<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Krim Delko on Medium]]></title>
        <description><![CDATA[Stories by Krim Delko on Medium]]></description>
        <link>https://medium.com/@kdelko?source=rss-e38d5b7b13c6------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*v_38mLTxpHcMQrm3.JPG</url>
            <title>Stories by Krim Delko on Medium</title>
            <link>https://medium.com/@kdelko?source=rss-e38d5b7b13c6------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 06:47:59 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kdelko/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Technology needs an ethics roadmap]]></title>
            <link>https://medium.com/@kdelko/technology-needs-an-ethics-roadmap-1264a2d0833d?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/1264a2d0833d</guid>
            <category><![CDATA[quantum-computing]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Wed, 14 Nov 2018 02:08:44 GMT</pubDate>
            <atom:updated>2018-11-14T02:08:44.029Z</atom:updated>
            <content:encoded><![CDATA[<p>Quantum Technology is the new frontier and nobody knows what it will do</p><p>New technology creates enormous social change. While we have very sophisticated processes to develop and commercialize new technologies (such as universities, government support and financial markets), we lack a roadmap for ethics. Technologies are let loose with very little oversight. And even when they are supervised, there is a lack of understanding of what we actually want (let’s define the “ethics roadmap” as the answer to the question what we actually want). So the outcomes are often arbitrary. Quantum technologies are the new frontier and serve as a case in point. There is almost no discussion about how to manage the progress and make sure, quantum technology serves us well. We can do better.</p><p>Quantum Technologies can be separated into two main buckets. One is quantum computing, and the other is quantum sensing. Both have enormous commercial potential and are therefore on the radar of science and finance. While it might be somewhat early to talk about inflection points and real money making opportunities, it is widely believed that quantum technologies will have a non-linear progression. In other words, once it gets rolling, it will quickly dominate academia, business and government. Hence, it pays to play, even if there are no tangible business opportunities on the immediate horizon.</p><p>The same non-linearity argument holds for the ethics discussion. We can’t just wait until quantum technologies develop and become relevant. It will be too late. We need to have a conversation now and build an ethics roadmap to deal the progress of quantum technologies.</p><p>Quantum Computing</p><p>The goal here is to use quantum effects in specific materials for information processing. Quantum effects are properties of tiny particles that exhibit behavior like for example electrons do. For example, they have different energy states, they have different spins or angular momentum. Any of those degrees of freedom can be used for information processing. You can encode information in so called qubits. For example, one qubit can have two states, spin up and spin down, or ground state and excited states. These states can be manipulated with radiation. This manipulation can be structured like in classical computers with gates. Gates are logical input-output tables that build the basic stricture of how quantum bits or qubits are manipulated. On top of these gates we have algorithms who’s goal it is to evolve the qubits in specific ways so computations can be done and results obtained. One particularly useful approach in quantum computing is to formulate a problem as a Fourier transform problem and then use the Fourier transform as a way to find a solution. This is the underlying concept of Shor’s algorithm and many other early applications of quantum computing.</p><p>Now, the physics community is working hard to build a functioning quantum computer. First, devices have to be developed that can exhibit quantum states. Second, those states have to be controlled with as little noise as possible. Third, the result has to be readable. Once we master those three steps, a quantum computer can be commercialized, We are still not there yet.</p><p>Quantum computers are good at solving optimization problems. They will most likely be useful for machine learning, since pattern recognition, principal component analysis and the search for local and/or global optima is key to machine learning. They can also be used for material science since material since questions can be formulated as optimization problems.</p><p>Here is why:</p><p>Given the set up of molecules and their Hamiltonians, what is the most optimal combination these molecules can have against an objective function, such as for example, a material that is light, cheap and sturdy.</p><p>Quantum Computing will have a large impact on society. In fact, it might have a much larger impact than classical computers. Therefore, ethical questions have to be formulated and an ethics road map must be discussed.</p><p>Quantum Sensing</p><p>Sensing is the less popular cousin of quantum computing. There is a joke in the quantum community, which goes a bit like this: “ A bad quantum computer is a good quantum sensor” You can also turn this statement on its head.</p><p>Now, in my opinion quantum sensing is the silent technology which will penetrate fast and be there one day without us even noticing. What is quantum sensing? It means you can measure stuff down to the quantum level, like down to the atom or electron level. You can measure changes in energy states, you can measure differences in radiation etc. The key here is that with quantum sensing devices you can measure tiny changes of things. You can be much more precise and see a world that until now has been hidden to us. Quantum sensing is like getting new eyes to see much smaller. For example, Jörg Wachtrup from the University of Stuttgart is working on devices which can sense electromagnetic pulses coming from our brain. In other words, he is working on a mind reader. The quantum sensors are so precise, they can measure the smallest differences in brain activity.</p><p>As with quantum computing, quantum sensors will have a large impact on society and therefore an ethics roadmap is necessary.</p><p>Ethics and technology</p><p>What is ethics? I will define it narrowly as the question to “What should I do next?” The answer to this question depends on what your incentives are.</p><p>Scientists do what’s next based on their utility function, which is get prestige, money and rise up the academic ladder.</p><p>Business people do what’s next based on the probability to make more money.</p><p>What’s lacking here is a conversation and road map on “doing what’s next” based on what is good for society, humanity or even a country. We don’t know what’s good for us as a collective. We don’t even know what’s really good for us individually, we only know how to maximize monetary and prestige gain. We need a broader ethics roadmap.</p><p>For example, scientists have been drawn to quantum computing because of Shor’s algorithm. This is a powerful computational proof that shows how algorithms based on quantum computers can be better than classical computers. Hence, lots of scientists are working towards quantum computers to implement Shor’s algorithm. But it might not be the right thing to do. Maybe we can do better. We could develop computers to simulate molecular behavior and hence contribute to advancements in material science. How do we know what to work on? How do we know what’s best to do next?</p><p>Often we just do things because others do the same. That is not a good roadmap either.</p><p>So, what should we do?</p><p>One way to solve this question is by doing what feels right. It might sound a bit Californiaish, but we actually often make decisions based on feeling. Our emotions are much better in evaluating complex situations than our brain. We should therefore do what feels right.</p><p>Many of us actually do operate that way. Especially in Silicon Valley lots of people say they do such and such to solve environmental problems, or societal problems. That is the right approach. But, careful. Often people think they do something for the right reasons but then when you actually look at their daily agenda, they do something completely different.</p><p>Google is a good example. You hear a lot of Google employees all the way to the CEO talk along the lines of “We want to solve big problems for the world”. But in their actual day to day life they do very different things. For example, find ways to stuff more ads into youtube, or compete with Amazon on media distribution. I call the gap between what people think they do and what they actually do the <strong><em>ethics gap</em></strong>. Goolge has a huge ethics gap. Other Silicon Valley giants do as well. In fact, the whole technology community suffers from a huge ethics gap.</p><p>What gets measured gets done — develop ethics accounting</p><p>The best way to decrease the ethics gap is to have a formal way of measuring what you do with respect to ethics. I call this ethics accounting.</p><p>Find ways to quantify what you do and compare it to your objective. If your goal is to reduce CO2 emissions with your technology then measure the impact. Make sure you actually deliver the result you wish for.</p><p>Having clear purpose is not just a societal necessity. I argue that even for the hard core scientist who only cares about her progress in the field, having a clear compass helps solving problems. For example. Instead of stating a goal such as, I want to build a better material to increase coherence, it would be beneficial to say “ I want to help build a model for simulating molecular behavior so we can build a better car battery”. This might sound far fetched, but having purpose, a clear compass and direction increases passion and excitement and hence contributes to better solutions.</p><p>This might not apply to all branches of science. Maybe fundamental science, where no applications are in near sight are different. But I feel that this is not the case. Some of the greatest scientist in the field of quantum physics sich as Heisenberg for Feynman always kept a keen eye on society and purpose, even though their field was complex and abstract.</p><p>Call for discussion</p><p>I call for a conversation about the ethics of quantum computing. The discourse can be held in context with organizations such as the <a href="https://futureoflife.org/">Future of Life Institute</a>, or it can be established as a stand alone subject. Quantum Technology is big and impactful. It deserves our attention.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1264a2d0833d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Quantenland 1]]></title>
            <link>https://medium.com/@kdelko/quantenland-1-1ac6090c38a7?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ac6090c38a7</guid>
            <category><![CDATA[science]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Fri, 19 Oct 2018 19:22:49 GMT</pubDate>
            <atom:updated>2018-10-19T19:22:49.630Z</atom:updated>
            <content:encoded><![CDATA[<p>Reality is far more interesting than idealized version</p><p>Quantum Computing has made huge progress. And to some extent it’s much faster than anybody actually thought. This is one of those situations where even an idealized version of realty can’t match up with the actual reality.</p><p>When a bunch of professors and grad students gathered at Yale to study solid state physics under the influence of radiation they came up with a good idea. Cool down aluminum to almost absolute zero and manipulate the resulting substrate with microwave signals. After a few early successes the team realized that you could use this thing as a chip for quantum computers.</p><p>Why? Because supercooled aluminum processed with some oxygen actually behaves like a quantum system. In other words, you can observe quantum like behavior. This is important. But not enough. You also need to be able to control this behavior. The team at Yale succeeded by using microwave signals to manipulate this quantum system. Together you have the basis of a quantum Chip.</p><p>Chad Rigetti was a member of this team. Today he’s assembled a team of leading scientists in his startup Rigetti Computing. The goal is to use the superconducting chip as the basis of a quantum computer and build software to enable others to use it. Rigetti is competing with leading firms such as IBM, Google and others to take leadership in Quantum Computing.</p><p>Andreas Wallraff, professor at the ETH in Zurich, was also part of that Yale team and is also competing with his grad students in the race to build better Quantum computers. Switzerland is actually very strong in this area. Next to Wallraff there are a number of professors at ETH Zurich, there is also Nicolas Gisin at Genava and Daniel Loss at Basel. Not far from Basel there is Joerg Wrachtrup at Stuttgart, who is also very advanced in Quantum systems, although his area of expertise is more quantum sensors.</p><p>The field is growing and Quantenland, the area from Geneva, Zurich to Basel and Stuttgart has some potential to to become the new hub for Quantum computing.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ac6090c38a7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Test drive Jaguar I-Pace]]></title>
            <link>https://medium.com/@kdelko/test-drive-jaguar-i-pace-d431c6c96dab?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/d431c6c96dab</guid>
            <category><![CDATA[tesla]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Fri, 28 Sep 2018 05:27:19 GMT</pubDate>
            <atom:updated>2018-09-28T05:27:19.231Z</atom:updated>
            <content:encoded><![CDATA[<p>Great car — too expensive</p><p>The Jaguar I-Pace is a great example of an electric car made by a traditional car company. The whole thing feels and smells like a normal car. Steering wheel and dashboard look and feel like a normal car. But when you press on the gas you realize, this is not a normal car. The acceleration is great. You feel like flying on the road. So, it’s a great new car built by an old car company. Here is one example of why the I-Pace feels like a new car in old clothes. The front doesn’t have any space, even though it’s an EV. Why is there not much more luggage room. My guess is, it’s because the car was produced on an existing ICE-production line.</p><p>Still, the car is better than comparable ICEs.</p><p>The problem with the I-Pace is twofold. First, it feels like Jaguar doesn’t really want to sell it. They’d rather sell you expensive Land Rovers or SUV Jags than this one. Not sure why, but that’s my feeling I got at the dealership. Maybe it’s because they expect a lot less maintenance revenue.</p><p>Second, there is Tesla. The I-Pace does not stack up to the Tesla. Not even the Model 3, forget about Model S and Model X. Also, the pricing in wrong. At 85,000$ the I-Pace is not competitive with Tesla. Maybe it can compete with comparable ICE’s, but not with Teslas.</p><p>Conclusion</p><p>The I-pace is better than comparable ICEs but doesn’t stack up to Tesla. There is no reason to buy an I-Pace for 85,000$ when you can get a much, much better Tesla Model S for the same, or a Model 3 for much less.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d431c6c96dab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Apple is a good investment]]></title>
            <link>https://medium.com/@kdelko/apple-is-a-good-investment-dc70f2755964?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/dc70f2755964</guid>
            <category><![CDATA[apple]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Mon, 17 Sep 2018 23:58:44 GMT</pubDate>
            <atom:updated>2018-09-17T23:58:44.770Z</atom:updated>
            <content:encoded><![CDATA[<p>Making money by iterating around market leading products</p><p>We sold Apple at 150. Now it’s 220. We are reconsidering our investment. Here is why.</p><ol><li>Apple innovates through iteration and makes a lot of money by doing it. With 1 Billion iphone users Apple has a huge based to iterate against. The good news for Tim Cook &amp; Co. is that this costumer base is sticky. People spend money on iphones and related services and they will keep doing so in the foreseeable future.</li><li>Health is a killer app and Apple has the right know how to fully exploit it. Apple’s core value proposition is safety and privacy. It’s not fancy products or anything like that. It’s the trust people have in using their devices. The Apple watch is a beach head into the health market and it shows how strong Apple can be in this segment. It binds people. Safety is key.</li><li>Functional org chart. Apple is not organized by products but by functions. This is huge for iterative innovation. For example, the team that does face recognition probably can add value in several other areas of the iphone/iwatch ecosystem. This functional approach allows Apple to gradually improve products and come up with incremental innovation. This is what large companies do and Apple does it very well at scale.</li></ol><p>Apple feels like Amazon at this point. It’s a flywheel company that does lots of things around a few core offerings. Those many things it does help it’s core offering get better and better and differentiate against competition.</p><p>It goes without saying that Apple is dirt cheap relative to Amazon, so either the latter is overvalued or the former is undervalued. The truth probably lies somewhere in between.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dc70f2755964" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Lessons from JiuJitsu - 1]]></title>
            <link>https://medium.com/@kdelko/lessons-from-jiujitsu-1-2d5845a6ecaf?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/2d5845a6ecaf</guid>
            <category><![CDATA[sports]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Tue, 28 Aug 2018 17:58:50 GMT</pubDate>
            <atom:updated>2018-08-28T17:58:50.897Z</atom:updated>
            <content:encoded><![CDATA[<p>Slow practice developing instinct for finish</p><p>It’s too good an art not to be shared. JiuJitsu is one of those rare things in life that sound good, feel good and get amazing results, all in one. The technique allows you to finish a fight without hurting anybody, keep you and the opponent safe and in most cases build long lasting respect between you and the opponent. It takes a long time and don’t be fooled into a quick lesson promising you self defense skills in an hour. However, if you put in the time you really do get lots of benefits, physically and mentally. This series (Lessons from JiuJitsu) is dedicated to the transfer of knowledge between the art of JiuJitsu and other parts of life.</p><p>One of the most useful drills in my training is what I call “Slow practice developing instinct for finish”. Let’s say there is a particular technique that allows you to finish your opponent, like an armbar.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/294/1*GrxmaXTD8VHr2oyzHokNPA.png" /></figure><p>While very effective, this technique is not easy to set up. So, when we drill this technique, we drill slowly. The partner moves around and every now and then gives an opening, so you can set up the armbar. It’s a simple idea, but incredibly effective. That way you don’t spend all your energy hustling to set up something that resembles the armbar. You have time and space to focus on the essential moves necessary to perform the armbar. After a while, you can increase the speed of the drill. But no matter how good you are , you should always go back and drill the slow version. It’s building mussel memory. The slow drill to develop instinct for finish really works.</p><p>I’ve applied this technique with my son. He is a goal keeper in soccer and 11 years old. I help him develop technical skills. One of the hardest things for goalies is the one versus one:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/259/1*hcItfC9P2vi_tvXKhPYIFA.jpeg" /></figure><p>I am applying the “slow drill to develop instinct to finish” for the one versus one skill training. We slowly approach the goal and give him openings, i.e. hints the ball might go right, or left or higher or lower etc. We go slow and then slowly increase the speed. Then we go back to slow and so on.</p><p>The drill is showing results. It doesn’t look pretty, but I am convinced the the mussel memory is being built up. Also, as a goalie you have time and space to build confidence for such situations, so when the real thing happens in a game, you get the feeling you’ve been there. It’s a bit like when you go back to visit your old school and think about how you were nervous when you took your tests. Today, you are much cooler about that. This kind of confidence helps a goalie master the one versus one. And I am convinced that the “Slow drill to develop instinct for finish” helps tremendously.</p><p>Lotos of people in the sports community emphasize talent and natural gift, which is good. But not enough emphasis is put on ways to develop players and give them special skills. Drills such as “slow drill to develop instinct for finish” are exactly the kinds of techniques that help people do things most would have thought can only be done by natural gift.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d5845a6ecaf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The role of government in Technology]]></title>
            <link>https://medium.com/@kdelko/the-role-of-government-in-technology-85cbb59f8fa2?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/85cbb59f8fa2</guid>
            <category><![CDATA[self-driving-cars]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Thu, 23 Aug 2018 04:14:14 GMT</pubDate>
            <atom:updated>2018-08-23T04:14:14.498Z</atom:updated>
            <content:encoded><![CDATA[<p>Direct initiatives are bad and framework building is good</p><p>This is the first article of a series about our thoughts on government and how it can help technology innovation. In this piece we focus on direct initiatives such as the DARPA self car challenge. We argue that such initiatives, while often perceived as beneficiary, in fact harm innovation. Instead of influencing outcomes directly by following its own agenda, the Government needs to focus on frameworkbuilding and enable innovation and entrepreneurship.</p><p>When the Defense Advanced Research Projects Agency (DARPA) announced a self driving car challenge in the mid 2000, lots of scientists jumped on the bandwagon. DARPA’s goal was to spark innovation for the development of unmanned vehicles that could be deployed in deserts and other rough terrains, where enemies have easy targets.</p><p>The DARPA challenge is considered one of the cornerstones of the development of self driving cars. We argue, that its not. In fact, it even had a detrimental effect on the industry.</p><p>When the DARPA challenge was launched, computer vision was still only a lab project. There was not enough computing power and not enough sensor technology at affordable prices. So the contenders in the DARPA challenge were forced to use other approaches. One of them is what the industry today calls sensor fusion, where lots of different signals such as radar, lidar, light and sonar are used to guide a car.</p><p>Today, computer vision is much more advanced and it is possible to build a car that focuses on computer vision as primary guide for driving. Also, the algorithms used today are much more flexible, since they are written in sort of self assembling way, i.e. the software is written in general form and then adjusts itself according to the data in a continuous learning process.</p><p>It would have been better for DARPA to wait 5 or 6 years and then launch a challenge. In other words, DARPA was too early. Had the agency waited, the contenders would have been able to develop much more sophisticated technology. Even today, many academics and practitioners are still stuck in the DARPA way of doing self driving. While not wrong, it is not the most efficient way to go forward.</p><p>The DARPA challenge example shows that government initiatives can backfire and have unintended consequences. Instead of pushing its own agenda, DARPA should have waited until the technology community was ready with technologies such as computer vision and better algorithms. It is often better to let the industry evolve organically than push a desired outcome.</p><p>We argue that the same applies to the moon landing. While the Apollo program has enabled lots of technologies, it has not achieved its primary goal, which is space exploration. The main reasons again is, that the technology just wasn’t there. Today companies like SpaceX are pushing the boundaries in space thanks to the availability of technologies. Today, the government can support such initiatives and create a fertile ground for other entrepreneurs and companies to flourish.</p><p>Conclusion</p><p>We argue that governments cannot decide when technologies are ripe for development. This is an organic process that most often takes time. Governments can certainly influence outcomes buy offering a healthy legal framework, stable financial conditions and good patent and visa policies. Also, governments can act as customers, which is particularly important for startups such as SpaceX. But direct initiatives to drive a technology just because the administration wants to have the technology are counterproductive.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=85cbb59f8fa2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why we like this talk]]></title>
            <link>https://medium.com/@kdelko/why-we-like-this-talk-bb1bd892c382?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb1bd892c382</guid>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Wed, 22 Aug 2018 03:44:36 GMT</pubDate>
            <atom:updated>2018-08-22T03:44:36.158Z</atom:updated>
            <content:encoded><![CDATA[<p>Prof. Sadoway is exactly the kind of person we want to support as investors</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FSddb0Khx0yA%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DSddb0Khx0yA&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FSddb0Khx0yA%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/c13ca04c33bf4c2af949f38aa2d2f475/href">https://medium.com/media/c13ca04c33bf4c2af949f38aa2d2f475/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb1bd892c382" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our investment in Intel]]></title>
            <link>https://medium.com/@kdelko/our-investment-in-intel-d0b8a2afa350?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/d0b8a2afa350</guid>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Wed, 22 Aug 2018 02:31:44 GMT</pubDate>
            <atom:updated>2018-08-22T02:31:44.237Z</atom:updated>
            <content:encoded><![CDATA[<p>The end of Moore’s Law, 5G and AI will drive the business</p><p>Intel trades at a PE ratio of below 12. They have no CEO and the competition is on their toes. That’s why the stock is inexpensive. However, we believe the story is and will become much more interesting than what’s currently priced in the market.</p><p>End of Moore’s Law</p><p>It’s very difficult to squeeze even more transistors onto a chip. That’s why Intel has turned from CPU centric to data centric. What that means in a nutshell is that Intel is driving performance with software, optimization and new memory solutions instead of just delivering faster chips. The latter is still important, but it’s not so easy to achieve. Hence, the performance improvements through various optimizations across the stack have more value.</p><p>One particular piece we find interesting is the Intel Optane Persistent Memory. It’s a kind of like DRAM but behaves like storage and costs less than DRAM. Intel is changing the memory hierarchy and thus offering new ways to run applications. For example, with this type of memory vast amounts of data can be processed in faster speed using less power. Particularly since AI has become a driving force in workloads, memory has become a bottleneck. With this new product Intel is disrupting the vast memory market.</p><p>With the end of Moore’s Law hybrid computing will become much more prevalent. This means that existing infrastructure has to be used more efficiently. Bottlenecks such as idle CPU time, interconnect and memory bus in the processor as well as networking and communication between workstations can be solved to improve overall performance. This is a great opportunity for Intel.</p><p>5G</p><p>5G has three major KPIs.</p><ol><li>Faster data speed</li><li>More bandwith</li><li>Less latency</li></ol><p>With 5G the line between the edge and the data center is blurred. You don’t care as much where you do your compute. Data can be moved fast and in vast amounts across the network. Intel has various initiatives to attack this opportunity. One is modem. But there is a lot more. They have edge compute solutions for AI such as Movidius.</p><p>But fundamentally, 5G will create a large data pool which can be exploited across the network. Hence, optimization solutions between the edge, the data center, memory and CPU will allow for faster performance and lower cost.</p><p>AI</p><p>The main opportunity in AI for Intel is compute and memory. As Intel AI chief Naveen Rao says, “AI is compute and Intel is compute so Intel is AI.” This is not just true, but also a necessity. If you don’t transition into the world of AI as a hardware company, you cannot survive.</p><p>Competition</p><p>The main drawback on Intel is their lack of 10nm success. This means that competitors such as AMD can use the 7nm capacity at companies like TSM to compete effectively. It is understood that Intel 10nm is equivalent to TSM 7nm due to density. This is a valid threat.</p><p>GPUs are also gaining share in the data center. While they have a role to play, it is unlikely going to eat into Intel’s business. In fact, the competition with with GPUs is forcing Intel to move faster can disrupt the AI market with innovations such as Optane Persistent memory. GPUs have to solve the memory bottleneck as well.</p><p>ARM is talking about an ARM based CPU. This is relevant since PCs in a world of 5G could work as virtual machines with less compute power and let the cloud do the heavy lifting. This will cost Intel market share in the PC business. 5G could bring a world of a truly network machine, a computer that just delegates tasks to the cloud and doesn’t have much performance on the board itself.</p><p>These are great risks. These are also the main reasons why Intel is trading at such low valuations.</p><p>We believe that there is more upside and Intel has a good chance to keep delivering high single or double digit EPS growth.</p><p>There is also the longshot of Quantum Computing. While not relevant right now, it will be relevant in 3 to 5 years. Intel is participating in the race and if they succeed this could propel the stock to all new levels.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d0b8a2afa350" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our view on the Tesla buyout]]></title>
            <link>https://medium.com/@kdelko/our-view-on-the-tesla-buyout-9551aff8e4e7?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/9551aff8e4e7</guid>
            <category><![CDATA[tesla]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Thu, 09 Aug 2018 03:31:36 GMT</pubDate>
            <atom:updated>2018-08-09T03:31:36.462Z</atom:updated>
            <content:encoded><![CDATA[<p>Elon Musk announced on August 7th that he is considering buying out Tesla at 420$. The stock currently trades at 370$. We are sellers in the mid 400s.</p><p>However, if the deal goes through, we will have to tender.</p><p>Before we argue about the pros and the cons of the deal, let’s look at the probability of the deal actually happening. The best way to gage the implied deal probability is to look at the premiums of longterm Tesla options. For example, the January 2020 500 Call went from 40$ to 20$. So the market is still ready to pay 20$ for an option which will be worthless if the deal goes through. However, the market is paying considerably less than before Musk announced his intention to buyout Tesla.</p><p>Given the options market pricing, we are faced with several options.</p><p>A. We stay long the stock and cash out at 420$ if the deal happens.</p><p>B. We buy Jan 2020 500 Calls for 20$. If the deal goes through we loose money on the calls. If the deal doesn’t go through we most likely make money, since the premium will jack up. Also, buying the calls gives you optionality, what if the deal doesn’t go through but somebody else decides to buy Tesla at ta much higher price? Tesla is a very volatile stock. That makes sense, since the company is full of optionality. Buying that optionality at a low price could be an interesting strategy.</p><p>C. Covered call. We stay long stock and sell the Jan 2020 460 Calls for 28$. If the deal happens we make 50$ on the long and 30$ on the covered call. We make 80$ on a 370$ position. The calls add 8% ish to the return.</p><p>D. Sell Calls and Puts. You can sell the Jan 2020 450 Calls for 35$ and sell the Jan 2020 350 Puts at 55$. If the deal goes through you make 90$.</p><p>The options market is pricing about 50% chance of a deal happening. For us that means we can take a bet, will the deal happen, and if so, at what price. There is enough premium in the market to take a bet.</p><p>Currently we are long the stock. We might add an options position.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9551aff8e4e7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why we went back into Tesla]]></title>
            <link>https://medium.com/@kdelko/why-we-went-back-into-tesla-f6e36646e8fd?source=rss-e38d5b7b13c6------2</link>
            <guid isPermaLink="false">https://medium.com/p/f6e36646e8fd</guid>
            <category><![CDATA[tesla]]></category>
            <dc:creator><![CDATA[Krim Delko]]></dc:creator>
            <pubDate>Mon, 06 Aug 2018 04:33:35 GMT</pubDate>
            <atom:updated>2018-08-06T04:33:35.199Z</atom:updated>
            <content:encoded><![CDATA[<p>When we sold Tesla last November, we had two major concerns. First, we weren’t sure if they can solve the manufacturing problem with the Model 3. Second, we were concerned with key people leaving the company.</p><p>Now, we’re less concerned and we took a a sizable position in Tesla, since both of those risk factors have been alleviated. The manufacturing issues are getting solved. This is a slow process, but the company is making steps forward. That’s a good thing. The risk is still high as reflected in the high risk premium on the stock and the bonds, but it’s better than 9 months ago.</p><p>The issue with key people leaving is not as bad as we thought. Recently we attended a talk by Tesla AI chief Andrej Karpathy and it made us think differently. Karpathy introduced his concept of <a href="https://medium.com/@karpathy/software-2-0-a64152b37c35">Software 2.0</a> which is a very interesting approach to developing software. It’s highly likely, however, that this produces a clash within the company, since more seasoned software professionals might not embrace this concept as easily. Anyway, we don’t want to dwell on the pros and cons of Software 2.0. But the key here is that Musk and Tesla are trying something very new and whenever people change course, there will be road kill. This is an angle we haven’t considered before. This is probably a large part of the reason why there is a lot of churn in Tesla’s higher ranks. It doesn’t make it better or worse, it’s just a logical explanation for high churn.</p><p>Now, we are pretty happy that Tesla seems to have improved the production issues. But we believe the key to the success of Tesla will be software and specifically two aspects of AI. One is the self drive and the other is advanced automated manufacturing. Our core thesis on Tesla is that they will advance manufacturing to a point where competitors will not be able to follow, a bit like Amazon in retail.</p><p>This is harder to achieve with the self drive since there are others even today which are as good if not better than Tesla. But we like the Software 2.0 concept and are willing to take a bet that this will give Tesla an unfair advantage. If Karpathy succeeds with his idea to implement this in production mode in a real company with real customers, it could become a massive advantage.</p><p>Given how big the market for transportation is (it’s even bigger than retail), this offers lots of upside for Tesla. The risk is high but the reward is much more interesting. We are shareholders again and it feels good to be back.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f6e36646e8fd" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>