Transcript: Vijay Pande at Splash Health 2016

Splash Talk Vijay Pande (General Partner, Andreessen Horowitz)

At Vator Splash Health 2016, Vijay Pande (General Partner, Andreesen Horowitz) shared his presentation: "When software eats bio." Here's a transcript as well as the video version. 

[Introductory music]

Pande: So, I’m excited to talk to you about some of the recent events that’s been going on in Andreessen Horowitz and our interests in the intersection between software and biology. So, first, a little bit about me. So, I’ve spent the last 15 years as a professor of Stanford and also an entrepreneur in the Valley. And I’ve had the joy and luxury to be able to geek out on a wide range of topics, from Computer Science to Biology, to Physics and to Chemistry.

But then, really, the parts that I’ve been very passionate about is the intersection between what software and Computer Science can do in Biology and Health Care. One of the projects that I pioneered at Stanford was the Folding@home distributed computing project. With Folding@home, the idea was that when we started over 15 years ago, in October of 2000. The idea was that computers can have a huge impact in Biology; but really, only if we have enough computer power, that there was so much promise and so much excitement, but really limited by the available computing power at the time.

And so, what we did then, is we actually got people to donate their computer power. Actually, we created the most powerful super-computer in the world. And actually, even in 2007, getting a Guinness World Record for the first computer which had a petaflop. In many ways, the exciting thing about Folding@home for me was that we were able to anticipate what would be interesting trends in the future, because what was very doable for us then, but impossible otherwise, now becomes really quite routine, as Moore’s Law continues on.

And also, on the entrepreneurial side, I’ve been involved in several startups, one of the more recent ones, Globavir BioSciences, takes a lot of these concepts and applies it to therapeutics and infectious disease and imuno-oncology where we use computation to be able to very quickly identify new compounds, especially for drug repurposing and to move within months or under a year from having nothing to something that’s ready to go to Phase-II clinical trials.

What I want to talk to you about today though, is a particular trend that we’ve been observing, which we call Bio 2.0. This trend is comprised of a confluence of many things that come together, really at this very unique point in time. So first off is a trend I was alluding to before which is Moore’s Law. I think everyone is familiar with Moore’s Law, the fact that the cost to compute has been exponentially decreasing over time.

What’s intriguing about exponential decreases is that they really catch us off guard, that ten years of exponential decrease can turn what used to be extraordinary into something that really is ordinary or commonplace or even free. So to put in perspective, for example at Folding@home, we got this Guinness World Record in 2007 for the most powerful computer in the world. And now, that same amount of compute power is just $300 a day at Amazon. That’s really the amazing thing; eight years, and it goes from being sort of world class record breaking to something that is achievable by everybody.

What’s interesting is that there’s not just one sort of Moore’s Law type of law. In a sense, it's sort of a family of laws. It’s not just that the cost to compute is going to zero, but also other things, like the cost of storage is going to zero and also the cost of mobile is going to zero. Right now, in China, there are mobile phones that cost basically $20 that are as comparable in power to the original iPhone. The interesting thing about this is when we think about mobile, this is a huge infrastructure that didn’t exist before, essentially a little supercomputer that’s in everyone’s pockets right now that’s connected. And we’ll talk about that it enables really new types of opportunities and new companies will arise from this.

Finally, one aspect that’s maybe a little under-appreciated is that the cost of mobile going to zero has another implication, which is that there are all these associated parts with mobile, that are part of the mobile supply chain, that also come along the ride for free, so you know all the cameras, the motion sensors and all those types of things. These things can be put in other types of devices. What we’re seeing is a lot of interesting hardware comes about by taking advantage of the mobile supply change, things that are already there and just slicing and dicing them in new ways.

What’s intriguing about this is I’ve been talking about Moore’s Law and we might be tempted to say, “This is just a computer thing, right? This is not really related to Biology.” What I think is intriguing is that there are biological equivalences to Moore’s Law. For instance, the cost of genome sequencing has been dropping dramatically. Actually, the interesting thing is it’s not exactly dropping like Moore’s Law. It’s actually dropping to zero faster than Moore’s Law.

What does this mean? Well it means that the extraordinary is now cheap. The first human genome is $3 billion in 1997. Nowadays, the cost of goods is probably $300 and I think we’re going to see that moving to $30. And so again, what used to be like mind blowing and amazing, like we go into this once, becomes everyday, cheap, free.

It’s interesting  to think what happens when we put these things together, the cost to compute going to zero, the cost of mobile, the cost of storage going to zero, the sensors, all the stuff in the supply chain, and also, the fact that in many areas of Biology, the cost of these interesting Biology areas are going to zero. You put this together and this is a confluence of events that has the opportunity to disrupt certain aspects of traditional Biotech.

I want to make it clear that it’s not my claim that this will disrupt everything in Biotech and there’re many things that are very challenging and difficult problems in Biotech, but there’s an interesting opportunity here that I’ll elaborate in the rest of my talk.

And so, when we start to see what happens with these macro-trends, we starting to see startups that have a certain composition. First off, is that we’re seeing a lot of startups that are in the Bo-healthcare space, but the key thing is that these are actually intrinsically software companies. They have software at their heart. That often means that they have interesting machine learning and cloud computing and they operate like software companies.

A second area which I’ll elaborate on in a little bit is the area of what I call Cloud Biology. And so, what Cloud Biology is, is the opportunity to do real life Biology experiments, but in the way that we typically think about calculations in the Cloud. Let me put it this way. Think about the early days of startups and software. You’d have to build up this huge server farm which costs maybe $20 million or something like that before you can do anything at scale. And now, with Cloud Computing, you can actually go to Amazon or Google Cloud, and with basically a laptop, electricity, wi-fi and a credit card, you can build a product and build up to scale.

The intriguing thing is that it’s not just that you don’t have to build out this $20 million server farm before you know that this works. You actually also have the elasticity that you can use just as you need it. And also, a very light touch that the fact that you can do this with a laptop and a credit card means that you don’t have to go through a complicated business depth procedure or anything like that.

Compare that to Biology right now. You could outsource Biology and Chemistry as you would in a Biotech regime. Right now, before Cloud Biology, this was a very high touch operation. It’s not a laptop and a credit card operation. What Cloud Biology does is that it creates something analogous to Cloud Computing, but instead of computers in the cloud, these are typically robots or sensors or other types of devices that run real life Biology experiments. And so, what’s intriguing about this is that we would get the same types of properties that you’d get from Cloud Computing, and for computing, as you would get for Cloud Biology or for Biology.

And finally, another area that comes up over and over again in these new startups is new approaches to handling FDA risk. Traditional biotech has to go through a traditional path of dealing with clinical trials and so on. We’re seeing a lot of companies that are taking alternate approaches, whether that means going after non-clinical indications, the clinical associations or other approaches. I think handling the FDA in different ways in an aspect of this trend that we’re seeing amongst these companies.

And so, in a sense, I think what then you look at all the parts together, to me, I see this picture, Biology today reminds me of software in 2005, that the opportunities are there and what we expect to see over the next ten years is an explosion akin to the explosion we saw in software 10 years ago. Okay, so that’s where the backstory and the big picture trend. Now, what’s interesting is what does this mean? What would you expect to see? There’s a couple of implications.

First off, is that machine learning is starting to become very pervasive. A great example of this is deep learning. The issue with machine learning before was that, before something like deep learning is that, machine learning was held hostage to one’s ability to come up with the right features. In a sense, you almost have to know the answer to the problem before you can have machine learning solve the problem for you. One of the seminal papers by Jeff Dean and Andrew Ang and co-workers was applying machine learning, especially deep learning, to YouTube videos.

What happens in deep learning is that we have this complicated sets of one neural net feeding into another, feeding into another and when you put YouTube videos at the left hand here, what happens is that along the way, you start building up features such that the human being doesn’t have to come up with the answer to featurize things. The deep learning algorithm will do it by itself. So the case of videos, there’d be like low-level geometric features, but the intriguing thing is that other things pop up automatically, like this face, there’s a lot of faces in YouTube videos, of course. This becomes something that’s automatically identified. What’s intriguing, of course, is since this is YouTube videos, we know there’s really one primary purpose for all videos on the internet, which is cats. And so, of course, the cats show up as well as they must. Okay, but this is YouTube videos into deep learning. What does this look like for Biology?

For Biology, you can imagine, let’s say in the case of drug design, maybe the low-level things are deciding what would be drug like or not drug like and higher order things would be whether it’s a GPCR-agonist inhibitor or so on and this could be anything. This could even be, let’s say, applying radiology and being able to tell whether this is a tumor or not, or looking at pathology. Anything that’s either visual or data heavy, genomics for example, are very natural approaches to be put into these types of processes. 

And so, what’s emerging is, due to all the advances in machine learning that come from the fact that we finally have the compute power, and you can argue that this architecture was first invented 20 years ago to some extent, but what we didn’t have was the computing power or the storage to be able to really take advantage of this. I think we’re in a unique period of time due to these macrotrends that this becomes viable. It’s becoming viable in areas outside of Health and Biology, but it’s a very natural pairing for Health and Biology where there’s such a magnitude of data and a great need.

There’s a second aspect that’s starting to emerge and I don’t know if you guys remember this commercial from maybe a decade or more ago where the Resee’s Peanut Butter Cup, “you've got your peanut butter mixed with my chocolate.” In my mind here, the peanut butter are the people working in Biology and Healthcare and the chocolate’s the computer scientists. If you prefer, you can think of it the other way around. It still works.

The idea is that we see some companies where the two are separate, where the Biology guys would come up with the genomics and hand it over to Computer Science guys. The successful companies that we’re seeing these days are ones where the founders can really go full stack in the Biology and Computer Science. This is actually a very different situation than it was 20 years ago. I think the biologists are very familiar with Computer Science in a way that they really weren’t 20 years ago and that the computer scientists have almost like a kinship and a draw to Biology, almost like Biology is the programming nature of sorts.

And so, this ability to go full stack I think is present in the companies that we see that really can have the biggest impact. The companies that seem to have a sort of more average impact are the ones that sort of take the genomics out of the box, machine learning out of the box and run it. Those are the companies I think will do well, but not the ones that will be growing big.

And finally, I think about a part that intrigues us and gets us excited, is the idea that we can do things finally from a much more rational point of view. And again, this might not be true for everything immediately, but as time goes on, there’ll be much more engineering versus empiricism.  And so, a great example here is the Bay Bridge. For those of you that don’t live in the Bay Area, this is a bridge that’s relatively new. It was built due to earthquake issues. You can see the old bridge right behind it. When this bridge was built, it costs pretty much the same amount as what a drug would cost, about $3 billion. You can imagine, if we didn’t understand how to engineer bridges, it would be kind of a scene disaster. Imagine like if we had some idea and we built a bridge, and then we’d have to do clinical trials on the bridge. We would send mice over the bridge to see if it’s safe first and then we’d send people who really want to get to Oakland to go over the bridge.

Audience : [laughs]

Pande: And then maybe, it would get pass regulations and then people will go over the bridge. So it’s ridiculous, right? That same way that that seems absurd is the way that I think 10 years from now or 20 years from now that many aspects of the way we do things now will seem so antiquated. That’s because we finally have the power, both in terms of computers and algorithms and data to be able to tackle this in a much more engineering-like approach.

Okay, let’s summarize this. Maybe one way to summarize this is from a table that’s actually from Peter Thiel’s book, Zero to One. He spent some time talking about why he loves software startups and hates biotech startups. He makes a pretty powerful case, that biotech startups involve uncontrollable organisms and software startups involve just code which can be perfectly understood, issues of being heavily regulated versus unregulated, expensive versus cheap and so on.

But what we’re starting to see is that these new generation of bio and healthcare startups really look like software startups. In all the ways that we love software startups. These startups, even though they are in the Bio and Healthcare space, work like software startups. That, I think, is part of the great opportunity.

So finally,  this is interesting compared to the sort of traditional biotech approach.  I don’t know how many of you are familiar with the term “Eroom’s Law”. Anyone? Some of you. For those of you who haven’t heard of Eroom’s Law, this was something that came up in nature a few years ago. “Eroom” is the word “Moore”, as in Gordon Moore, spelled backwards. The reason why it’s Moore spelled backwards is that it’s Moore’s Law backwards. That the cost of the drug, instead of exponentially decreasing over time is exponentially increasing over time. This is something that’s been born our for several decades now.

It’s kind of clear why this is, in a sense that, just like I mentioned, a sibling to Moore’s Law, there’s such in storage and bandwidth and compute and genomics, there’s siblings to Eroom’s Law, like the cost of college tuition is also exponentially increasing. In a sense, the reason why Eroom’s Law exists is basically [for] people like us, well-paid professionals that are highly skilled, highly trained and therefore, very expensive and increasing at this exponential rate. And so, when we look at the companies that we're excited about this Bio 2.0 trend, in a nutshell, (which is maybe the crudest way to put it), Eroom’s Law and everything associated with that are things that I know what the future of that looks like. I know what time goes to infinity looks like there. And it looks very expensive.

In Moore’s Law, I also know what this looks like. And so, we’re really especially looking for companies that live in the Eroom’s Law space, in terms of what their applications are, but that can take advantage of the Moore’s Law curves. Even if those companies are break even with standard technology today, it’s very clear what the futures of those two companies would be.

Okay, I’ve been talking sort of very abstractly in some ways and in the remaining time, I want to talk about specifics that sort of put the meat around this and to go over some case studies. There are three areas that come to mind to demonstrate this thesis. First is in terms of traditional therapeutics; we are really all familiar with these issues, that they are very slow to develop, they’re expensive, you have to deal with toxicity and are naturally highly regulated because of those issues. What’s emerging now is a new type of therapeutic, so called “digital therapeutic” involved with digital health.

And so here, there’s a portfolio of companies of ours on health which I think is pioneering the area of digital therapies and digital health for pre-diabetics. The idea here is that there are many issues in modern medicine that modern medicine does well. Such as if I have a bacterial infection, if I didn’t take an antibiotic, maybe I’m dead in two weeks, if it’s a serious infection. I take a pill and I’m magically better in two weeks. That’s the power of modern medicine, which is really wonderful. That approach works for some things, but it really doesn’t work for others. I think what we’ll find, when we think about this maybe 20 years in the future looking back, is that there are many areas for which the pill solution, at least, shouldn’t be Plan A.

These are the areas of behavioral solutions, issues like pre-diabetics, depression, anxiety, sleeping disorders. You can imagine the standard medical model works really well for something like an antibiotic. You can have a great mouse model for an antibiotic. I don’t think you can have a great mouse model for PTSD, or for anxiety, or for even necessarily for Type-2 diabetes, because the way we have Type-2 diabetes would very different from the way a mouse would have it. And so, what’s intriguing is that this whole area of digital health takes existing therapies that are often found in places like Stanford or UCSF, that work, and these are behavioral therapies, but they are typically very expensive and available only to a few people.

Like the Stanford Sleep Center, it does amazing things, but maybe 20-30 people are able to access it a year and it costs $5,000 a year. What companies like Amana does it the area of digital health is they take an existing behavioral therapy and they allow it to scale. Instead of tens of people, this in principle, could be tens of millions, or really hundreds of millions, instead of thousands of dollars, this becomes hundreds of dollars.

So, how is this possible? It’s possible because of the huge built-in infrastructure we have with mobile already. In a sense, so much of this issue is, in a sense, could be solved just by saying, your doctor gives you the talk when you’re getting towards Type-2 diabetes , you should exercise and you should eat better, that should be the cure to Type-2 diabetes. But that really doesn’t work, at least for most people. And so, with these behavioral therapies, due to a combination of a tightly-knit social, and coaches, and email, and analytics, including things like scales, Amana’s able to come up with a full package that really transforms ones behavior using this built-in resources. Amana’s not the only example. I think we’re seeing many companies in this space in areas of anxiety, depression, PTSD, sleep, and so on.

The second aspect of these companies which I think are intriguing is that they have not just the claim that this could help, but because they can monitor things quantitatively, they can do the equivalent of things like clinical trials. So you can compare a placebo, which is your doctor talking to you, versus a leading drug in the space for Type-2 diabetes, Metformin, versus the lifestyle effect. And actually, if you look at which ones works best, which is closest to zero, least amount of number getting the issues, this lifestyle approach and monitor approach can actually have greater efficacy than the leading drug in the space. I think that’s the hope here. The hope is not just that it’s comparable to a drug in efficacy, but it can even exceed the drug in efficacy. On top of that, obviously, this doesn’t have the toxicity issue that a drug would have.

Now going down the line, you can imagine very interesting things where you can combine digital therapeutic with a small molecule therapeutic for synergistic effects and new types of cocktails. That’s more in the future, but even today, I think my recommendation would be something like this; would be, the Plan A and the drug be the Plan B. That’s one area of digital therapeutic and digital health.

A second area that we’re seeing is the area of Cloud Biology, and it’s interesting to compare Cloud Biology and traditional Biology. If you went to any lab, like in Berkeley or UCSF, or Stanford today, and you went to a top lab in Biology, it'd look pretty much like this: it’s like teams and teams of people working with their hands, pipetting and so on. If you think about it, it’s almost like a scene out of pre-Industrial Revolution times, except for the computers and stuff, but in terms of the manual labor.

There’re reasons why manual labor has challenges. It leads to all these irreproducibility problems that we see. This is actually a true story from one of my colleagues that I was talking to at Stanford a few months ago is that, they were trying to track down an experiment that was extremely variable. They didn’t know why sometimes it worked, and sometimes it didn’t work. They were like going crazy, they were pulling out their hair for what it could be. They started recording literally everything, what people were wearing, even the dyes on lab coats can vary and can lead to differences. They tracked it down to particular Tuesdays when the grad students had tuna fish sandwiches for lunch and the amides from the tuna fish sandwich was getting into the material and causing some contamination. 

Something like that is sort of insane to imagine tracking down, but is a very common thing. You could imagine comparing that to the area of Cloud Biology, where robots are driving everything, and where the research is not about pipetting, it’s about programming. You literally write a program that gets executed in the robots to do the experiment. Reproducibility looks like re-running the program. And in this sense, can be completely perfectly reproducible. The counsels that companies I’ve been an advisor of and they have a large Cloud Biology backend, but there are several other companies like Emerald Therapeutics, Transcriptic, and Malserra, they’re applying these technologies to be able to make this available broadly to other people.

Okay so finally, an area that I want to highlight is an area where traditional medicine is getting overloaded with data. This could be in radiology, dermatology, and genomics, in cancer therapeutics. Really, it’s this data challenge that is a huge challenge for a human beings, but is actually something very natural for data science and machine learning. One of my favorite examples of this is in cancer, that often, we say that researchers around the world are looking for a cure to cancer, but ironically, that’s actually completely false in the sense that the problem isn’t that we don’t have a cure to cancer. Often, the problem is that we have many cures to cancer and the challenge is what is the right drug, for the right patient, at the right time?

Often, even if you’ve got the right drug in the beginning, the tumor can metastasize or change and go through mutations and suddenly, that’s no longer the right therapeutic. And so, this challenge of matching drug to patient is a very difficult one for an oncologist to do. Even amongst the best oncologists; it’s a very natural one for machine learning and genomics. In this space, we’re starting to see a large combinations of companies that involve a combination of deep machine learning with genomics, such that they can be able to understand individual patients in ways that were really not weren't possible before.

Maybe one last point about that is that, one thing that I think is poorly appreciated, is that the quality of care throughout the country, for something like oncology, differs very greatly, whether they’re in a landmark place, or a small clinic. And the ability to drive this computationally isn’t just increasing the care or decreasing the cost, but it’s also bringing the highest quality care to everybody.

Okay, so finally, let me wrap up and talk about what are the implications for this, and what does this all mean. There’s one last sort of Moore’s Law, sort of sibling, but I think is interesting and while it’s not completely, directly related to what I’m talking about, I think it makes an interesting point. This last one I wanted to talk about is actually the adoption of solar energy in the world. This is an example of something that’s sneaking up on us. For instance, the amount of solar power that’s used in the world or a fraction of solar power in the world has been increasing exponentially for several decades and doubles every seven years. It’s basically five percent right now.

It’s intriguing what 20 years from now will look like. 20 years from now, we’ll be basically three doublings, so a factor of eight. And so, eight times five is 40, so roughly, half of the world power will be provided by solar energy, if this trend continues. I think there’s no reason to think that it wouldn’t. Thinking about the impact of that right now, like the cost of oil is decreasing, just due to slightly less use of oil. It’s interesting to consider that it sounds like science fiction, but it’s not possible to imagine that 20 years from now, the cost of energy itself will also go down to zero, or certainly start its way there.

That just feels like a science fiction statement, that certainly remains to be proven. But it’s an example of how transformative, and exponential, the cost can be just for 20 years, how much the world can change. It doesn’t even have to be 20 years; it could be 10 years. My favorite example is that I’m very much addicted to is using my phone to everything that I do. I don’t think I can do my job without it. I think that’s true for many of us. I couldn’t have gotten here. I came over in a Lift and I couldn’t have gotten here. I couldn’t have done my email or anything. I can’t even use two factor multiplication without it.

It’s really something that has snuck up on us. It’s only been less than ten years since the iPhone was even developed. And so, what I’m intrigued about, is due to this confluence events, the cost of computers, the cost of mobile, the cost of storage, and cost of genomics. It will really create a different world that will sneak up on us very quickly, and that the companies that can take advantage of these Moore’s Law curves, and switch things from Eroom’s Law over to Moore’s Law, will be the ones that are shaping that future.

With that, I’d like to thank you and I think we have three minutes left for questions. Thank you very much.

Audience: [claps]

Host: You want questions? Q & A?

Pande: Yep, please.

Host: Anybody out there? Got to have a couple of questions. Anybody? Here we go.

Audience member: Thinking about your time at Stanford as a professor and then moving into the space you're in now, you mentioned Eroom’s Law, and education being one. I understand Google’s going to come out with a PhD program. How do you see education? Trying to educate and identify new engineers that are also geneticists and combining those? And who’s going to survive that?

Pande: Yeah, it’s intriguing to think about these sets of things. I think, the challenge for education is two-fold. It’s education versus training, and so in my mind, a true education is learning how to learn. If you can do that, you can be a computer scientist and you can pick up a Biology textbook, or you can be a biologist and you can pick up a Bazian statistic textbook and you can start to absorb it. But then, also, there is the training aspect just because once they get out they have to be able to dive into things.

And so, I fully expect that we’re going to see a lot of alternatives to the university system for a variety of reasons. I think that’s actually very exciting. If you think about it, right now, there’s sort of a limit of what universities can do. Stanford can admit thousands of undergrads, but can’t admit millions of undergrads. What can we do to scale these things? MOOCs were an attempt to play a role in this, but it’s interesting what else we can do, especially since Stanford can educate thousands a year when there’s millions to educate and that’s the challenge.

Audience member: [indiscernible 00:28:51] what do you see the lab of future looking like?

Host: Yeah.

Audience member: We work in trying to find a lab as a base for [indiscernible 00:28:57] wet labs, what do you see the labs’ role [indiscernible 00:29:06]?

Pande: The question is what does the lab of the future look like and I think there’ll be much  more automation. We’re seeing that already just as a general trend. And automation's both in terms of cost, but also, I think the reproducibility issue is, I think poorly appreciated. Basically, if, and this has been highlighted in various journals, most Biology coming out of studies are not reproducible. There’s a variety of reasons for that, some are statistical such as a low P-value threshold. But even beyond that, it’s just very difficult to do these experiments, and the ability to have it automated really just takes that all away and allows us to share protocols and share code. You can imagine, just as code sharing allows software to sort of be more than sum of the parts, and to extend for people copy, paste and edit and modify, I can imagine we’ll see that as well.

Now, it won’t be something where it’ll be hard to completely automate all of it. Even a small degree of automation for the more repetitive tasks and challenging tasks, that alone probably like 30 percent automation can handle 90 percent of the challenges.

Host: One more question.

Jackson: Deshawn Jackson from the Entrepreneurship Center of UCSF. I just have a quick question about how does this translate into leveling the playing field for when you’re trying to side for these tech startups or these software companies, what do you think that this computational power, how do you think it translates to dollars for startups getting off the ground quicker and faster?

Pande: Yeah, I think where this translates is the idea that right now, in the software side, we could give a bunch of smart grad students or young entrepreneurs $3 million and they can use that with cloud computing and so on and get to a product. And then, at later stages of investment, we can assess like how they are selling, how they’re building up the sales and basically, at each stage, we can think about funding based on data-driven metrics of their performance. It’s very different from biotech where you only have revenue at the very end and it’s very difficult to de-risk in those ways.

I think this levels the playing field in the sense that we’re going to see better startups that can be invested in and raise money like these software companies where along the way, they’ll be able to move much more quickly to product and proof of concept and even potentially to revenue very early. That’s something that would be very key to the health of the company, but also to the de-risking of the company going forward.

Host: Great. All right, thank you very much.

Pande: Thank you.

Audience: [claps]

[END]

Thanks to our main sponsors KPMG, Javelin Venture Partners, SAP Startup Focus, Bread and Butter Wine, Artis Coffee, Scrubbed, Stratpoint and Healthiest. 


Editor's Note: Our annual Vator Splash Spring 2016 conference is around the corner on May 12, 2016 at the historic Scottish Rite Center in Oakland. Speakers include  Nigel Eccles (CEO & Co-founder, FanDuel, Andy Dunn (CEO, Bonobos), Libby Schaaf (Mayor, City of Oakland), Mitch Kapor (Founder, Kapor Center for Social Impact), Oisin Hanrahan (Co-Founder & CEO, Handy), and more. Join us! REGISTER HERE.