MemComputing Demonstrates Brainlike Tech That Can Crack Encryption in Seconds
hello everyone and welcome to the Future inter review podcast where we talk with leaders in Tech science and business about the future of technology and its impact on society and the global economy I'm your host Barrett Anderson the COO of future inter review which The Economist has called the best technology conference in the world and I'm here today with John Bean who's the CEO of M Computing and Fab Fabio Lorenzo traversa who is the co-founder and CTO of M Computing both of whom are joining us at future in revieww in Los Angeles just a few in just a few weeks to talk more about me Computing but John Fabio welcome it's so great to have you here with us you as well thank you so much thank you so much uh so we're talking today because of a special demo that you did using your technology um meem Computing and how that demonstrated the potential to break 2048 encryption Technologies in less than a second um for those who are watching this this podcast who don't know anything about encryption can you explain to them why that is important what is the so what of this of this demo that we're about to talk about so I'll I'll jump in quick here Fabio so are the internet uses something called RSA 2048 to uh as encryption for the keys so it has a public and private key that it exchanges and the private key uh is one that we don't know about right so it so to in order to break encryption so you could break into the internet or or Communications or what have you you would have to be able to crack that private key the way the private key is made up of two large prime numbers and uh the the way that you could break the key is called prime factorization basically you you identify the the two prime numbers now with current technology ology it's expected to take 400 trillion years to break this so that's why it's considered extremely safe however uh you know there are different means that are trying to go after and break that Quantum Computing hopes to do it one day but with me Computing uh a new technology new computer architecture that was invented by Dr traversa here at UC San Diego we demonstrated that we can uh at scale we can actually break the say 2048 in uh um in subse time now that's we're not at the scale today but it's just about you know two years away that we would we'd be able to do that with with funding right right right so Fabio can you tell me more a little bit more about me Computing what is it how does it work and what is it about that architecture that allowed you to make this kind of breakthrough where others have not been able to Sure uh so normal architectures for computing are our computers right right they um their building blocks are Gates they are called logic gates and they work in a sequential fashion so you give inputs and they return outputs so our uh uh new uh Computing architecture instead is uh um is based on a different concept of gates we call them self-organizing Gates so those new gates instead of being uh sequential type of object so input output they are they have these terminals that are input out agnostic it means that they can actually support a superposition of signals that carry input and output information and uh they work like reorganizing their voltages so you can build network of those Gates and you literally embed problems onto them and uh and solve problems using the the the physics of those Gates so this reorganization of voltages is actually the physics of the gates so their Dynamics and so we use exactly this principle to solve problem so instead of using an algorithm implemented on these logic gates that we have today in computers we take the problems and we embed on on this new uh uh uh architecture uh that is self-organizing and use the physics instead of the algorithm to solve problems and so the result is higher output uh the result is that especially for problems that are classified as combinatorial problems which means that you have to explore all possible combinations to find the right one uh uh uh usually with a computer the algorithmic approach is exactly that so checking all possible combinations and finding the correct one now instead if you use M Computing and you use this physics that solves the problem instead of the algorithm what happens is that you don't check all combinations anymore but the physics uh naturally goes into the right one now this is clearly much more complex than what I just said but more or less gives you the idea so you skip this process of checking all combinations Comin exactly but instead you you leverage the physics to go exactly into the combination in that that that you need to solve your problem or is the solution of your problem now this is a technology that you have personally have been working on for quite a long time oh yeah yes yes we introduced the uh self organizing G so we introduced this concept of M Computing in 20134 self organizing gates in 2015 and so yeah it's almost 10 years that we developing this and at what point did you realize that it might have this capability or this level of of you know in encryption or de encryption well we always kind of knew and the reason is very simple uh I have to admit that we were uh um like uh um uh initially what we we were thinking is okay you have a quantum computer and you use the short algorithm to solve the factorization and it leverages this this way of having you know the entanglement many people talk about this entanglement entanglement is exactly uh uh this this uh uh correlation between the Q bits that you leverage you're leveraging physics to perform computation so our thing was okay let's leverage classical physics to do the same and our first goal actually was okay if you we leverage classical physics to can we really factorize so since the beginning actually we were thinking about uh uh uh these uh problems related to cryptography but we never really worked on this actively because once we started uh like developing we saw that uh it actually can be applied to many more different problems much quicker and easier in an easier way like for example scheduling planning all these very hard combinatorial problems but that they are very important for industry and so we started applying to these these other these other aspects was the low ending fruit if you want so how when when you were you were you did you discover this on purpose sounds like yes or yes it's has been a a a a pretty long process uh uh so the goal was uh was so we started uh with the in memory Computing which is a a more um it's a broader concept uh our first AR was were for example on on what we call the dynamic Computing random access memory which is a ram that can perform some logic we started doing these things so it was very basic we were not thinking about combinatorial problems we were thinking just to start having a computational memories so memories that could perform some certain type of computation possibly basic computation but still some computation from there we started like defining uh and when I say we because I did all this part of the work together with Max Dent is a professor at UCSD and we we developed this together and uh uh uh it was the we we we started defining much more abstract concept which is the universal M Computing machine which is basically the the computational model that describes a computational memory okay and from there we discovered okay so you have now this abstract model what are the properties because usually mathematicians do this no you you have you have you have several definition axioms if you want that describe a theory a theory and from there then you say okay I have all of this what I can do with this and you start writing theorems and blah blah blah and so we realized okay wow the computational memory in its most ideal form is actually so powerful that is equivalent to something that is called a non-deterministic touring machine but in practice means that it can solve very efficiently combinatorial problems that otherwise require exponential Resources with a normal computer so that was all Theory at that point complete Theory like we don't even know how to realize this in practice it was like okay we have some mathematical theorem that tells us this but but but how we actually do and so from the then we started thinking okay how we realize this concept in practice and we came out with the with the self-organizing gates which is a uh embodiment of of the universal me Computing machine and so John what at what point did you did you join mem Computing and and come in on this process so I was a entrepreneur and Residence at UC San Diego I've done a few startups they asked me to join them and evaluate their IP and to look for something to spin out and when I was there I met with Dr Fabio traversa and Dr Max dentra and they presented the technology to me and it was very clear to me that the capabilities were something that you know I didn't want to see sit on the shelf right we needed to take this out and try to commercialize it and uh and yeah we all agreed with that and the the university um you know blessed us spinning it out and we've been working at it ever since and when you think about the potential for so as the CEO you're I would imagine you're more focused on the business side of things when you think about the potential for this technology I know you know this is a very exciting discovery that you've made um and frankly kind of terrifying if you're worried about your ability to encrypt anything online or have any kind of private conver ation or transaction um but are there other do you see M Computing pursuing this and following this path are do you see you know this is a thing that maybe you've done and and you spin off and continue on with meem in some other capacity or how are you thinking about this from a from a business perspective to drop back for a second me Computing is a new computer architecture and we're really a deep tech company everything is about the circuit development using the mem Computing based technology so we've actually uh patented quite a few different chips and a long run we see ourselves like an arm we'll license we'll do the design we'll evaluate you know build the chips and design the chips but then ultimately we'll license them and other people will build them themselves right so got it and but now stepping back there when we went out in 2019 we've been solving these uh these um routing scheduling type problems for large companies BP locked Martin Chevron uh also for Air Force Space force and NASA so huge the you know huge comp uh uh comp complex computational problems that they can't solve optimally today that we can solve for them in a minute so uh um the could you give could you give an example of like what type of problem would you would you would fit in that category so the types of problems uh are the easiest one for everybody to understand stand is think of UPS trucks FedEx trucks Amazon trucks an individual truck has to do about 120 deliveries a day to generate an optimal route so they generate routes they obviously are delivering packages they generate routes but to generate an optimal route would take them years and years and years of computer time that's something that we can do in minutes and we've done for some of these other companies so it's a uh the the the problem is considered intractable and the advantage is now if you think about it uh in fact this from from UPS's own marketing if they could save one mile per vehicle per truck per day for a year it's $50 million right so so and what we've shown to the companies we work for and done projects for is we're demonstrating that they can save tens of millions of dollars on an annual basis because we can give them an optimal solution versus an approximation right so but for the for specifically for the prime factorization from the beginning we have have intended it for use by the US government and for control by the US government because of what you noted is that it's pretty frightening what it the implications that it could mean if it got on the outside so it's not our intention to complete it on our own and go after a market uh necessarily we we'd really rather uh see this in the hands of the US government and and let them control it so this is a I mean as you've just alluded this is this is a a a big deal right it's a it is a big deal and the level of funding that you would need to complete and time that you would need to complete this is not significant in the grand scale of things not at all you know if you compare us to Quantum Computing so we're actually delivering today sorry a little bit of a commercial but we're delivering the performance that Quantum Computing hopes to deliver Quantum Computing hopes to break to do prime factorization but they're at least 10 years out and and frankly they've been 10 years out for the last 30 years so but but the government puts hundreds of millions of dollars into Quantum Computing on an annual basis worldwide billions of dollars on an annual basis go into Quantum Computing so there's thousands of of scientists work physicists like Fabio and and Max working on this and they've been doing it for decades we've been doing this for five six years commercially and uh uh with seven people and and with very little money and and so anyway the the the short answer is that it would be a fraction of what they put into Quantum today they it's a rounding error in what they would need to invest in us to get us to take it to the rest the rest of the way and why is like why are you allowed to talk to me about this right now if that's the case well it wasn't our choice so we've we've had requested so we we did this work with the government we actually did it uh it under the guise of a working with an Air Force intelligence group and we had always been telling them and other government officials that work with that we think that this should go top secret and go black and they should you know control it and uh and so we actually we had this capability back in December of last year and and haven't publicized it because we didn't want to um and then ultimately a so but we were we're um you know sharing it socializing it within government agencies and we're talking to everybody and anybody we can talk to and so happens that one of the government agencies who invited us to talk to them give them a presentation and demo made the presentation and the demo live public it was public so they forced our hand which in some ways is good because now we're getting some recognition for the things that we do greater recognition but um it's still there's you know the issue with what this ultimately means and how it should be handled so you at the moment are basically seeking government funding to kind of like retract this information and this this just not retract it but to to re maybe move it back into like a little bit more of an obscure situation where it can be completed finalized used that yeah that would be our preferred method uh we are also talking to large Aerospace organizations because they may have one they have they possibly have the funding they have have de Pockets but they also have the relationships with the government so they might be able to help us take it the rest of the way and with their credibility and their contacts so I'm curious you know this is I think people often think of government funding as like there's so much of it and it's very advanced in your opinion how is it that this uh demo has been able to to slip through the cracks in this way from a security perspective and and maybe I guess a better question is how like who would you be looking to who do you think is the natural entity to fund and Advance this kind of work like who would you like to connect with well ultimately I think this belongs in an intelligence agency right so uh NSA CIA are the obvious ones to to man it and control it and uh figure out what they want to do with it um because it has huge National Security implications and the the the Strategic Benefit that the United States would have uh over their adversaries they haven't had since World War II that's what this delivers but the I think the challenge is just that we're out of the mainstream we're a small company out of San Diego and you know it did it came from UC San Diego it didn't come from Stanford or MIT and we just uh we just haven't gained the awareness and haven't gotten to the right desk yet for somebody to recognize that this is something they should try to pull in well we hope that uh by bringing you to to Future inter rview in just a few weeks um and and by introducing you to some of the kind of folks at at that conference we can help you make those connections um are there things that you would want to say I mean this is a public podcast is are there things that you would want to say to the general public about this technology or about thinking about encryption that this has kind of surfaced uh just you know it it sounds scary uh and it is scary that ultimately it could be developed but but we aren't able to do it today so nobody is is cracking your Amazon account or your bank account or anything like that um and ultimately again that's that's why we think that this is something that should be in the hands of the government um rather than us we don't I'd rather I'd personally not like to have the responsibility yeah yeah that's a lot it's it's a big weight to have on your shoulders right abut having all right well thank you both very much for your time I look forward to spending some time together in person in in LA in just a couple of weeks and um thank you for all of the work that you've been doing it's it's really extremely important and I I very much appreciate what what it is that you're trying to do thank you and thank you for inviting us to the conference we're we're excited absolutely yeah thank you should be great yeah we'll be
2024-01-01 01:29