this year the GPS was jammed uh they lost GPS for hours and hours uh more recently just a month ago quite a number of planes in the Middle East were jammed uh they lost GPS in fact it was even worse then Francois in that particular case some were GPS jammed some were GPS spoofed where a wrong signal of GPS was pushed into the airplane traveling at at uh you know 8900 kilometers an hour that is a very dangerous uh scenario and so we have to have a way of navigating around the world without GPS in fact that's what Quantum sensing and AI together can give us we can have a Quantum sensor that senses the magnetic field of the earth that magnetic field is unique on every spot across the Earth and then combined with AI to pull the signal from that sensor we need the AI there we can actually have a device and have built a device that is now flying on quite a number of government and private planes that actually can navigate without GPS also for for heart disease well exactly the same technology I just mentioned for navigation can also be used to improve our cardiac Diagnostics heart disease fruis as you all know is the number one killer in many countries in the world and so how do we improve outcomes for heart disease number one is Diagnostics right now we have ECG we have ultrasound there's a variety of initial Diagnostics when someone has chest pain but they don't give us enough information and in fact in most cases in hospitals around the world you have to wait for Two Plus hours to get a particular blood test to see is the heart did it have some muscle breakdown right some some attack on the heart and in fact what can we do in the first 5 10 minutes we can do not an ECG electrocardiography with the new advent of what we've developed with AI and Quantum sensors together we can have MCG Magneto cardiography this detects not the elect field of the heart but the magnetic field of the heart and this is going to be a breakthrough for cardiac Diagnostics absolutely and proactive medicine is one of the the key we want to go to preventative rather than just reactive exactly um Jack to finish we talk about lots of our gen AI with' all test that what has your view is it A revolution is it an evolution well gen of AI I think is a very powerful tool particularly for certain kinds of applications if you have a Content application and you want to generate a lot of content let's say you're a bank and you want to create a lot of research reports gen I think could create the first draft of that report and do 70% of the work let's say with the human coming in to finish off the 30% and polish it I know a number of screenwriters and Hollywood that are using gen right now to become much more productive writers and they're writing for multiple shows at the same time so I think gen is a very powerful tool that has many interesting applications but FR really it's just the beginning of applications of AI when we think about all the areas in terms of applying AI to many other fields Beyond just content generation this is where other forms of AI will really start Rising we are just at the beginning of this phase of the AI revolution in particular by combining AI with simulation AI with Quantum this is going to be a very powerful combination of both AI tools that can train on past data and simulation tools tools that can drive to the Future this combination is going to really revolutionize the whole world in fact Quantum will bring new signals very weak signals and AI will bring correlation between those signals to move forward the research correct absolutely correct Jack last question what is your ambition and visual for sandbox because we are I'm very excited when I listen to you where do you see yourself in three for five years from now well sandbox AQ has already grown very very significantly I'm very proud of all of our teams fris our teams have really excelled Beyond any Milestone uh that we internally Set uh be it in simulation uh for drug Discovery and Battery Technology in other areas be it in cyber security to protect the world's data a very key mission for us uh or be it in Quantum sensing our mission at sandbox AQ is to be the most impactful company in the world and we Define impact in terms of the positive effect we can have on all these different areas is using Frontier Computing using Frontier technology key to that is our teams so the first answer I would give to you about where do we see ourselves in three to five years we see ourselves building on The A Team that we have already hired and a players hire other a players and so we we depend on the wonderful incredible teams that we have to bring in additional people internal to the company but also fris key to the next 3 to five years for SBX AQ is partnership we are already having a number of Partners now around the world here in France in the UK in the US in Canada and we're going to have more and more partners because we cannot do this ourselves we have to partner with large companies and startups and Nos and many others and governments in many cases to make sure that we can have the impact we want to on these critical areas Jack thank you for sharing your passion Vision um I get I've got a great IDE for you for Christmas yes Jack Quantum Computing and applied resource it's a very light read FR very light yeah I I just read 10 page each because after I need to understand exactly so thank you so much Jack for joining us and congratulation what what you've accomplished in close to year and half thank you for well
2024-05-01