<?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[EMBEDDED VENTURES - Medium]]></title>
        <description><![CDATA[Embedded Ventures…The VC Remix — Defense, Dual-use, Edge Compute, Quantum, Communications, Manufacturing, Robotics, Biotech, Battery/Power, Sustainability - Medium]]></description>
        <link>https://medium.com/embedded-ventures?source=rss----ce419602c934---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>EMBEDDED VENTURES - Medium</title>
            <link>https://medium.com/embedded-ventures?source=rss----ce419602c934---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 07:40:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/embedded-ventures" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Mechanical CAD: Yesterday, Today, and Tomorrow]]></title>
            <link>https://medium.com/embedded-ventures/mechanical-cad-yesterday-today-and-tomorrow-981cef7e06b1?source=rss----ce419602c934---4</link>
            <guid isPermaLink="false">https://medium.com/p/981cef7e06b1</guid>
            <category><![CDATA[software]]></category>
            <category><![CDATA[hardware]]></category>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[mechanical-engineering]]></category>
            <category><![CDATA[engineering]]></category>
            <dc:creator><![CDATA[Jessie Frazelle]]></dc:creator>
            <pubDate>Tue, 23 Mar 2021 18:11:39 GMT</pubDate>
            <atom:updated>2021-03-23T18:11:12.980Z</atom:updated>
            <content:encoded><![CDATA[<p>Computer-aided design, or CAD, has been around since the 1950s. The first graphical CAD program was called Sketchpad and came out of MIT¹. Since then, CAD has become essential to designing and manufacturing a hardware product. Today there are multiple different types of CAD. This article will be focused on mechanical CAD, used for mechanical engineering.</p><p>In digging into the history of computer graphics you will find some interesting connections between the most ambitious and notorious engineers. Ivan Sutherland, who won the Turing Award for Sketchpad in 1988, had Ed Catmull as a student. Ed Catmull and Pat Hanrahan won the Turing award for their contributions to computer graphics in 2019. This included their work at Pixar building <a href="https://renderman.pixar.com/">RenderMan</a>, which was later licensed to other filmmakers. This led to innovations in hardware, software, and GPUs. Without these folks, not only would we not have mechanical CAD, we also wouldn’t have the level of animated films we have today. We wouldn’t even have GPUs!</p><p>Modeling geometries has greatly evolved over time. Solids were first modeled as wireframes by representing the object by its edges, line curves, and vertices. This evolved into surface representation using faces, surfaces, edges and vertices. Surface representation is very valuable in robot path planning as well, not just modeling solids. Wireframe and surface representation contain only geometrical data. Today, modeling includes topological information to describe how the object is bounded, connected, and its neighborhood².</p><p>OpenCascade, Parasolid, and ACIS are all what is known as a boundary-representation kernel, most people refer to them as B-rep kernels. A B-rep model is composed of geometry and topology information. The topology information tends to differ depending on the program used. B-rep file formats include STEP, IGES, NX’s prt, Solid Edge’s par and asm, Creo’s prt and asm, Solidworks’ sldprt and sldasm, Inventor’s ipt and iam, and AutoCad’s dwg.</p><p>Visual-representation, or vis-rep, models tend to be much smaller in data size than b-rep models. This is because they do not contain as much structural or product management information. Vis-rep models are approximations of geometry and are composed of a mass of flat polygons. Vis-rep file formats include obj, STL, 3D XML, 3D PDF, COLLADA, and PLY.</p><p>CAD programs tend to use b-rep models while animations, game development, augmented reality, and virtual reality tend to use vis-rep models. However, the two are interchanged frequently. For example, if you had a b-rep model you were using for manufacturing but wanted to load it into Apple’s ARKit for some animations, you would first convert it to COLLADA, a vis-rep file format. The file should already be a bit smaller from dropping all the CAD data. If you wanted to make it even smaller you could tweak the polygon counts on each of the meshes for the various parts.</p><p>The tools we build with today are built on the shoulders of giants! But there is a tremendous amount we could do to make them even better. At some point, mechanical CAD lost some of its roots of innovation. Let’s dive into a few of the problems with the CAD programs that exist today and how we can make it better.</p><h3>Single-threaded</h3><p>Since most CAD kernels are built on cores from the 80s, they were not built for modern systems. Even if you have the latest CPU or GPU it won’t do much to help the performance since most of these programs are single-threaded³, or have single-threaded aspects, and have no awareness of a GPU. Sure, some of these kernels have been updated since the 80s but their roots are still tied to their predecessors⁴.</p><p>That does not mean that all CAD kernels are _entirely_ single-threaded. Parasolid is multi-threaded but that still means if you are importing or exporting to a file format other than Parasolid, you might have just switched back to a single-threaded process. Another example of a multi-threaded kernel is <a href="https://www.implicitcad.org/">ImplicitCAD</a>, which is written in Haskell.</p><p>One problem with making a whole CAD program multithreaded is the different file formats. For example, a STEP file, whose format dates back to the 80s⁵, pretty much mandates the need for a single-threaded process⁶. Additionally, most parametric CAD operations are single-threaded. However, the open-source project <a href="https://solvespace.com/index.pl">SolveSpace</a>, which uses non-uniform rational basis splines (NURBS), has some parallel operations.</p><h3>Duplication</h3><p>In software development, when you want to get the contents of a memory address, you use a pointer. This allows you to reference the same content of memory over and over again without the expense of copying the content itself.</p><p>Some products built in CAD may never duplicate a part of their model ever — lucky for them! For folks who do have multiple similar parts in their CAD designs, most CAD programs are creating very expensive copies of the parts.</p><p>For example, let’s imagine a model of a server rack. The _default_ method of copying a part⁷ in Solidworks, as well as many other industry wide CAD programs, is to copy the entire contents of a child model to a new model. So if we have 32 sleds in our server rack and we use the default copy method in Solidworks, we have 32 of the exact same model in individual copies. This is very expensive! As you can imagine, each sled has a bunch more models inside, and then those models have child models as well. This exponentially increases the work on the kernel and on our program to even load our model in the first place since the program does not know these are all the same thing.</p><p>If we think back to software development, what we really want is a form of pointer to the model. In the CAD world, these are called instances. Then we can have one copy of the model stored, and all the other instances are actually just a reference to the original copy. This also saves the user a bunch of time! Imagine having to update parts of models in 32 different locations when a part in a sled changes. No one could describe this better than Albert Einstein when he said, “The definition of insanity is doing the same thing over and over again, but expecting different results.”</p><p>While the default copy method in Solidworks is very expensive, they do offer another option that is more in line with how pointers operate. However, since this is not the default, you can imagine most people might not even realize there is a better way. This might seem mundane and the answer should be “just use the other way”, but products should try to make the default path lead to the least amount of pain. Instead of having two methods for copying, have one! Make the default method act more like a pointer (or instance) _until_ the geometry, surfaces, or topology of the copy (not the main) has changed. In which case warn the user that this will now act like a unique part aside from the main copy. Or the user might have mistakenly meant to apply those changes to all the copies in which case they should apply the changes to the main copy.</p><p>There is another huge problem with this as well. Each individual CAD program has their own way of implementing and referencing instances. So if you export your program from one CAD program to another, likely you will still have 32 individual sleds versus the concept of 1 sled and 31 references to the original sled with only the xyz coordinates changed. Some programs offer ways to import instances but it all relies on the type of file format you are importing and if they have the support for that.</p><p>As an additional cherry on top, even if you are using instances you are still at the mercy of the single-threaded kernel and likely none of the copies are going to render in parallel.</p><h3>Version control</h3><p>For software teams that are accustomed to using git, having the ability to diff, fix merge conflicts, and work as a team in parallel on the same file is a huge time saver. For mechanical CAD, there are a number of startups trying to solve this problem.</p><p>Instead of re-inventing version control for CAD, folks who use git today want to continue to use git and not have to add another tool to their workflow. The way CAD files are today, there is no way to push them to a git repo, have several people modify the file, and resolve merge conflicts⁸. For all the startups working to solve version control for mechanical CAD, this is why they had to reinvent the wheel.</p><p>However, in a world where we could make a kernel that fully utilizes a modern CPU and GPU, can we not also use a file format that is human-readable and would allow for resolving merge conflicts? When you ask yourself “what is human readable and works well with git?”, the first thing that comes to mind is a programming language.</p><p>The other great win from using a programming language is: even if you don’t use git or want to use git, there are many different options already for version control of human-readable files. Additionally, integrations with GitHub and other version control tools could be extended with wasm support so that diffs could be visualized as renders as well.</p><h3>Programmable</h3><p>Let’s think back to our example of the rack of servers. If part of our rack contained complex math that we were calculating in a program like Mathematica, we would have to continuously re-evaluate the math in another program and update it in our model. If instead we could program in the CAD program itself, then we could do all our calculations in one place and the model would update if anything in our equations changed.</p><p>For our rack of servers, each sled has network cables that connect to the back of the sled. Using the GUI it is quite hard to make these perfectly aligned with the connector on the sled. Someone would have to sit with the model for an hour or so just tweaking each individual cable to be perfectly aligned. That is a huge waste of time! Rather, if we could program the alignment of the cables, we could ensure each was perfectly aligned with the connector.</p><p>The need for programming becomes even more acute if you want to do mesh or topology optimizations. Unfortunately, most optimizations are implemented through GUI click interfaces, and given their complexity to define, can often be more trouble than they are worth. Today, some of the industry used programs allow for scripting, but their APIs are COM based and as you can imagine built in the 90s. But it’s great they even offer it for the ones that do! Thank you AutoCAD for being the first CAD command line interface I got to use!</p><p>For the modern world, it would be great to generate sdk clients for the CAD program in every language. Much like how API clients are generated. This way, anyone could program in any language! It greatly lowers the barrier to entry since you wouldn’t have to learn a new language. This would allow for complex math to be done in the CAD program itself versus using Mathematica, MATLAB, or Wolfram Alpha.</p><p>A few scriptable CAD programs exist today and are great examples of paving the way for this: <a href="https://www.implicitcad.org/">ImplicitCAD</a>, <a href="https://libfive.com/">libfive Studio</a>, <a href="https://www.openscad.org/">OpenSCAD</a>, <a href="https://github.com/CadQuery/cadquery">CadQuery</a>, <a href="https://www.freecadweb.org/">FreeCAD</a>, and <a href="https://github.com/cbiffle/ruckus">ruckus</a>. <a href="https://www.blender.org/">Blender</a> has a great console interface. <a href="https://threejs.org/">Three.js</a>, while not CAD oriented, is also another great example of 3D programming languages. Jonathan Blow’s <a href="https://oxide.computer/podcast/on-the-metal-9-jonathan-blow/">Jai</a> is for writing systems-level code, and a great example of creating a language thinking heavily about performance⁹.</p><p>Most of the mechanical engineering community is tied to the GUI, so you’d also need a way to generate code from GUI interactions. This is quite similar to an HTML “point and click” GUI that generates code on the backend. This allows folks who want to script to script and others who want to click can click. Both worlds can be happy — code on the left side, render on the right, just like a markdown editor.</p><p>If there is an SDK client for the CAD program and underlying kernel, you can imagine a very rich ecosystem of plugins and tools emerging much like the ecosystem that surrounds VSCode, vim, and emacs. Most CAD editors used for products are very closed off and don’t allow for this type of community-based development and sharing. Plugins could be written for any use case, for example: mesh/topology optimizations and supply chain system integrations. This includes the functionality for finding parts, creating bills of materials (BOMs), and computing lead times for parts of the model. Today, this is usually done in separate programs or even spreadsheets.</p><p>I’d personally love plugins to support a command+P function. In most programs, when you want to print something, you hit command+P¹⁰. For mechanical CAD, when I want to “print” my model the underlying program should discover all the 3D printers and machines on my local network (or plugged directly into my machine) and send the parts of my model that are compatible with that machine to be printed. You could even take this a step further — in a fully automated factory with robots, the program should set up and start the assembly for the model and all the parts.</p><p>Speaking of 3D printing, let’s dive in a bit to the STL file format. The STL format was defined in 1987 and its namesake comes from stereolithography, the first method of additive manufacturing. STL files represent geometry in a series of triangular surfaces. Since STL is a vis-rep format, it does not hold any data about internal structure, color, texture, or any other CAD data that a b-rep format would contain. Modern 3D printers have innovated past the simplicity of the STL format. For example, if you wanted to print a full color model, you should use a VRML (Virtual Reality Modeling Language) file or you’d need an STL file associated with textures in order for the printer to add color and texture to your object. Plugins can ensure that your printer gets the right data for your specific model, without the pain of conversion and make sure no materials or textures are dropped.</p><h3>Testing</h3><p>The test flow of CAD models usually consists of running simulations. These can be for various different things but we can use airflow and thermals as examples.</p><p>In the software world, after pushing our code updates, typically a CI is run on our changes letting us and our teammates know if we broke anything and our code is safe to merge. CAD should work in the same way. If we make changes to a model, our simulations should run in a CI to let our teammates know if our code is safe to merge. Most of these simulations are very compute intensive so being able to offload the simulations to the cloud or remote servers would also be ideal.</p><p>Much like VSCode and other editors have nice plugins for offloading tests to other computers, a modern CAD program should have the same.</p><h3>User Experience and Design</h3><p>After trying numerous different industry CAD programs, most have one thing in common: a user interface that looks like it is from the 90s. It is a bit ironic that a tool used for mechanical design has not considered the design and experience of its user interface. Most CAD programs are in need of a makeover, however there are a couple outliers that do interface design well. <a href="https://www.shapr3d.com/">Shapr3D</a>, an iPad CAD app, has a great design and very intuitive interface. <a href="https://www.sketchup.com/products/sketchup-pro">SketchUp</a> has a more intuitive and beautiful design.</p><p>Additionally, CAD applications need to be native on MacOS, Linux, and Windows. Native applications built for their specific platform perform better than ones built with Electron¹¹ and the like. Especially for a program as graphics heavy as CAD, it is best to utilize the underlying OS graphics mechanisms to get the best performance possible. Today, you can only use CAD programs on the single OS that is supported by that specific program. Additionally, most of them use archaic GUI frameworks that truly show their age.</p><p><a href="https://www.onshape.com/en/">OnShape</a> changed the mold by offering a software-as-a-service (SaaS) CAD program. This allows expensive compute processes to be easily offloaded to the cloud. This was truly revolutionary and a great idea. However, this limits users ability to work offline. In contrast, native apps can work offline but also have the ability to offload workloads to the cloud when you are connected to the network.</p><p>By focusing on having an intuitive design that does not fall into a trap of complexity, both new users and pro users should be productive. Just like I would use vim for a side project and also professional projects, I would expect my CAD tool to work for building a toy for fun, just as well as it would work for a complex project. A lot of this comes down to the interface design and extensibility through plugins.</p><h3>A better tomorrow</h3><p>Each of these aspects needs to be thought through for a new CAD program. To date, not one existing CAD program has solved all of these problems.</p><p>We owe so much of the amazing innovations of computer graphics to brilliant folks like Ivan Sutherland, Pat Hanrahan, Ed Catmull, John Carmack, and so many others. I can only hope some truly revolutionary changes are headed to the world of computer-aided design in the same way that computer graphics pioneers paved the way for rendering, animations, and virtual reality.</p><p>The hardware industry is desperate for a modern way to do mechanical design. A new CAD program built for the modern world would lower the barrier to building hardware, decrease the time of development, and usher in a new era of building!</p><p>[1] <a href="http://images.designworldonline.com.s3.amazonaws.com/CADhistory/Sketchpad_A_Man-Machine_Graphical_Communication_System_Jan63.pdf">http://images.designworldonline.com.s3.amazonaws.com/CADhistory/Sketchpad_A_Man-Machine_Graphical_Communication_System_Jan63.pdf</a></p><p>[2] A neighborhood of a point is a set of points containing that point where one can move some amount in any direction away from that point without leaving the set.</p><p>[3] OpenSCAD, and everything built on CGAL are all single threaded.</p><p>[4] I am sure there is a lot to learn from these codebases. But as someone who has seen many old code bases, this can lead down a dangerous path.</p><p>[5] <a href="https://www.iso.org/committee/54110.html">https://www.iso.org/committee/54110.html</a></p><p>[6] Additionally, you cannot read a step file sequentially, you need to load the whole thing into memory and then resolve it.</p><p>[7] Using “copy” and “paste”.</p><p>[8] You could, but it would be the opposite of fun.</p><p>[9] This is not yet open to the public, but he has talked about it extensively.</p><p>[10] Creo probably has the closest thing to this functionality but lacks an open ecosystem.</p><p>[11] That being said, VSCode is a very nice Electron app.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=981cef7e06b1" width="1" height="1" alt=""><hr><p><a href="https://medium.com/embedded-ventures/mechanical-cad-yesterday-today-and-tomorrow-981cef7e06b1">Mechanical CAD: Yesterday, Today, and Tomorrow</a> was originally published in <a href="https://medium.com/embedded-ventures">EMBEDDED VENTURES</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Lack of Venture Capital Investment in U.S. National Security]]></title>
            <link>https://medium.com/embedded-ventures/the-lack-of-venture-capital-investment-in-u-s-national-security-edc4234da3bd?source=rss----ce419602c934---4</link>
            <guid isPermaLink="false">https://medium.com/p/edc4234da3bd</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[military]]></category>
            <category><![CDATA[government]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[Jordan Noone]]></dc:creator>
            <pubDate>Fri, 19 Feb 2021 17:53:35 GMT</pubDate>
            <atom:updated>2021-02-19T18:43:25.431Z</atom:updated>
            <content:encoded><![CDATA[<p>By <a href="https://www.linkedin.com/in/jordan-noone/">Jordan Noone</a>, Co-Founder &amp; General Partner at <a href="https://embedded.ventures/">Embedded Ventures</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bQDILUvKxFcKrp81zDTNGA.png" /></figure><p>It’s been more than five years since founding <a href="https://www.relativityspace.com/">Relativity Space</a> at the beginning of 2016. During that time, and during the months since stepping out of my operational role there as Chief Technology Officer, I’ve reflected on a number of concerning trends I saw with the direction of venture capital investment in the United States. Seeing these trends was part of my motivation for starting <a href="https://www.embedded.ventures/">Embedded Ventures</a> with my co-founder, <a href="https://www.linkedin.com/in/jennabryant/">Jenna Bryant</a>. (You can read more of our founding story in her blog post: <a href="https://medium.com/embedded-ventures/embedded-ventures-skunkworks-of-deep-tech-venture-capital-b494eb296ad4">Skunkworks of Deep Tech Venture Capital</a>.)</p><p>There is a lack of venture capital investment in United States national security, especially at the early stage. Dual-use technologies, which are technologies having both civilian and military applications, are not receiving adequate funding or receiving the rapid innovation provided by startup talent. Examples of these dual-use technologies include: next generation electronics; advanced and high temperature materials; machine tools; transportation hardware; communications hardware; energy storage, generation and transmission materials; robotics machinery and materials; and space technologies.This funding void is occasionally filled by patient and strategic capital, however this capital often originates from adversaries of the United States.</p><p>In 2006, 55% of United States venture capital investment was in software-based technologies. In 2017, 92% was in software-based technologies.¹ This is partially a result from software startups requiring less capital, having lower technical risk, and providing a path to quicker growth returns. Seeing the opportunity from investing in software startups, the venture capitalists are reducing risk exposure and drawing forward their investment horizons. These software startups are less focused on developing intellectual property, are often consumer-facing, and with rare exceptions, provide zero value to United States national security.</p><p>With such a focus on software startups, many venture capital funds lack the ability to evaluate startups that have their own scientific developments. Without the early sales traction common in software startups, investment decisions rely on an understanding of the technology being developed, a vision for how transformative the developments and their implementation will be, and the ability for the founding team to develop the company, the team, and the tech.</p><p>The majority of venture capital funds will pass on the opportunities to even evaluate these startups, due to their lack of talent and confidence outside of software. Investment from these funds may come once market timing is confirmed through sales traction, however this traction is often late in the development cycle for hardware startups, significantly limiting the sources of capital available for to support their early growth.</p><p>Unfortunately, there is also a decline in research and development funded directly by the Federal Government. Over 50 years, we have seen federally funded research and development drop from a peak of 2% of gross domestic product in 1964, to less than 1% of gross domestic product today. This is the opposite trend of what has been seen by our international partners and competition.²</p><p>Additionally, corporate research and development has succumbed to short-term shareholder pressures. The iconic corporate labs that developed the base of our modern infrastructure, such as Bell Labs and IBM Research, no longer exist. These organizations, responsible for translating basic science to corporate applications, fell out of favor. Instead, as a more relaxed antitrust environment emerged in the 1980s, corporations began looking at mergers and acquisitions for their technology needs, beginning the rapid demise of the corporate research labs.³</p><p>The breakthroughs developed by the corporate research labs provided centralized, general purpose, practical research and solutions. This is quite different than the subdivided research, with specific purpose, that is done by smaller groups and startups. A healthy innovation ecosystem requires both, and the United States is currently missing the corporate research labs. (The demise of the corporate research lab and the impact on country-wide innovation will be the topic of a future post.)</p><p>With the decrease in federal research and development funding, and the demise of the corporate research lab, there is an opportunity for early-stage startups to carry the burden of fundamental scientific developments. We define these startups as “deep tech” startups, and created Embedded Ventures to support them.</p><p>The majority of deep tech startups can provide either direct or indirect value to United States national security. This can be achieved through a variety of means, including: providing technology solutions to the Federal Government, moving forward the state-of-the-art in their specific field, or expanding the talent ecosystem developing dual-use technologies, among others.</p><p>Embedded Ventures is uniquely positioned with our conviction towards this mission, and our team of engineers, scientists, founders, and startup operators to find, mentor and grow these startups in a way that helps protect the freedoms we enjoy.</p><p>In closing, two quotes I enjoy that highlight both the importance, and our ethical perspective, on investing in national security and the future of our country:</p><p><em>″​The core mission of our company always was to make the West, especially America, the strongest in the world, the strongest it’s ever been, for the sake of global peace and prosperity…” — Alex Karp, Palantir</em></p><p><em>“We made a principled decision that we’re not going to withhold technology from institutions that we have elected in democracies to protect the freedoms we enjoy…” — Satya Nadella, Microsoft</em></p><p>¹ Pitchbook Data.</p><p>² American Association for the Advancement in Science, Historical Trends in Federal R&amp;D</p><p>³ The changing structure of American innovation: Some cautionary remarks for economic growth</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=edc4234da3bd" width="1" height="1" alt=""><hr><p><a href="https://medium.com/embedded-ventures/the-lack-of-venture-capital-investment-in-u-s-national-security-edc4234da3bd">The Lack of Venture Capital Investment in U.S. National Security</a> was originally published in <a href="https://medium.com/embedded-ventures">EMBEDDED VENTURES</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Embedded Ventures: Skunkworks of deep tech venture capital]]></title>
            <link>https://medium.com/embedded-ventures/embedded-ventures-skunkworks-of-deep-tech-venture-capital-b494eb296ad4?source=rss----ce419602c934---4</link>
            <guid isPermaLink="false">https://medium.com/p/b494eb296ad4</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[diversity]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[embedded-ventures]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Jenna Bryant]]></dc:creator>
            <pubDate>Tue, 02 Feb 2021 20:01:31 GMT</pubDate>
            <atom:updated>2021-02-02T20:01:31.734Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*U3NH0qdfzI6LcVGtjnmvMw.png" /><figcaption>Embedded Ventures …The VC Remix</figcaption></figure><p>By <a href="https://www.linkedin.com/in/jennabryant/">Jenna Bryant</a>, Co-Founder &amp; CEO/General Partner at <a href="https://embedded.ventures/">Embedded Ventures</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LJom53HwMlsv1lP5IJJjAg.png" /><figcaption>Jenna Bryant &amp; Jordan Noone</figcaption></figure><p>You might be more likely to see a triple Jovian Eclipse on Jupiter than you are to see two fund managers like <a href="https://www.linkedin.com/in/jordan-noone/">Jordan Noone</a> and <a href="https://www.linkedin.com/in/jennabryant/">me</a>. A woman who at 22 years of age launched a career of pairing early-stage startups with the most in-demand engineers in the world, while simultaneously appearing on TV shows like <em>Sleepy Hollow</em> and <em>The Detour</em>. And an engineer who at 22 became the technical Co-Founder of a future unicorn hard tech startup that was building the world’s first 3D printed rocket, <a href="https://www.relativityspace.com/"><em>Relativity Space</em></a>. We’ve experienced the magic that happens when you’re surrounded by skeptics, but you still roll your sleeves up and turn an idea from nothing to something. Our atypical and serendipitous industry experiences combined with our passion for the communities we represent, led us to the launch of <a href="https://embedded.ventures/"><strong><em>Embedded Ventures</em></strong></a> — a new option for early-stage complex tech startups seeking capital…aka <em>The VC Remix</em>.</p><p>A positive result of the “impostor syndrome” that I have grown to embrace is the ability to zoom out at any moment and identify what legacy processes need to be updated. A few unrelenting observations stand out (and my stubbornness/loyalty to the pursuit of fixing them) as questions we wanted to address as we crafted what mark Embedded Ventures would make on our industry:</p><p><strong>1)</strong> <em>Why is the industry that helps drive innovation so far behind when it comes to diversity? Even the data proves that diverse teams perform better… so what’s the hold up?</em></p><p><strong>2)</strong> <em>Early-stage, hard tech investing is about predicting how science is applied and turned into the technology of the future, and evaluating the technical competency and leadership qualities of the Founders. This begs the question: Why do most early-stage VC’s have pure finance backgrounds instead of hands-on technical, engineering, and startup experience?</em></p><p><strong>3)</strong> <em>Hard tech startups need resources beyond what investment and networking provides. They need expensive, scarce physical resources like lab space, rocket launch pads, and manufacturing facilities. So why is there limited or no collaboration between early-stage startups and large corporations who can grant access to these resources and possibly even be the startup’s first customer?</em></p><p>The Venture Capital industry’s claim to fame is seeing trends first. This might be true, but just for technology trends. Despite decades of diversity, equity, and inclusion progress in the United States, most Venture Capital funds still have the diversity makeup that funds did fifty years ago — NONE. One root of this archaic makeup is how the industry wrongfully prides itself on “warm introductions only,” limiting the growth of their network to individuals two degrees of separation away. What happens when someone outside of that network has an amazing idea for a company or has a unique way of thinking that would lead to lucrative investment returns? They get left behind.</p><p>The majority of Venture Capital firms rarely, if ever, invest in early-stage hard tech. The areas we consider falling under hard tech are <strong>advanced/edge compute, quantum systems, defense &amp; security, communication infrastructure, logistics/supply chain, mining, battery/power systems, material science, advanced peripherals, aerial mobility, dual-use tech, drone technology, manufacturing, robotics, biotech/biochem and sustainability</strong>. Most pass on these startups at the early stage, hoping that someone else will invest in them. They’ll participate once there is industry hype or customer traction. Luckily, a few brave early-stage firms have taken chances on this more complex group of technologies, even with limited hands-on technical depth. They often utilize their existing network to evaluate the technical depth of these founders and help determine if technology in this space makes sense. What happens when you put deep tech founders and technologists in the driver seat to evaluate these early investment opportunities, instead of bankers? You understand the potential of each hard tech startup years ahead of others.</p><p>Building a hard tech startup is extremely different than building a SaaS company. In addition to having more visible market timing indicators, there is a thriving ecosystem of low barrier-to-entry infrastructure partners for building software companies — the highlight being cloud computing services like AWS, Azure, and Google Cloud. There is no such equivalent for hardware. In some cases, there are only a handful of organizations that have the infrastructure you need and you can’t afford to replicate it (rocket launch pads, for example)! We set out to tie together corporations and startups in a way that is underserved by today’s options.</p><p>Our early benefactors are interested in keeping our operations under the radar. We are not planning on publicizing much more than this with you at this point, hence the “skunkworks” semi-secrecy of the majority of our operations.</p><p>Jordan and I realize how uniquely positioned we are, between our personal experience and talented community, to help shrink the gap between idea and action for changing these legacy systems. We can’t do this alone, and we are so thankful to be surrounded by some of the most well-respected leaders from every part of the industry as Advisors. We will all work together to break the current ideologically heterogeneous constituencies.</p><p>This is a mammoth opportunity, and we know it will take time to see the results of our work. I also assume that the industry won’t change as visibly as I’d like it in my lifetime. No matter what, we will do our absolute best to pave the way for this next generation of VC firms through Embedded Ventures, and will be a guide to those who follow.</p><p><strong><em>…I love that with this audience, I don’t need to say it louder for the people in the back!</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b494eb296ad4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/embedded-ventures/embedded-ventures-skunkworks-of-deep-tech-venture-capital-b494eb296ad4">Embedded Ventures: Skunkworks of deep tech venture capital</a> was originally published in <a href="https://medium.com/embedded-ventures">EMBEDDED VENTURES</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>