00:00welcome to the a 16z podcast I'm Michael
00:03Copeland and we are here in the room
00:05with three people to talk about
00:07computational biology and how compute
00:10meets biology and and generally speaking
00:13how healthcare can get better through
00:14through means of technology and to help
00:16us do that we have a 16 zsv J Pandey we
00:19have Jeff Kindler who is the former CEO
00:21of a little company you might have heard
00:22of Pfizer and who's now an advisor and
00:26investor in in the biotech space and
00:28finally Andrew Raiden who's the CEO of
00:292za which is a portfolio company of ours
00:32which is in the computational biology
00:34space not surprisingly so welcome guys
00:36it's great thank you we have seen
00:38technology and software in particular
00:40you know from our vantage point seep
00:42into all kinds of industries right so
00:45bursting it in finance we're seeing it
00:47in driverless cars and what there's been
00:50this promise in the healthcare space
00:52that technology will come in and revamp
00:56all these kind of very expensive very
00:59time consuming processes to help us get
01:02to better health in the end but better
01:04therapeutics better you know testing for
01:06that matter maybe Jeff let's start with
01:08you why is this problem so hard and how
01:12you you've been in this space for a long
01:14time and kind of seen it from all sides
01:16where are we headed now and you know how
01:19do we get there sure well first if I
01:21could just go back and give a little
01:22analogy to another industry that I think
01:24is appropriate so if you think of
01:26Hollywood the movie business and the
01:281930s and 40s the studio's had everybody
01:31as an employee of the studio the crew
01:34the cast the writers the directors and
01:36they just cranked out movies today the
01:39movie business is one in which the
01:41studio's act as syndicators financers
01:43phonate Sears commercial oriented
01:46organizations and they put together
01:48every movie as a package with people
01:51that don't necessarily work for them I
01:52think Pharma is in the process of making
01:54a similar shift for many years until
01:58even relatively recently they were one
02:00of the full one of the few fully
02:02integrated industries in the sense that
02:05they did everything in-house from
02:06Discovery through the end of a product's
02:08life cycle over the years starting with
02:10clinical research organizations they
02:13this they have come to realize that some
02:16of the things that they think of as cost
02:17centers are better done by people on the
02:20outside who are doing it for profit and
02:23therefore probably do it better more
02:24efficiently and so forth I think it to
02:26answer your question directly Michael
02:28why hasn't it happened sooner I think to
02:30a large extent it's because the farm
02:31industry has been so profitable and so
02:33successful and is so has been and will
02:36always be so concerned with control and
02:39quality and the rest that they probably
02:41come reluctantly to outsourcing over the
02:43years because it just wasn't consistent
02:45with the way they did things but what's
02:46happened in the last few years is
02:47they've come under greater earnings
02:49pressure they're really looking at their
02:50balance sheet or income statement and
02:52they're coming to realize that there's a
02:53lot of costs in their system that can be
02:55driven out and services provide more
02:58appropriately and efficiently by people
03:00on the outside Andrew that gets right to
03:02you I mean you are one of those folks
03:05yeah and and here's what's interesting
03:07too is that you come to this not from a
03:10medical background yeah or chemistry for
03:12them so yeah just just to kind of you
03:14know add a little more color to that
03:15comment our company owns a refrigerator
03:18and microwave that is what we own and
03:20I'm guessing you don't actually use that
03:22for anything medical right and so we
03:26acquire resources on demand and so this
03:28this ability to go out and get millions
03:31if not tens of millions of dollars of
03:33infrastructure whether that's computer
03:35whether that's lab resources use it in
03:37that that moment in that moment might be
03:3910 minutes worth of computation and then
03:43release that resource and don't have to
03:44pay for it again that enables startups
03:46and and small companies to do things to
03:49have all the types of resources that the
03:51big guys have without having to have the
03:53huge bank account so in some sense it's
03:55this unbundling of pharma but it's also
03:58this AWS kind of like look we're just
04:01gonna take the services that we need
04:02when we need them and pay for that and
04:04that's it exactly yeah yeah I think I
04:07just would add to that I think that
04:08there's we're at a great moment in time
04:10because I think that technology and
04:13Veejay can speak better to this than me
04:15but the technology and the capability
04:17from the cloud is so much greater than
04:19it was and that comes right at a moment
04:21in time when Pharma can use it best
04:23there's there's a tremendous need in
04:27standardization and automation of its
04:29processes and historically you know
04:32they're doing some of the things that
04:34they do in the same way they did decades
04:36ago so this is a very propitious moment
04:38to have technology meet the
04:40pharmaceutical business and and help
04:43them both vijay as Jeff pointed out
04:45Pharma hasn't had really change because
04:48the financial pressure certainly wasn't
04:50there and there may be that pressure now
04:54but breathing things back so for example
04:58the cloud and the banking industry the
05:00banking industry is like whoa whoa whoa
05:01why would we ever do that turns out now
05:04the banking history is all up in there
05:06in the cloud what will it take for for
05:09Pharma to get in there in the same way
05:11yeah you know there's a couple of
05:12interesting aspects here first off is
05:14that this isn't something that has to
05:16happen immediately that what will happen
05:18is one will see gradual and gradual
05:20penetration of these services in first
05:23in pilot projects than larger and larger
05:25and we already see this with compute and
05:27forma that farmers using Amazon at first
05:29there was maybe concerns about security
05:31other things and what I've heard
05:33confidentially from people is that AWS
05:36actually has better security than often
05:39the way a lot of people have set up
05:40themselves that's the benefit of them
05:42being an expert in cloud computing that
05:45they can put the resources not just to
05:47have something cheaper but better and on
05:49the cloud biology side where we're not
05:51talking about compute but we're talking
05:52about real life biology in the cloud you
05:54could imagine the same type of thing
05:55that it could be done not just cheaper
05:57and not just lower overhead but actually
05:59done better and I think when something
06:01is cheaper and better then it's actually
06:03a very strong value proposition and
06:06something being very new will mean that
06:08it won't have a mediant adoption but as
06:11people start to see it I think they'll
06:13start to want more more of it so I'm
06:15involved with a company that's still in
06:17stealth but you'll hear a lot about it
06:19and what it's doing is addressing a sort
06:22of small corner and not much spoken
06:24about part of the pharmaceutical
06:25industry which is animal models animal
06:27models today are done pretty much the
06:29way they were done in the 1950s their
06:31very high touch they involve a lot of
06:33human subjectivity in accuracy and
06:35reliability it still involves people
06:37going in and measuring the size of a
06:39rat's foot and that kind of
06:41and this company has digitized that with
06:44the heavy dependence on idea of us as a
06:46matter of fact they have automated it
06:48they have made it more reliable they're
06:50using sensors and cameras and big data
06:53to make that whole system both as vijay
06:55says both much less expensive and much
06:57more reliable that's a that's a good
06:59example of a part of the pharma industry
07:00that was historically done either
07:03in-house or by somewhat commoditized
07:05outside providers and really no
07:07technological innovation for years and I
07:09think that's a precursor of a lot of
07:11I want to get into why there's pressure
07:14in pharma but Vijay you mentioned cloud
07:16biology and Andrew so those two words
07:20don't seem to go together but and just
07:22in the way Jeff that you describe this
07:24kind of like there's actual real animals
07:26but somehow technology interfaces with
07:28them what is cloud biology where are we
07:32today with it and and sort of where do
07:35we head there are several companies in
07:37this space that are pushing the envelope
07:39of cloud biology and um when I think
07:41about it I think of it as doing
07:42real-life experiments so I don't want to
07:44confuse this with doing simulations or
07:46calculations you know these are
07:48real-life experiments whether that be
07:49animal models with mice or in vitro
07:52experiments but the difference here is
07:54that it's something that's completely
07:55systematized for example a lot of
07:57in-vitro experiments can be driven by
07:59robots and what's intriguing here is
08:01that when you drive something by robots
08:03biology now doesn't look like humans
08:06pipetting bench after bench
08:07it looks like program and many of these
08:09companies actually you literally to use
08:11their service you write a computer
08:12program so what makes us better is not
08:16just the fact that it's cheaper but
08:18there's a huge disaster and you know
08:21real challenge in biology right now for
08:23reproducibility that many biology
08:25experiments are just not reproducible
08:27and it's crazy all the different
08:29variables that can come into play but
08:31the beautiful thing about a cloud
08:32biology like setup is that since its
08:34programming on robots you have the best
08:37chance for reproducibility that
08:39rerunning the experiment is rerunning
08:41the code making a small change in
08:43experiments making small change in code
08:45and rerunning at sharing the experiment
08:46with some other colleagues is sharing
08:48the code this actually now brings the
08:50ethos of programming into the biology
08:53realm and that's part of what
08:54we save when we talk about it being
08:55better one of the things that that
08:57fascinates me though is that programming
09:00is programming right code is is code and
09:02yes it can be elegant
09:03but then when you when it meets the
09:05world of biology or chemistry people are
09:08complicated right biology is complicated
09:10how can code account for that yeah
09:13that's an interesting question so I
09:14think there's there's often a
09:16misconception about what software can do
09:19you know and I've often heard people
09:21talk about you know the intelligence of
09:23the machine and somehow it's going to
09:25magically replace you know human inside
09:27and I'm here to tell you as a computer
09:29scientist that's that's not happening at
09:31least not this week right so check back
09:33next week yeah exactly so you know I I
09:35see the software is something that
09:38brings great assistance to the
09:40researcher right and so what the
09:41software can do is look at billions of
09:43data points in an instant and provide
09:45some insights as to what might be
09:47interesting to pursue versus what may
09:48not be and that's very different than
09:51having a human sort of think about the
09:53creative process engage in sort of their
09:55knowledge of biology and sort of
09:57interpret what the computer is providing
09:59to help make an ultimate decision right
10:01so for me software is is it is a tool
10:03it's a very powerful tool it's a way to
10:04accelerate the process and I think also
10:07there's there's a there's a part about
10:08software development that is this very
10:11iterative approach right and so when
10:13software developers go and build
10:14something it's not like manufacturing a
10:16car where you produce it once and you
10:18hand it off but rather you make changes
10:20to that infrastructure weekly daily in
10:23some cases right and so that idea where
10:24you make a change we're in an experiment
10:26run a test and do that very rapidly
10:28right in conjunction with doing things
10:30on the laboratory side and having these
10:32interfaces to these these biological
10:34mechanisms that allows you to iterate
10:36and learn very fast right and that's
10:38where the power of software kind of
10:40connects into the biological work Jeff
10:42you straddle both the startup world and
10:44you know clearly your background in
10:45large pharma how do those two what what
10:48what andrew describes how ready is the
10:51large pharmaceutical industry to embrace
10:54that well I think they're getting there
10:56and it's gonna happen regardless of
10:59whether they embrace it and I just want
11:01to add to what Andrew was talking about
11:02the big data element of this is is
11:04enormous no pun intended but
11:07now we have a huge amount a huge number
11:11of data sets that for the most part are
11:13pretty siloed when you can start to
11:15combine them and bring analytics to bear
11:18on them the insights that you gain as
11:21compared to kind of the human trial and
11:23error approach of the past is huge I
11:25mean one recent example is Watson
11:28healthcare at IBM I just bought a
11:30company called trevean for 2.6 billion
11:32dollars it's a company that has
11:34cloud-based data sets for more than 8500
11:37life sciences companies and millions of
11:39lives that it has data on when you
11:42combine those data sets with the other
11:44data sets that IBM is acquired recently
11:46and IBM Watson's analytical capability
11:49you're gonna learn things not just about
11:51drugs but about humans that are beyond
11:54anything anybody had before so going
11:56back to original question yes there's a
11:58there's always going to be a human
11:59element to this of course is always
12:00going to be interpretation and judgment
12:02and scientific discrimination and the
12:04rest but the ability to have data from
12:06which to make these judgments well
12:09structured accessible analyzed
12:12interpreted is is a game changer from
12:16where we were even say ten years ago at
12:18the end of the day what we want is
12:20better health outcomes so I I wonder if
12:24you've seen this story before in another
12:26industry or or maybe the other way to
12:29describe it is like what are we looking
12:30at and how are we going to benefit him
12:33what is that going to start to look like
12:34you know actually I would disagree that
12:36we don't just want better outcomes we
12:38want better outcomes at a lower price
12:40it's a combination which actually is the
12:42difference between sort of an advanced
12:44and our really major step for it you
12:46know there's many different industries
12:48in which compute has made huge advances
12:50and we could think about things as the
12:52rise of machine learning in in many
12:54different disciplines and this is
12:55something which is obviously outside
12:57healthcare but could make a huge impact
12:58you know it's really kind of shocking
13:00that computers now are getting routinely
13:03better than people so you know years ago
13:06was at chess and now it's at go now it's
13:09an image recognition that computers
13:11routinely do better than individuals do
13:13these advances are key milestones
13:16towards the role of and the capabilities
13:19of machine learning but we even see this
13:21if you think about companies like lyft
13:23or uber what makes those interesting is
13:25not just that's cheaper and not that
13:27it's easier but that combination of that
13:29you just have the cellphone Network
13:31there and you can call a car and it can
13:33be sort of effortless that's what
13:35creates this magical experience and the
13:36analogous thing that could happen here
13:38is that the machine learning can allow
13:40computers to do things that humans just
13:41couldn't do before but there's certain
13:43requirements we need the data and we
13:45need to be able to still have the right
13:47directions to point this to and so the
13:49human part is certainly going to be
13:50there but the opportunity is clear yeah
13:52it's interesting yet data is that that
13:54is my world right when without the data
13:56we you know as computer scientists just
13:58can't operate and so there are tons of
14:01public data sites available the NIH
14:04array expressed in the European Union so
14:07there is a lot of data that you can
14:08actually get your hands on no agreement
14:10required right millions of assays just
14:12go ahead and download we've also found
14:14that there's a lot of academic
14:16institutions that have data that that
14:17they're happy to share with folks but
14:20they're up there also is just tons of
14:23data that's that's hiding in the silos
14:25of large Pharma straight and so we look
14:26at some of the farmers that we've
14:28interacted with and they're there you
14:30know they say look this this is our gold
14:32right we're not handing this out no
14:33offense it's not you we're not giving it
14:35to anybody and we meet other companies
14:37like Santa Fe that is doing more in
14:39terms of sharing data with folks I
14:41believe that the way we're going to be
14:44able to I think make progress as a
14:45society and being able to process them
14:47the state and find new insights is
14:49through data sharing and so whether that
14:50comes with some sort of a business
14:52agreement or some way that we can
14:54aggregate those data and be able to give
14:57back to those who have contributed most
14:58in terms of the data that they have
15:00sought you know it's it's something that
15:03I think eventually is going to happen
15:05just because the benefits are going to
15:06outweigh you know that the cost of just
15:08holding it tight to you I mean it's an
15:09open source kind of approach to it where
15:11large corporations can get involved and
15:13they can help facilitate the access to
15:16that data or is the creation of that
15:17data and then they get the benefit in
15:19different ways yeah or they don't even
15:20have to release the data one can run a
15:22eye on the data in behind their firewall
15:26and use that to generate features or the
15:28things that could come out so I said I
15:29think there's a lot of creative ways to
15:31handle the IP challenges I think another
15:33aspect of this that bears on farmers
15:36that it's true at Pharma has a lot of
15:37data but the fact is the payers probably
15:40had more and better real-world data than
15:42the pharma companies do and one of the
15:44reasons that the pharma companies will
15:45eventually in my opinion have to do data
15:48sharing as Andrew suggests is that the
15:50payers are going to aggregate their data
15:53and they're going to start to have an
15:55analysis and interpretation of what
15:57drugs work and how they work and who
15:59they work on that the pharma companies
16:01are going to have to contend with and
16:03they won't have enough data by
16:04themselves to be able to do that so I
16:06think data sharing is inevitable and I
16:09think you know you already see large
16:11payers aggregating databases and as I
16:14said in many cases knowing more about
16:15the drugs and the drug companies right
16:17one of the reasons Big Pharma is so big
16:19is because it's so darn expensive right
16:22and you know you see this again and
16:24again we're a smaller biotech company
16:27has a promise in molecule or something
16:28and then it gets you know they can't
16:30push it cross the finish line so Big
16:32Pharma comes in and buys them and helps
16:34them get there does it make the idea of
16:36Big Pharma less attractive and does Big
16:38Pharma become smaller do we start to see
16:40a different industry well I I think
16:42you're onto something there and I think
16:44if we were having this conversation five
16:47to ten years ago the the status quo you
16:50described was probably there and if you
16:52think about it 10 15 years ago pharma
16:54companies had drugs with 80 percent
16:57gross margins that were making tens of
16:59billions of dollars and growing at
17:00double digits a year didn't worry too
17:01much about cost you know add another
17:03hundred million dollars to a trial so
17:05what if it's going to add billions of
17:06dollars to the drug sales that's changed
17:09the risk of drug development drug
17:11discovery has increased therefore the
17:13shareholders expect greater returns
17:15therefore the pharma companies have to
17:17think about cost until Vijay's point
17:19it's not just better outcomes but
17:20achieved it more efficiently meanwhile
17:22another trend that's going on at the
17:24same time is the increased growth of
17:27small virtual pharma companies that
17:30don't have all this legacy
17:32infrastructure that Big Pharma started
17:34within themselves outsource and
17:36specialized themselves so for example
17:39I'm an investor in a specialty pharma
17:41company that has about 150 million
17:43dollars of EBIT on it's got ten
17:44employees because it never built that
17:47like my movie analogy it puts together
17:50each deal as it needs it with contract
17:52research contract manufacturing contract
17:53sales and so forth I think to your
17:56question Pharma needs to get more and
17:58more like that whether the big ones can
18:01get there you know it's going to be a
18:02major test of them whether they can get
18:05smaller and leaner in order to grow I
18:07think some of them are on the way to do
18:09that others are still thinking in legacy
18:11terms and also I don't think it has to
18:13be mutually exclusive I think there will
18:15be changes in pharma but also there's
18:17just other models we've seen actually
18:19really intriguing other types of models
18:20models where foundations like the CF
18:23Foundation has played a huge role in
18:24pushing for new CF drugs CF cystic
18:27fibrosis okay yeah and so that's an
18:29intriguing area because this cystic
18:32fibrosis foundation can actually do many
18:34things to help push something forward
18:35they can help recruit patients and help
18:37with clinical trials as well as do
18:39funding for these trials and so I think
18:41that's one model that is outside the
18:43traditional pharma model but I think
18:44we'll see more of also there'll be other
18:46cases where there'll be maybe more
18:49orphan issues where there is no drug at
18:52all and so a clinical trial will be less
18:54expensive because the number of people
18:55you have to drive it through be much
18:56smaller there's not going to be the huge
18:58return that there would be for a
19:00blockbuster drug but that the price
19:02performance will work out just fine for
19:03about tech an interesting little
19:04experiment that you can do to sort of
19:06test the degree of maturity with regard
19:09to outsourcing is to ask firm of people
19:11what is the core competency of a
19:12pharmaceutical company and what you
19:14discover is you get a lot of different
19:15answers some people say research some
19:18people say sales some people say
19:19commercialization whatever and the very
19:22variety of answers that you get suggests
19:24to me that they haven't quite figured
19:26out what their real role in competency
19:27is and in my opinion the thing they're
19:29really good at that I I find it hard to
19:32imagine being disruptive in the very
19:34near future is that translational stage
19:37from late stage assets to initial
19:40commercialization figuring out what the
19:41market wants and again by analogy did
19:44the movie studios that's what they do to
19:45virtually everything else can be
19:47virtualized no pun intended and I think
19:50that will happen and I think we'll see
19:52more and more different models as Vijay
19:54says and the the farmers that succeed
19:57are going to be the ones that outsource
19:59more and more of their cost centers and
20:01really focus on what they do best does
20:03efficiency in all these areas get us
20:05more toward personalized medicine - you
20:08talk about from drugs but I mean
20:09there's always an orphan for a drug
20:11probably yeah you know there is some
20:14really tantalizing possibilities here
20:17that 80% of all cancers could be cured
20:20with existing drugs if we caught them
20:22early enough and we knew to give them
20:24early enough and so what we're also
20:25seeing is huge advances on the
20:27diagnostic side from advances in
20:29genomics or proteomics metabolomics and
20:31that clinical side actually I think will
20:34interface very nicely with what we're
20:35talking about and the diagnostic side
20:38will interface nicely with the
20:39therapeutic side and that could be a
20:41very different world I mean the
20:42challenge there is getting Diagnostics
20:44to be consuming more accurate trend they
20:46were before but that's an interesting
20:47challenge and also that itself is an
20:49interesting machine learning and
20:51computer science challenge to take these
20:52new data sets to combine which you'd get
20:54from proteomics and genomics and other
20:56take everything you can get and make the
20:58most accurate predictions and then give
21:00the best drugs we have now that itself
21:02would be a different role that doesn't
21:04require anything on the pure farmer side
21:05yeah and I think we need to really over
21:08time they also build enough data to be
21:10able to pull those things off I mean
21:11like for me ultimate personalized
21:13medicine as I as I walk into my pharmacy
21:15I say I'm not feeling well
21:16biology in biological samples are
21:19collected data is generated and I'm
21:21formulated a solution right on the spot
21:22right that's ultimate personalized
21:24medicine and so we can't do that today
21:27because the ability to generate enough
21:29quantity of data to be able to make a
21:31prediction to give someone a drug on the
21:32spot just you know that the costs are
21:34too high right and so we as data
21:35scientists are sort of looking at larger
21:37populations and we're starting to go
21:38down the path of being able to do
21:40subpopulation on disease and individuals
21:42and so eventually I think we'll have
21:44enough data that those sorts of things
21:46become very easy to do but for today
21:49right now I think we're just we're more
21:51headed down that path rather than that's
21:52something we can we can pull off at the
21:54moment I mean as a health care consumer
21:56that sounds great I walk into a pharmacy
21:58I walk out with you know what what I
22:00need for what ails me the sort of
22:02hurdles liability wise and otherwise is
22:05that a scenario that we get too easily
22:07or easily is the wrong word ever oh well
22:10we definitely get there and not easily
22:12and not soon but everything's headed in
22:15direction and if nothing else the
22:16customers and and patients will demand
22:18it I mean we have a generation of people
22:21today that are used to being able to get
22:23services and goods on demand very
22:25quickly very inexpensively without a lot
22:27of intermediaries and there's no reason
22:29they're going to accept the kind of
22:31health care system that their
22:32grandparents lived with which is of
22:34functionally what we have today so I
22:36think we will get there there will be a
22:37lot of disintermediation there will be a
22:39lot of companies that will fail as a
22:41result of this it probably will have a
22:43salutary effect on pricing and you know
22:47there's a there's an absolute need for
22:49it I think history suggests when there's
22:50a customer demand and a patient need it
22:53ultimately gets filled I think health
22:54care is is has historically been and I
22:58hate to say it's different from other
23:00industries but in some respects it is
23:01it's an incredibly complex ecosystem
23:05with you know thousands of different
23:07interests sometimes in conflict whether
23:10it be hospitals doctors payers patient
23:13groups pharma companies the government
23:15it's also been a system that
23:17historically has had intermediation
23:18between the customer and the ultimate
23:20provision of care drugs being just one
23:23example of that and it's been an
23:25industry where there's been
23:26understandably and rightly a huge focus
23:28on quality and control and as a result
23:31of those things it's been a little bit
23:34slow to change and just to give you one
23:35example that I think people in the tech
23:37industry might find remarkable it was
23:39only relatively recently that the FDA
23:41allowed clinical data to be entered
23:44directly into computers as opposed to
23:47put on paper because there was a
23:50reluctance to trust the quality and
23:52integrity of that data which I think we
23:54would all agree is kind of silly because
23:56it's much more reliable than paper but
23:58just the fact that that only happened
23:59recently suggests that when you're in a
24:01in an industry with a regulator that is
24:04quite rightly and understandably
24:06concerned with quality it can be slow to
24:07change but I think that change is going
24:09to accelerate I think it's not going to
24:11be long before we see things along lines
24:13Andrews talking about yeah and I think
24:15you know what's really exciting about
24:16the future of having data Sciences get
24:19involved in some of these things
24:20especially we're talking earlier about
24:21preclinical studies and animal models
24:23which you know as it turns out a mouse
24:25is not terribly predictive of what might
24:27actually happen in human
24:28in cancer in my silene time exactly
24:30right and so when when we look at some
24:33of the data models and what we can do in
24:34terms of the predictive power of those
24:35models you know I see a future where
24:37we're getting rid of the animal because
24:39what we can simulate in the computer is
24:41going to do a better job of predicting
24:42what's going to happen in that human
24:44being right so what does that mean we're
24:45talking about you know the FDA right
24:48we're gonna we're gonna convince the FDA
24:49we're just going to put it all in the
24:50computer then we're going to drop it in
24:51a person right like that's the types of
24:54changes that I see coming in the decades
24:56to come and and the good news is we're
24:57building that capability and that that
24:59is part of our future but to your point
25:01it's going to take a while to get there
25:02right there's a lot of hard evidence
25:04that we're going to have to produce
25:05we're going to make sure that you know
25:06these things are safe and reliable but
25:09it is where we're going right now
25:10there's no doubt also for many the
25:12reputation of Silicon Valley is
25:13something like move fast and break
25:15things it's when it's when it's your
25:18internal organs involved yeah yeah yeah
25:20and so so and I think it's important to
25:23emphasize that even though there's very
25:25much a lot of this look about a culture
25:27of applying computation and innovation
25:30involved here it's understood that this
25:31is a different type of space and that
25:33there's gonna be new types of challenges
25:34I can't wait to get into my driverless
25:36car and go to the pharmacy and stand in
25:38front of the machine and it spits out a
25:41pill just for me and maybe you won't
25:42have to go the farm yeah I think it's a
25:45blood test at home and something arrives
25:47later that day Jeff I don't mean to
25:50paint you with the big farmer brush but
25:52I'm going to anyway you and Andrew are
25:54sitting here side by side andrew is
25:57definitely a product of computer science
25:58and here in Silicon Valley how is it
26:01that you guys are getting along not in
26:03this room but how is the culture of
26:05where you come from Andrew and meeting
26:07the culture of you know Jeff that you
26:09used to inhabit and still do in the
26:12early days and maybe even to some extent
26:15today of interactions between the valley
26:17and Pharma I'd be in rooms where those
26:20two people were meeting and one speaking
26:21Greek and one speaking Latin there was a
26:23just a fundamental cultural divide they
26:25didn't understand each other's business
26:27and industry Andrew can speak to whether
26:30or not he experiences that today but I
26:32think it's got to change on both sides
26:34you know I used to when I first started
26:37doing investing in this space ten years
26:41from Silicon Valley that made no sense
26:43from a healthcare perspective and a
26:45total failure on the healthcare side to
26:47understand the value of computational
26:49benefits so I think that's changing I
26:51think it's got to change like it has in
26:53every other industry you know the
26:55history of of industry in general is
26:58that the disruptors usually come from
26:59the outside and one of the one of the
27:02challenges Pharma has had is its own
27:04success it hasn't had an existential
27:07crisis the way IBM did in the Gerstner
27:09era for example which is one of the few
27:11examples you can think of where a big
27:13industry really changed most of the time
27:15it's Amazon taking over Barnes and Noble
27:18so I I imagine especially going back to
27:21my comments about virtual pharma
27:22companies I imagine a lot of this
27:24innovation will come outside a big
27:26pharma but the big farmers that succeed
27:28will be the ones that get it Andrew yeah
27:30how is it when you sit down in the in
27:31the halls of big pharma I'm gonna I'm
27:33gonna put it out there right now I'm I'm
27:35happy to admit this I'm a weird dude you
27:39know when when software engineers and
27:42mathematicians think about problems we
27:44think about things in just way different
27:46ways and so one of the things that I
27:48think is just so interesting is that
27:51self-driving cars did not come out of a
27:54GM or a Ford or an automobile company it
27:56came out of a software company came out
27:58of Google right and so that way of
28:00thinking that culture that you know sort
28:03of stepping back from from the problem
28:04and just twisting it in a very different
28:06way I think is is quite compelling right
28:10there's just sort of this this different
28:13view on problems that I think you know
28:15can offer insights and and offer new
28:18abilities to produce things that that
28:19maybe others haven't considered and it's
28:22almost some of these things feel like a
28:23matter of time but there will be hiccups
28:25you know before Amazon there was web van
28:27and you know after web then you might
28:30think well you know this internet
28:31thing's not going to work out like you
28:32would never be useful for home delivery
28:34or anything like that but lo and behold
28:36comes a company that takes these
28:38advances at the right place in the right
28:39time and does something great the
28:41there's almost like a deterministic
28:44nature of Moore's law that costs of
28:46compute and cost of genomics on these
28:48things decreasing exponentially that
28:50actually is something that's really hard
28:51to avoid and so in my mind it's not
28:53really a question of if but
28:55this happens and that I expect to
28:57actually the next 10 years could be
28:58really fantastic towards those ends
29:00Jeff Andrew thank you guys so much thank