 So hello everybody I was Jake just waking up this morning twisting my hat a little bit That's maybe the reason why I walk a little bit like a robot. So it's not part of the show. It's what's the twisted neck So we're Gemma this and we design a chip that works like the human brain I'm pretty sure you guys know all about the amazing opportunities that arise from the use of AI today And that are gonna arise from the use of AI in in some years And I'm like a really tech-positive person because I think that we are gonna need this innovations that will come from AI To solve the really big problems that we are facing in the future however, I've bad news for you today because There is a problem with the current AI technology and the current AI hardware because a AI hardware is way too inefficient to operate Okay, the slides are not moving. So I have to do it without slides Is you see chat GPT is consuming more than 100 million US dollar every every month for operation costs for its servers because the energy Requirements of the hardware are way too high or if you look at Tesla The newly developed dojo chip is taking more than 30 days for one single training cycle It means you gather all the data you feed them into the system and then you have to wait more than 30 days While consuming tons of energy until the training is done. Afterwards, you can look at your Neural network and see if it was fruitful or not And if it's not fruitful you have to re-engineer this and that's the reason for this is that there's a separation in The architecture today between the CPU the information processing entity and the rum the storage and this separation between I don't know. There are the slides No This separation between both entities Leads to a communication bottleneck because both entities have to communicate all the time with each other and the most of the time and energy within the processing for the neural networks is going to the actual Communication and not the processing itself. You can think of it like Driving up to work six hours a day for just a two hours work day There you see that the most of the time and the energy you you afford is for the driving and not the actual work during the two hours so That is perfect So you see on the left side the separation and now we come to the right side So what we do at Gemma this is we use a novel component called a memrist us To adopt the information processing mechanism of the human brain because the human brain is like a natural blueprint it is very very intelligent and It is only running on the energy of light bulb But using the memrist or the memrist or Can store and process information in the exact same place and we build neurons and synapses in a very abstract way To adapt this information processing mechanism You can think of How people develop the plane years ago They looked at the bird and they looked at it how it flies around and then they were trying to adapt the Mechanisms that it uses so if you look at planes today, they don't look really like birds But you can see some coexistences. So for example, they have wings and With this with this example you can think a little bit of how we do it when we look at the brain and try it's adopt this mechanisms for our hardware so Our product is an AI chip that is 100 times faster and 100 times more energy efficient than state-of-the-art technology Adding up to a 10,000 times reduced energy consumption overall Because we use standard processing mechanisms with CMOS production lines We are as cheap as a chip in a smartphone and Because our chip is really really small and very energy efficient We can nearly build it into every mobile device enabling decentralized AI training on my smartwatch on your mobile phone Or on any sensor you could think of so Or vision is that there will be will be a future when every processor on every device will come with an gem of this AI chip and Why is this important or why do we think that is? Changing something because what you do and what you all have probably at the moment is you have a smartphone with a series So if you have an apple it's a series on it and the series You're talking to is except the same series as my series because how we do it today is we We gather all the data on all mobile devices Send them to a centralized place where the neural network is trained and then we deploy them on every small Smartphone so your series is just a mean of all data gathered everywhere But actually and that's for me is sometimes a little bit funny or smartphones are not smartphones I would rather call them stupid phones because They are just everywhere the same so I want my smartphone to understand me I want to understand my German sarcasm and maybe my German accent so With our chip this will be possible because we can really train neural networks on devices so To push or chip into the market we license it together with semiconductor manufacturers and they help us to push into different market applications for us there are two income streams and a license fee and the per chip royalty the per chip royalty goes to the Company every time a chip is produced with or architecture You can think of it like having a smart having Samsung and they want to build a new TV and within this TV They want to have like an AI middle that's able to learn so they can go to the semiconductor manufacturer ask them to buy or Chip design and then it will be integrated in their devices So they have a better product the semiconductor manufacturer had better sales and we get money for every chip produced So so far we have quite great traction and we drop on more than 50 years of research on or research Institute back in Bochum in Germany and We are now funded with more than 3 million euro and non dilutive by the German government Where we were able to build up a team of approximately 15 people and eight people full-time and One year ago. We were able to finish our first simulation of our chip design to prove that the Theoretical aspects behind it are really working and one month ago We were able to build the first neurons and synopsis and hardware to really replicate what we did in the simulation Next part will be to really build the first IC chip In the semiconductor manufacturing line and then start scaling it up from there I think we are the right time at the right place with the right idea because Europe is heavily demanding More chip design from its own from its own countries because as you see at the moment They are just 2% of it and as we see at the moment There are many many programs pushing AI and semiconductors start us within this place. So we have a great place for funding We have a strong team. So there are two of my co-founders Dennis and Enver and they got the background Of Bayer inspired computing and they were developing this idea this chip design during their PhDs Then we have Daniel Kruger. I think he's over there today So if you want to talk to him you're you can just come around and he's an IC day Engineer he has an experience from the Harvard University and work for a brain-implant startup. I myself I'm a former management consultant and responsible for all the non-technical stuff We have a quite great team of advisors We're really like to highlight Jamie Urquhart Jamie Urquhart was a co-founder of arm And he was able to bring arm to one of the 200 most valuable companies in the world in adapting the same business model as we do So we're here today because we're gonna raise our first seed round by beginning of next year So if you invest or come around talk to me talk to Daniel And yeah, if you if you're a company wanting to use our chip in the future Just come around or drop me a message on LinkedIn. We are Genesis and we sign a chip that works like the human brain. Thank you