Good morning, Alex, how are you? Good morning, Young, I'm very well it's nice to see you here! So, tell the audience about your investment fund ideas. What are some of the things that you're looking at how big you are and what are some of your recent investments you get really excited about? So, Amadeus Capital Partners is a venture firm here in Cambridge and London in the UK and we've been around for about 23 years now. We've made more than 165 investments in technology companies all over the world and we focus on deep technology. So, technology with
a lot of science at the sort of cutting edge of innovation. The early stage team which I helped to manage here at Amadeus particularly focuses on the first couple of financing rounds, so called Seed and Series A financings which startups need to get themselves established. You know, we start to see in the history of Amadeus, our firm, a number of different what we would now call unicorns, so you know, billion dollar plus exits from investments we've made including CSR, four scout and network computing network security business and also more recently businesses like Improbable and Graphcore which have also achieved this kind of valuation. Very exciting! Some of the topics that I want to discuss is really around the key mega trending core technology.
As you know, 2020 was supposed to be a great year but it turned out to be a pandemic here. But when I was coming into 2020 I was thinking that there are four C's that are influencing our lives. And the first C was consumer and how the consumer behavior and experience is really changing the way of buying things, experiencing products, the way we communicate. And then the second C was really around what I call cyber security, as more and more people are connected in cloud. What does that mean in terms of their data, privacy, security? And then the third one is about cloud. It seems to me that we're all working on the clouds now and that we're connected and the cloud business is the one that is really growing to be a much bigger size than ever and driving our economy. And the last C
was China, given there is ongoing geo-fencing that is going on but technology in particular. How do you work with two different economies in terms of standardization, in terms of partnership, in terms of IEP, in terms of access? So these are the four mega C's that we've been discussing in the beginning of 2020. Of course, now we have one more C which is Covid-19. So those are the five subjects that I want to kind of focus on this morning. So, start with your view of how do you see AI impacting our lives and how do you see that AI is actually delivering in reality? Because I know you've been doing a lot of early stage investment in AI and other related deep technology areas. And
given your background as the one of the earliest and longest venture capital from out of Cambridge, Hermann Hauser and others have been building this. Alex, and you've been running it now and I just would like to get your perspective of what you see in AI. Sure, so I think the AI environment has been evolving a lot in the last couple of years and first of all people started out believing that delivering complete solutions with AI would be feasible. In practice, I think that mostly what's really happening is that the solutions are around using machine learning to provide some analytics function. But very little full stack solutions are being delivered. Think about autonomous vehicles. Many of the autonomous vehicle businesses are having to reconsider or
consolidate. And you know that's because finishing the whole project from top to bottom in the time scales that were originally envisaged hasn't proved feasible. On the other hand, machine learning for data analytics is becoming really a significant factor in a lot of companies lives and problems of course that arise from that involving areas like bias detection, algorithm drift, you know these sorts of things need work and startups are starting to address them. So we're finding quite fertile opportunities for investing by looking at the infrastructure of the AI market where machine learning needs to be deployed rather than the complete solutions in AI.
So it's in a way, if you kind of use the football analogy, we are still in a very early part, first half, I guess. I think that's right and I think that, you know, in technology there are many optimists and we're excited to build the future. But reality usually intrudes pretty quickly, you know. We think we're heading for a win just because we have a good start to the match but actually the reality is it's a long game and many things will happen along the way. And we have to
accept that the market is evolving around us. So tell me about when you look at the European landscape, the AI companies that you see, what are some of the interesting killer apps that you see and what are some of the companies that you get excited about? So I think in the particular context of the current situation with the pandemic, one of the most important developments has actually been in the area of Covid-19 detection and the development of new treatments. It's very clear that one of the great things which AI can do is to do things like rapidly discover genomic markers which indicate when some kinds of people may be more susceptible to the Covid virus than other people. And also in the drug detection area to identify molecules which can be therapeutic and which can be used to help people find treatments. Of course not just for covid but also for diseases more generally. So we have for example here in Cambridge a great business called Congenica. Congenica's business
is to do the genomic analysis of very large amounts of human data and use that to detect new kinds of treatments for rare diseases. And we've also a business that was invested by Amadeus but it's actually in Valencia in Spain called QUIBIM. That business does analytics on images of for example X-rays or MRI scans and it became very important quite quickly in the pandemic to help doctors detect early signs of the Covid virus in people who were suffering from it, even before they were able to feel themselves that they actually were suffering seriously. So I think those kinds of technologies in the healthcare market are really attractive and you know good evidence of the opportunity that AI holds to improve the quality of all of our lives. So, Alex, I think one of the interesting things about the emerging applications and others of which clearly benefit humankind is about the how you train the data and using those data you can discriminate and being able to filter out what is possible, what isn't and it seems to me those things are now even better than what humans can detect. Because machines can detect each things with much better accuracy than humans of what I understand in imaging applications. So I'd like to get your perspective where are we?
So I think you're absolutely right. I think one of the great attractions of using AI is that it's capable of sifting through incredibly large volumes of data, much larger than humans would be able to do. And then to give support to humans to help make important decisions. If we think about the healthcare applications, for example screening lung X-rays to detect early signs of diseases like Covid, the important thing is to support the radiologist and to help the radiologist to pay particular attention, perhaps to some area of the X-ray or to think more seriously about an X-ray which looks borderline to a human, but which the algorithm can detect is actually showing serious signs of disease. So in this way the AI is supportive of the expertise of the human and at the same time it's able to do much more analysis than the human could feasibly do in a short period of time. The thing that we also should think about is who gets all the data, right? Even the bigger company, the bigger you are the more data you can have and they'll become a better model and then therefore it actually has an advantage for platform players. So do you see
a breakthrough behind just data-driven learning to other ways to make it more efficient? So I think the important thing to be aware of is that new machine learning algorithms are emerging which actually work on significantly smaller amounts of data. At the moment, the history has largely been that more data is better and greater accuracy comes from greater amounts of data. And, of course, first of all that potentially deprives some users of the opportunity because they don't have access to the data. But it also actually is very inefficient, it's not very satisfactory. From a sustainability point of view, data centers have become notorious in, you
know, consuming huge amounts of energy. So these new algorithms are emerging which allow effective machine learning to occur on much smaller amounts of data and there's also an interesting trend in how the mass is actually done. The idea of binarized neural networks, so actually using neural networks where only one bit of information is being processed instead of much wider 16 or 32-bit numbers, which have been the tradition, this can increase the efficiency of these algorithms and so both the requirement for data and the efficiency of the algorithms are actually improving. And these trends, I think, will continue and make this kind of machine learning accessible to a much wider base of users. So it's not just a big large database application but it could even apply to edge computing where you need a smaller set of data and being able to also run it with much less compute intensive by narrowing the bid you can be able to run a much more power efficient way of running the AI applications. So that's really interesting, I'm sure you're looking at some exciting startups in
that area. For entrepreneurs out there: If you have great ideas around that, contact Alex. I'm sure he'll be excited to talk to you about this. One more question around this, it's about AI bias. So as we entrust more and more of our tests on AI, such as screening bank loan applications or rating job applications or whatever the data sorting through AI. We can see more and more risk from biases and AI models. How are the setups or others addressing these issues? This is really important because, you know, clearly the tradition in computing, there's a good phrase to describe the situation "Garbage In Garbage Out", if the data which you're feeding machine learning algorithms doesn't accurately represent all of the people that you're going to need to ultimately study when you're making inferences from the data, then the machine learning algorithm is going to deliver the wrong answer. So it's really important first of all to
curate the data which is being provided in the first place. The second critical thing is to have techniques available to detect when bias or drift in those algorithms so that means the algorithm is becoming less accurate over time. Then, if either of those things are happening, we need to know. There are a number of startups which are offering this kind of analytic capability to monitor the quality of the work carried out by the algorithm. We've invested in a business called Seldon
which has an open source stack widely used by people for deploying machine learning algorithms and Seldon's alibi tool allows people to understand how was a decision made by a machine learning algorithm, what was the basis for the inference. And that helps the operators of that algorithm to develop a clear steer on whether or not there is some kind of bias or drift occurring and as a result to take the appropriate measures to prevent that from being a problem. Here is a very interesting area it's probably too early to say but it's something that we have to watch carefully because clearly we've been living with a lot of bias for a long time. What we don't want to do is AI and machine learning and others accelerate even more to that direction because the data the training is actually feeding that into our sidecar. So I'm glad that you're looking at that as a part of your thinking. How do you make sure the bias is not built in in our thinking and our data and our AI algorithm? So let me change it a little bit to now quantum compute.
I know Hermann called me about five years ago talked about quantum computing so it's been a while that we've been thinking on this concept. So what is the state of quantum startups in Europe? And what are the European unions as well as the institutions that are working on this issue, are they collaborating, are they competing, what's the state of union? So the European quantum scene is very fertile and what's exciting about it is that there are so many different quantum technology application areas, so everything from quantum photonics, quantum sensing, right through to quantum computing is delivering really interesting capabilities which were simply impossible in previous generations of technologies. In the quantum photonics area being able to make sensors which are accurate to detect even as little as a single photon of light can enable imaging with a level of detail which has never been possible before. And it's also possible to build secure communications links using quantum photonics with single photon emitters and single photon detectors, we can essentially make a optical communication link which will be uninterceptable without us knowing that somebody has interfered with it. And that kind of security is extremely valuable. In the quantum computing area computing is really at an early stage. There are six or eight different technologies which exploit quantum properties to carry out computing and as, you know very well, Young, some of the world leading work in this area is delivering only maybe 50 qubits or so of capability right now. So it's demonstrating the possibility of quantum computing
but it's not really yet delivering functional computers which are capable of working on the hard problems in the real world. I think that's the trend which is going to change over the course of the next, probably only two to three years. We're going to start to see quantum computing scaling up into the hundreds then thousands and tens of thousands of qubits. I think most people would agree that when we can get to one million qubits it will start to be possible to carry out really meaningful work with these kinds of quantum computers. So that's probably five years away
but it's actually a horizon which is collapsing pretty quickly. So, you know, just kind of give background to readers of our audience of our program, clearly the benefit of having large number of qubit is about being able to solve problems simultaneously. So think about linear algebra when you have multiple values variables and you want to solve the same time by having this compute requirements in classical computing, clearly we're using zero and ones to identify and be able to manage to take a huge number of iterations. But if you have, imagine it's not zero to one, now you have
many many numbers between zero to one, so you can be able to use all those things at the same time and deploy that and being able to solve very complex linear equations - I try to simplify it for the readers - but I think that is the benefit of having a large number of quantum bits, I guess. Absolutely. This ability of the quantum computer to be able to perform effectively the equivalent of many millions of simultaneous calculations which would be required on a classical computer, means that the ability of the quantum computer to solve problems which were almost unsolvable in classical computers, is really a huge step forward. And so this has applications in areas such as cryptography, in breaking codes but also in things like modeling the way proteins work inside human cells. So helping us to discover new drugs or identifying the chemical reactions that
catalysts go through to accelerate various kinds of industrial processes, enabling us to develop new materials more rapidly than would be possible without this kind of computing capability. Great, so what it does is being able to solve very large variables very quickly with lower power, higher accuracy and these are some of the advantages over class compute. Obviously, quantum compute isn't for everything, right? There are certain things class compute will still exist, still go on but quantum compute can be a complementary to the traditional compute that we've been living with. Let's change the subject a little bit to security. If you look at today, the number of issues that we are dealing with with cyber security is increasing.
As more and more people are working at home, the number of instances is now at a record, the number of ransomware has been going up and they are even going after mission critical space like hospitals and schools and others. So what is your view of the whole cyber security space? Thank you. So, certainly, the security environment has been evolving fast in the last few years. As you mentioned, I've worked myself in the cyber security area for many many years. And what I've seen most recently is that the traditional model where we tried to keep the bad guys out of our castle by building a really high wall around the outside of our organization, this model just doesn't work anymore. Nowadays, we're all carrying mobile devices with us, we're all connecting from home into the
office and so, actually, the wall is already full of holes and is starting to be torn down. So we can't have a defensive model for security any longer, which thinks about just keeping the bad guys on one side of a wall. Really, the significant challenge here is all about people being trained to detect when they are being offered, maybe an email which is a phishing attack, or somebody is trying to persuade them to give up their password details by talking to them over the phone. These kinds of human connections are actually where most significant security
breaches happen. There's a person in the loop, they click on a link or they answer a question, when actually, if they looked really carefully, they could probably figure out they shouldn't do that. So the challenge is to help people to behave better, to help to first of all train them, also to have detection on the network. So we know when breaches occur it's no longer a question of believing that we're going to keep the bad guys out completely. We actually really have to accept that breaches of networks will occur, so we want to detect them early and we also want to have tools in place that help us to slow down or stop those attacks after they've occurred. So that's really
where the focus of a lot of investment activity is now occurring in the cyber security area. So, if you think about cyber security analogous to, let's say a virus in a way that we cannot avoid the virus. If we're going to have it, it's a question of how are we prepared, how do you detect it and how do you response to that. I guess, it's probably what you're saying. Because virus is just not going to be able to protect it. Always unseen, difficult to predict and very costly to creating ongoing
vaccines around you all the time, that you don't even know what's going to come at you. Right. So, just as people really can't avoid disease, it's not practical to live inside a bubble all the time. So we have to have an immune system and we have to have an ability to detect when a
foreign virus or bacteria has entered our body and then to fight it off. And so this is the function of the various mechanisms inside people which help us to defend ourselves against disease. And so, as a metaphor, as a way to think about cyber security, it's quite helpful to think about the enterprise security in the same way to have both systems which detect when an infection occurs and also to have systems which fight it off. But we do have to accept that we are going to
sometimes get sick and so it's necessary to plan for that. And so some of the other implications of that are things like taking good care of the critical information inside your organization, so you actually keep it off the network in places which are better defended when it has really high importance. And you don't allow duplicate copies of sensitive personal information about people for instance or corporate intellectual property to be lying around on the network and able to potentially be picked up if a breach does occur. That kind of good hygiene is actually just common
sense in making sure your enterprise is less vulnerable. So virus and knowledge is pretty good, about having a good hygiene, being able to have the right response and being able to attack the virus. Once you detect it with your antibody system, they can be able to do autoimmune response. All those things actually do have a lot of analogies to our human body. Tell me, in your view, what are some of the exciting companies in cyber security that you've seen? So, I think that the kind of technologies we've been discussing here, which allow us to detect when those attacks or infections occur, they're really important. We invested in an exciting business called Sension which has the ability to detect when traffic on the network contains suspicious behavior and flag it up to the IT security team. So they can
take actions against it. Another investment is a business called Exonar. Exonar helps us to classify all of the data on a corporate network and figure out when it's a duplicate of sensitive information we probably shouldn't keep lots of copies of, so help us to focus the protection of for example encrypting files to only really do that where necessary for particularly sensitive information. Because it's cumbersome and difficult to use encrypted files, most people will recognize the annoyance of having to type a password in to unlock a document they want to read, it's a nuisance to do that, you only want to do it for those documents which really really need it. But figuring out which documents those are and how many copies of this critical information you have in your enterprise actually turns out to be pretty hard. So businesses like Exonar help to actually make that a reality by classifying the information first and then enabling people to decide which information is the most important and protect it to a higher standard.
Great, well thank you for sharing some of your great companies out there. Well, I'm so happy that we can have a conversation around this key technology mega trends that are impacting all of our lives and I really appreciate all the help you have given to me and also my team. I look forward to continuing our journey and have great holidays! Thanks very much, Young. It's nice to chat with you here and we appreciate your support. Wishing you healthy and happy holidays and a much better 2021. Absolutely. Thank you!
2021-01-21