Curing Hair Loss With Sean McClain

[Music]

Welcome to FYI, the 4-year innovation

podcast. This show offers an

intellectual discussion on

technologically enabled disruption

because investing in innovation starts

with understanding it. To learn more,

visit ark-invest.com.

[Music]

Ark Invest is a registered investment

adviser focused on investing in

disruptive innovation. This podcast is

forformational purposes only and should

not be relied upon as a basis for

investment decisions. It does not

constitute either explicitly or

implicitly any provision of services or

products by ARK. All statements made

regarding companies or securities are

strictly beliefs and points of view held

by ARK or podcast guests and are not

endorsements or recommendations by ARC

to buy, sell, or hold any security.

Clients of Arc Investment Management may

maintain positions in the securities

discussed in this podcast

>> from Arc Invest. This is FYI, the

four-year innovation podcast. With me

today, we have Kathy Wood, our hero and

savior, CIO and CEO of Arc and

and Sean Mlan, returning champion. Sean,

welcome back to the pod. Sean is the

co-founder and CEO of Absi. Sean, remind

us what is ABSAI? What are you doing?

Uh, tell us about AI, drug discovery,

drug development, please.

>> Yeah, absolutely. Well, thanks for

having us on for a third time. So, we

are a generative drug design company. We

are using uh generative AI to actually

design molecules in our case antibodies

to some of the hardest uh drug targets

that are out there. And I think if you

look at you know what AI is doing in

general, it's you know making things you

know faster, better, cheaper. But one of

the pieces that you know generative

design uh within drug discovery allows

us to do is actually open up this fourth

dimension which is going after new novel

biology. Um being able to go after

biology that has previously been

undruggable. And we've been able to

actually show success with some of these

targets in in partnerships. One of them

was in partnership with Caltech. There

was this Caltech had made this really

great discovery where they found this a

highly conserved region in the HIV virus

called the caldera region and no one had

actually figured out how to drug that

because it was this very deep crevice

and our own immune system couldn't

figure out a way to drug it. humans

couldn't figure out. But we were able to

actually use our AI technology to design

an antibbody that could bind deep with

inside that crevice potentially leading

to a neutralizing antibbody to all

clades of HIV which no one had ever done

before. And I think another like prime

example of this is in our partnership

with Almall. There is this particular

ion channel for dermatological

diseases. And this ion channel had been

known for 30 plus years. And everyone

has, you know, been figuring out ways to

block this particular ion channel. And

we were able to again use this

generative design platform to actually

design an antibody for the first time

ever that could actually block this ion

channel. And you know, these are just

some of that kind, you know, some of the

the proof points that we've seen already

in kind of unlocking this new novel um

biology kind of in addition to making

things, you know, faster, cheaper, uh

and and better.

>> Sean, just uh on that last point, the

ion channel, many people who are going

to be listening to this would like to

know your ability to do that, to block

that.

What who did that help? who did that

benefit?

>> So, I can't necessarily dive into that

particular program just due to

confidentiality, but I'll I'll um I'll

paint a picture on a different ion

channel, which is a potassium ion

channel, KV 1.7 and 1.8. And this is a

well-known ion channel in pain. And

there's actually great human genetics

data out there that individuals that

have these ion channels

knocked out don't feel pain. I'm sure

that you know you've heard of the fire

walkers in you know Southeast Asia. And

the idea is is if you can actually

block this particular ion channel, you

would actually eliminate pain and you

would actually have a non-addictive

pain medication that isn't uh opioid

driven. And for us, these are kind of

the hard challenging, you know, problems

that that we want to solve. And so

that's kind of kind of to give an

example of you know uh in this case an

ion channel actually regulates uh pain

>> and on the HIV side potentially this

would provide the fuel for a potential

HIV vaccine that would be universally

useful.

>> Yeah, that'd be universally useful

across all the different variants of of

HIV that exist today. It's always been

hard to drug the the virus because it's

so rapidly changing and evolving. And

the discovery that Caltech made was this

highly conserved region that doesn't

change from, you know, the different HIV

viruses that evolve over time. And so,

uh, if you're able to actually target

that, then you could essentially wipe

out all the different variants, uh, of

HIV. And again, no one actually had, you

know, figured out how to, uh,

necessarily go about, uh, finding a a

vaccine in this particular way.

>> So, these are huge markets. How long? So

you've seen success. How long in in

terms of research are we looking at you

know six years 13 years how long before

we would see something on the market

>> to talk about you know how quickly we

can you know use generative design to

actually get to a you know phase 2 proof

of concept in in humans. I'll focus in

on ABS 2011. So this is uh an antibbody

that we are developing towards the

prolactin receptor and this is a really

cool mechanism for hair regrowth. And

essentially what we've found is that if

you're able to block the prolactin

receptor, you can shunt the follicle

from this essentially dormant phase

where you get apoptosis and and

regression and the hair no longer, you

know, grows to actually shunting the

follicle back into the antigen state or

this active growth state. And

this is, you know, this would

essentially be able to, you know, regrow

hair as well as repigmentation.

And, you know, we started on this

program roughly, you know, two to two

and a half years ago. And from, you

know, start of the the program to our

phase 2 proof of concept readout next

year in in humans will be roughly three

and a half years to get to a phase 2

proof of uh concept readout. This is

truly, you know, unheard of. I mean,

normally it takes three and a half three

and a half to to five years just to get

one drug in the clinic and then, you

know, takes years after that to then get

a proof of concept uh readout. And, you

know, not only were we able to do in

three and a half years, but we were also

able to do it with a $15 million

investment where normally it takes 50

to$und00 million just to get it into the

clinic and then billions to actually

develop it once it's in the clinic.

>> That's pretty wild. I know I backed into

that question. Uh, but we really wanted

to talk a little more about AI. Am I

right?

>> No, that's fine. I mean, one, let's just

stop there for a second. Hair loss is

actually your hair follicles just

falling asleep, right? Basically, they

just like fall asleep. And this is like,

you know, you plug in kind of like into

an electric socket in the cell like wake

up, wake up, you know, and then it wakes

up and keeps growing or that's the

promise. And it's very different from a

Roane or you know the existing

treatments on market which which aren't

like precise keys that plug into the

cell.

>> Exactly. It's actually regenerative

biology. So essentially what ends up

happening is you get prolactinmia on the

scalp essentially a buildup of prolactin

and that essentially drives the follicle

into that kadagen state again where you

get the apoptosis and regression and the

stem cell is no longer active and the

the follicle itself is as you mentioned

Brett essentially dormant and by

blocking the prolactin receptor we're

actually restimulating the stem cell

growth

and actually allowing that that follicle

that's existing to start to grow hair

again as well as to actually get

pigmentation

back as as well. And you know, these are

kind of some of the exciting themes

that, you know, we have within our

portfolio is is focused kind of on this

this regenerative biology where we can

actually figure out how to reverse some

of the damage that has been caused by us

just aging.

>> May I just ask one quick question on

that? How did you happen upon that

particular application? What did you

back into it? was or was it we're going

after this?

>> This was actually uh discovered by our

chief innovation officer on Andreas Bush

who was the uh cso at at Bayer and uh

you know AI in this case cannot take

credit. This was a a total serendipitous

uh discovery.

Bayer was actually looking at the

prolactin receptor for indometriosis.

And when they ran the animal studies,

they they actually saw the mice that

were on the drug regrow their hair

faster than the control arm. And that

led them to then dive in and figure out

that actually the prolactin receptor is

not only involved in indometriosis, but

is involved in the hair regrowth. And

you know and and so essentially Andreas

took that knowledge to to ABSI and then

we used our generative design platform

to design kind of the best antibbody to

to ultimately address that particular

target and and be able to actually see

this um you know biological pathway come

to fruition in in in patients.

>> And Bayer didn't pursue it in part

because they couldn't find something

that efficiently plugged in. Right.

>> Exactly. they developed an antibbody but

it wasn't commercially viable. So in in

order to get the efficacy that was seen

in in the Stumptail Macaks which you

know completely regrrew their their

hair, they were going to have to to dose

24 injections over a uh six-month

period. uh versus, you know, we're going

to be, you know, dosing every two to

three times over that 6-month period.

And so being an out-ofpocket, you know,

cash pay, you know, product that just

ultimately wasn't going to scale. And so

we were actually able to develop a drug

that could be, you know, uh, at a at a

price point that would be commercially

viable.

>> That's like the key

if you imagine. So we've done modeling

on AI impact on drug discovery. We think

you can, you know, over time shrink time

to market and and and shrink your cost

to get to market in aggregate. Right

now, from the likes of you all, there's

pretty good evidence that the

pre-discovery, everything up to the

clinic can be done more efficiently. You

can get into the clinic more quickly,

you can do it more capital efficiently,

and um then there's an open question

because most of the AI developed drugs

are not yet commercialized. It's once

you're in the clinic, are you going to

be able to be more efficacious there so

you have a better odds of getting all

the way through?

>> Well, in this particular one, isn't is

the question, wait a minute, either the

hair grows in back with pigment or it

doesn't and you know, is that is that

efficacy be will it be determined in

phase two and you don't need to go to

phase three? Do you cut that out?

>> Yeah. So, so the great part about this

is that we're actually able to run a

combined

phase one phase two trial where we're

going to be able to look at both safety

as well as uh hair regrowth. And since

you know most of the you know

individuals that have you know AGA or

antigenic alipcia are you know healthy

individuals we you know we essentially

can you know run combined uh phase two

or phase one phase two study uh looking

again both at safety and and hair

regrowth. And then we would then run a

pivotal study uh after that. And the

great news is that if the readout looks

good next year, we will be filing our um

BLA with the the FDA and seeking

approval around 2030.

And so that's actually from a from a

time to market that's actually a very

very quick turnaround time. Uh and and

again I think uh it really goes to show

that the you know how AI can ultimately

kind of collapse overall timelines and

and increase overall success.

>> Yeah. So Brett has uh led uh uh our

research on how much AI uh will be able

to shrink the time. And I think if if

you're successful on this uh our

research is spot on because we've said

the average time uh for a for a drug up

to now without AI has been 13 years.

You're saying about eight years, am I

right? Seven or eight years. And

>> yeah, seven or eight years. Yeah.

>> Yeah. We've been projecting that by

2030. So, bravo, Brett.

>> Oh,

>> yeah.

>> Let's not judge it based on one drug,

but

>> it's okay.

>> I'll take the win. Sure.

>> Yes. There you go.

>> It is really exciting. This this will be

one of the first AI designed antibodies

that will have a phase 2 efficacy

readout. And so, this will be one of the

the first of of its kind. And so, we're

we're really excited. I mean, not only

for Yeah. the the readout you know for

for the drug for patients but also I

think a really monumental time for for

AI to drug discovery as well.

>> One last question and this again is back

to to the research. We have said that if

you include failures the average cost

to bring a drug drug from discovery to

market today is about 2.4 billion.

How much do you think this taking this

drug to market will cost you?

>> It'll be a fraction of that. I mean, if

I look at the the, you know, total in

investment,

it's easily going to be

sub$1 to $150 million, you know, to to

ultimately get approval on on this. So,

you know, a fraction of that, you know,

$2 billion.

>> Of course, that's conditional on

success. So the, you know, again, don't

judge based on one drug. It could be

that Sean and Absai are just very good.

>> I know, but I'm just trying to help

people understand what the what we think

the possibilities are. Here's, you know,

this may actually fit into our eight

years. Now, our our cost estimate is 600

million per drug, including failures.

So, you know, if if you uh skipped a

beat along the way, it could be as much

as that. But if not, oh my gosh, that's

that's unbelievably good.

>> Yeah. No, absolutely.

>> What's interesting about drug discovery

to me is like not only can you

potentially shrink time preclinally or,

you know, discovery, etc.

And you know, if you get to market with

just a key that plugs in more with a

tighter coupling to the lock, then you

don't have to give the patients as many

doses. So, you know, an injection every

two to three months as opposed to it

sounded like an injection every week.

>> Every every week to two weeks. Yeah.

>> Yeah. And and um it means that kind of

the potential for any kind of like

toxicity effects are lowered because

you're able to give a lower dose and

still like tightly coupled and it's not

going to plug in anywhere else. This is

a key made just for this lock. And so

then you know no matter the disease, it

could be hair loss, could be whatever.

Um the the promise is that then these

really are precision medicines whereas

before we were kind of relying on kind

of random walk through nature to like

find the way in which the thing fit.

>> Yeah, 100%. You kind of hit the the nail

on the head there. I mean I I like to

say that you know we've really gone from

now this paradigm of searching for the

needle in the haststack to actually

being able to create the needle. In our

case, it's an antibbody based drug, but

we can now actually start to test

hypotheses that weren't possible. Yeah,

you can actually start to engineer in

the features that you want. Bind in the

tumor micro environment in acidic pH,

but you know, don't bind in in healthy

tissues. Activate this pathway, block

this pathway, and you can really have

that that precision when when you're

going about, you know, drug discovery.

And it also allows you to fail faster if

you can actually quickly and rapidly

test these these hypotheses and figure

out which of these, you know, actually

have, you know, legs or, you know, which

of them you you should, you know, kill

and, you know, you know, the the

biology, you know, didn't pan out.

That's hugely valuable. And I think that

that's, you know, what I think a lot of

times is is overlooked is is that

ability to fail faster and have that

that learning that learning loop uh

dramatically decrease because the you

know the the shorter that that learning

loop is you know the the faster we're

going to and and more cost effectively

we're going to be able to to uh you know

get these drugs to patients. you were

going to the chip companies and we we're

struck that, you know, that both Nvidia

and AMD understand

the promise here and want to be a big

part of unlocking the secrets of life,

health, and death and and and so forth.

So, I know AMD invested in Absi and

Nvidia and recursion.

So maybe you can talk about that a

little bit, how that how that happened

and and the similarities or differences

between the two of you.

>> Yeah. No, absolutely. I mean with Nvidia

investing in recursion and AMD and

investing in ABSI, it it really goes to

show how, you know, deeply committed,

you know, these chip manufacturers are

to, you know, healthcare and life

sciences. not only committed but I think

actually see you know massive upside and

and potential and I I think it's you

know both you know AMD and Nvidia I

think really have the the the vision

that healthcare could be you know the

the next kind of big unlock for for them

and you know the compute utilizations

that you know they're going to see I

think are going to be astronomic

you know relative to to others just

because of the complexity of biology

it's so complex and it's kind of like

the perfect

use case for AI and and and deep

learning. And you know, we've had

partnerships with both, you know, in

Nvidia and and now AMD. AMD, you know,

earlier this year made uh you know, an

investment in ABSI. And one of the

reasons we decided to partner with with

AMD was, you know, one how committed

they they were to the the overall uh

space. But the the interesting part to

us was actually the the chips themselves

and the higher memory that they had. The

fact that these chips had higher memory

meant that we weren't going to have to

crop the input to the model. And so

essentially, you know, we're training on

structures and protein sequences. And by

cropping them, we're only putting in a

fraction of of the overall, you know,

target structure. You're getting limited

information. But if you have more

memory, you can actually train on the

whole structure of the the target or

protein, which enables you to ultimately

have, you know, a higher fidelity and

more accurate um model. You're just

essentially being able to get more

information into that model. And so we

saw that as, you know, really exciting

from, you know, a training perspective.

We're excited to be partnering with with

both of them and and ultimately we see

them being able to kind of uh at least

AMD being able to to unlock new kind of

features but also drive down our our

overall uh cost as as well on on the

training side.

>> If you look at your chip and your chip

utilization, is it right to think of

like, hey, I'm devoting some of this to

training and some of this to inference

in your data center? How is that working

and how does that trade-off occur? Yeah,

since we're an R&D shop, we're we're

still I would say predominantly on the

training side, but we are seeing more

and more ultimately um going to

inference and you know inference costs

are increasing. I think that's another

area where you know we see the AMD you

know chips being you know very uh

efficient on

on the inference uh side of things and

we do see that being more and more

important moving forward but I would say

you know predominantly it is training

with a you know a decent chunk going

into inference and really trying to

drive down those inference costs with

with the AMD platform. tell me about the

data mix, what you're feeding into the

model and then kind of like what's the

equivalent of a prompt for the inference

engine on the back end. Yeah. How does

this all work?

>> Yeah, absolutely. So, you know, we're

really working with like multimodal

models. So, you know, we're we're

training on, you know, the the structure

of of proteins. Uh we're we're also uh

training on the sequence of of

antibodies and then the overall affinity

to a particular uh target of interest.

And then we're also you know training on

you know various um developability um

parameters you know is this you know

particular molecule going to be a

minimal for for manufacturing can we get

you know high tighters with with this is

it going to have good solubility?

Is it going to be you know have low

cross reactivity with other proteins and

you know is it going to have a low

immunogenicity

profile? And so these are kind of all

the different data points that we're

putting into into the model.

And when you look at the the prompt, you

know, essentially what we're, you know,

inferencing on is, you know, taking the,

you know, let's say the the colder

region of the HIV virus, going back to

the the HIV case study, we're

essentially taking the that known

structure. We're telling the model we

want to generate an antibbody that will

bind to this particular area of of the

target which is called the the epitope.

And the model then is going to design

essentially the fingers of the antibbody

that are able to then bind to that

particular target of interest. So, you

know, essentially going kind of to your

lock and key analogy, we're essentially,

you know, giving the the model the the

lock and then it's designing the the key

to to fit in appropriately to to

ultimately give us kind of that biology

uh that we're uh looking for. And not

only is it designing the molecule to

bind to that target, but it's also

taking into account again all of these

developability parameters because it's

not about just making sure that you have

a potent, you know, efficacious

antibbody. You also need to make sure

that it doesn't react with the immune

system and have, you know, a tox effect

in in in the clinic. Uh you need to make

sure that you can manufacture it cost

effectively. And so there's all these

other parameters that are really

important that the model takes into

account as well when designing these

particular drug antibodies.

>> May I ask just one more question on the

on the AI? It seems to me that the two

chip makers, the largest ones

have chosen

you and recursion.

I think I don't I they haven't invested

to my knowledge. Brett, do you know uh

in any other? So, it really puts you in

a in an a very interesting and special

position, I think. Am I right on that?

>> Yeah, from a public standpoint, I think,

uh, you know, us and and recursion, I

think, are are the, you know, only

public investments that they've made. I

think you know we're you know you are

seeing you know in in Nvidia and AMD you

know I think invest in a lot of the

you know startups in in in the space but

in terms of you know public companies

you know were you know really I think

the only two investments that they've

made

>> very interesting

>> yeah biological data is like truly

astronomically big data strategically

they definitely if you imagine chips are

just a mechanism for like bulk

processing like you know large volumes

of data well here We are just scratching

the surface on understanding biological

systems and there's so much to kind of

like unpack and throw into those ships

>> in helping people understand that you

know who are trying to understand wait a

minute how are these two spaces coming

together. uh 35 to 40 trillion cells in

our body is all I have to say and and

you know large largest data project you

know

most important source of proprietary

data I think uh most profound

application of AI

>> yeah and I think one of the things that

is important to recognize is that

we have kind of taken all of the

lowhanging fruit that's out there of the

available data and I think we've done a

you know a pretty good job of figuring

out how we can utilize that that data

and in both drug discovery and and

clinical development but we are starting

to to reach a point in time where we do

need to generate new data to train the

these models and you know one area that

we kind of think of is like the

translatability so you know right now

we're designing the the molecule again

with a hypothesis in in mind but the

next step would be okay how Can we, you

know, use an AI model to actually

predict how that molecule is going to

behave in in humans to to ultimately

kind of, you know, re reduce uh, you

know, the the overall cost of of failure

and making sure that everything we take

forward is going to have the highest

success possible. And we're missing the

data that translatability data to

actually make the these models. And it's

difficult because you you have to

generate it from you know both animal

data as well as you know human data and

there's limited data there. And so how

do we go about generating kind of the

you know these key pieces of of data to

ultimately I think get some of these

important models designed of of of the

future. And so that's kind of an

exciting piece that we're looking at is

is figuring out how we can create new

technology and then use, you know, you

know, existing kind of simulation data

to ultimately create the data package

that's going to be needed for these um,

you know, these AI models.

>> Can I just ask one question on the

animal testing and animal research? Uh I

think that uh that the FDA is saying now

no more need for animal testing for

monoconal antibodies and and that may be

more. Is that going to change anything

for you?

>> Oh, so we love the fact that the the FDA

came out with guidance that said they

ultimately want to replace animal models

with AI models. And that is the

direction that we are headed. there's,

you know, steps that we need to take in

in order to to get there. I think, you

know, another piece is, you know, how

can we create AI models that can, you

know, predict immunogenicity and and how

it's going to ultimately behave from a

safety and talkert perspective. And then

there's actually predicting how

efficacious is the, you know, is the

drug going to be, is the the biology

viable? And those are kind of like the

two different paths that, you know, the

FDA is is wanting us to to go. And and

again, for us to get there, we're going

to need to continue to generate new data

in order to enable that. But the fact

that the FDA wants to go that direction

is is extremely positive because that

ultimately is it's what's going to, you

know, streamline and, you know, make,

you know, drug discovery and development

a lot more cost-effective. It

>> it's becoming AIcentric as an

organization, it seems to us. Yeah,

>> absolutely. And it's it's really great

to see

>> and one of the ways I think about it and

actually this will kind of segue into

the longevity discussion a little bit

but we you know there's this idea that

there's I think it's maybe 7,000

diseases. There's not 7,000 diseases.

There are there are many many more you

know I don't know about orders of

magnitudes more but at least 20,000

because that's how many genes you have

but really like practically you know

each individual like every cancer is its

own unique disease. We are we are

suffering from a huge der of precision

precision medicine obviously because we

have only diagnosed 7,000 diseases.

There's so much work that could be done

to like actually you know create the

puzzle pieces that help our body

function better.

>> Yeah, absolutely. And if you can

actually increase the probability of of

success and decrease the overall

cost, you can actually start to go after

smaller and smaller patient populations

until you get to the point of

individualized medicine. But first we

need to show you can increase that

success rate because that drives a ton

of o you know overall cost and you know

that you know will will truly I think

drive down the the cost to you know

start to look at smaller and smaller

populations and then again going to that

individual

>> right and and that's like the stage

we're at now is very much the okay prove

to me that it works story uh you know

you can and having proven that it works

that will attract a lot of capital

because kind of the apparent RORO return

on R&D could be very high. I mean, we

think it could be in kind of like more

technology landscape rather than

traditional drug discovery, which has

been a horrific use of like equity risk

capital over the past decade, you know,

getting 4% returns. I'd rather be in

cash, right? The promise of AI drug

discovery is you can do, you know, 20

plus% returns on R&D and then just like

massively expand the the target set

you're going after. Before we go into

longevity, Brett, maybe we we can stop

right there and and uh you know we have

over the last uh four to five years it's

been very tough in this space and I mean

we have been calling this and we truly

believe that you know healthc care is

the most profound application of AI

brewing out there and we've gotten

validation with Nvidia and AMD

buying into this space and you're a lot

of that in activity increasingly in the

private markets. Um why do you think as

you are talking to analysts

um let's say in the public uh markets

are you getting questions that help you

understand the misperceptions out there

or is the time horizon just too long and

they're hiding behind higher interest

rates or or or whatever other excuse the

macros are are throwing at them.

>> You're completely right. I mean it's

been uh you know 5 years of you know a

biotech uh nuclear winter and I I do

think that

you know spring is is coming I think you

know not only for you know biotech but I

think you know AI drug discovery and and

uh we we are seeing a lot of interest

pick up. I think one of the big pieces

that in investors and and analysts you

know want to see is that look we we hear

you that it can be you know better

faster and and cheaper and there's a

chance that you know you can you know

lower overall uh success rates. It's

great that you've shown it, you know,

preclinally, but we want to see, you

know, this be actually, you know,

efficacious in in humans and and and

actually translate have that

pre-clinical data translate in into

humans. And so, I think we do have a lot

of, I think, exciting wins that are

going to be, you know, put on on the

board. I think you you have Incilico's

readout coming up. You have recursions,

you know, readouts that were coming. You

have ours that are going to be, you

know, next year. And so we're gonna

have, I think, a a um a whole slew of,

you know, phase two proof of concept uh

catalyst coming out that I think will be

really strong proof points to show how

AI is really shaping drug development

and that we are able to start to

use AI to unlock novel biology that's

going to, you know, ultimately be

beneficial to to patients. But I do see

the turn in the market and interest

rates you know coming down I definitely

think are going to you know help help in

the space I think.

>> Yeah. But even so if you look at

interest rates they peaked uh they

peaked well I guess it was short rates

long rates peaked in in 2023

and still I think you know the lack of

M&A lack of strategic price discovery

just um put the kibos you know and IBIDA

losses even as rates were starting to

come down um most most time horizons it

seemed to us had had collap collapse and

that's changing. I just want to in the

last answer you said lower success

rates. I know you meant lower failure

rates.

>> Yes, lower failure rates. Thank you for

the correction on on that Kathy.

>> Yeah, there is an interesting

intersection. Well, one there's one

thing that people say as we're

constructing like our total forecast.

They're like, well, okay, you expect

there to be an inflection and basically

spend for medicine as medicine comes

becomes more precise. And it's like,

well, where does that come from? One

place it can come from is inefficient

spend for existing health care stuff. If

you have a hair loss drug in market

actually you can displace non-insured

spend which is somebody paying for a

hair transplant. Instead they can take a

drug and that's actually could be a lot

you know less expensive and more you

know you don't have to wear a hat for 6

months or whatever and do go through a

surgery. Do you think kind of that drug

and then other drugs that you

potentially have in discovery will be

more of this kind of like consumerdriven

marketing action that's and maybe

longevity generally has to move in that

direction.

>> Absolutely. I think you know some of the

things that we are working on that I

think can definitely be consumer focused

uh would would be in you know um looking

at like sarcopenia for for example. So

essentially muscle wasting. So you know

as as you age once you hit about 40 you

lose you 1 to 2% of both muscle mass as

well as uh strength. And by the time you

hit uh you know 50 to 60 you know that

can get upwards of 3% in in overall

muscle and strength uh degradation and

>> per year.

>> Per year.

>> That's pretty wild.

>> Per year.

>> No wonder I'm training so much.

>> Yeah. I better I'm just going to do some

leg lifts as you're talking. Just keep

going.

>> Yeah. The craziest part is like you know

you look at hip fractures and you know

in in the elderly

15 to 30%

of those that ultimately sustain a a hip

fracture have a one-year mortality rate.

And it really just goes to show how

important overall, you know, strength is

for ultimately preventing, you know,

very deathly uh falls. And so we are

working on uh a a drug uh right now or a

target uh that's actually known for for

longevity. Uh this particular target has

been actually shown to increase the the

lifespan of mice by by 25%. There's a

lot of various other kind of longevity

targets you can go after, but we felt

like being able to go after sarcopenia

and being able to show that you could

increase overall strength would be a

great initial indication and then you

could ultimately use this in a lot of

other, you know, indications that are

within longevity and kind of along this

this idea of being able to kind of

reverse the the damage that biology has,

you um ultimately acrudeed on us as

we've as we've aged.

>> Could that be an insured drug going

against sarcopenia maybe defined to some

subopul? It could.

>> Yes, we think that uh this could

definitely be uh a um you know both both

out of pocket but also insured as as

well. And I think different, you know,

other indications I think could be out

of pocket as as well, you know,

depending on how the doctor, you know,

wants to ultimately prescribe it.

>> Cuz it is interesting to think some of

these drugs, for instance, sarcopenia.

It's like you could imagine an insurer

will look at it and be like, well, this

person only becomes at risk of falling

and breaking their hip at the age of 74.

So, we're going to allow allow it to be

underwritten there. But here I am. I'm

like, well, I don't want to lose 3% of

my muscle mass. I want to be good at

tennis, right? So, I'm willing, you

know, you're trying to avoid me breaking

my hip 20 years from now. I'm trying to

avoid me like losing some speed on my

serve, you know, so I'm willing to to

kind of pay to avoid the disutility of

aging and potentially that becomes like

actually a bigger driver of the market.

If we're in a world of abundance, if

we're all we have humanoid robots taking

out the trash for us, we're riding

around inexpensively in robo taxis, we

have all this consumer surplus, what are

we going to spend it on? It's I want to

live forever. I'm going to spend all of

my money trying to live forever. That's

the the promise here.

>> And there's another thing. You've got

the GLP1s. Uh those that that

accelerates the muscle uh degradation,

right?

>> Yeah. And it actually causes balding as

well, which is really interesting.

>> Oh, wow.

>> You're going to be an antidote to all of

the problems that GLP wants. like they

they do a great uh number of good things

but uh there are side effects. Yeah.

>> It goes to show you that the consumer is

actually willing to to your point uh

Brett to actually pay for medicine that

is you know potentially going to make

them you know live longer and healthier

and is like more preventative in nature.

And it's going to be a really

interesting question on how you know

insurers you know approach this because

I think you know that's where we see

things headed is you know being able to

prevent you know let's say the four

horsemen from occurring which you know

is is cancer any neurogeneration uh

disease metabolic uh dysfunction disease

such as you know you know diabetes as

well as cancer and if you can

essentially you know prevent those from

from occurring you can ultimately save a

lot of capital uh in in the long term,

but is in you are the insurance

companies actually going to pay for for

that upfront if you can actually, you

know, show you can, you know, prevent,

you know, muscle wasting, you can

prevent cancer, so on and so forth. And

so, it's going to be interesting to kind

of see how, you know, insurance

approaches this and and are we going to

be going direct to consumer on this? And

is the consumer, you know, willing to

pay for it? And from what we're seeing,

I think some of these, I think the

answer is yes. the consumer will uh will

ultimately pay for it,

>> right? I mean there is like this um

principal agent problem and that

insurers have patients on plan for an

average of four years, but patients live

longer than I mean hopefully live longer

than four years, right? And so if you if

you're promising something that actually

will extend their life on the back end,

it's like, well, why am I on the hook

for this now? Wait till you're about to

die, then you can come to me, you know?

And so it it it definitely misalign kind

of like the commercial incentives in the

market and and again it kind of biases

people towards sick care. It's like find

the deathly sick patients and extend

their life by three months and uh rather

than be like let's extend everybody's

life you know by you know a couple years

with something. We've been doing some

research on this uh you know

prescriptions or therapies

less than 10% of the health care budget

which is closing in on 20% of GDP change

that 90% shift it more towards therapies

and a as we prevent uh you know

preventative med medicine and cures uh

we think will take share from the sick

care and it's in the insurance company's

interest to do so.

>> Yeah. I mean the the the expense that

you know is accured in in the hospital

and you know the last you know few years

of of life and you know all the managed

care is just uh you know outrageously

expensive you know compared to you know

the actual you know therapeutic

development. If you can prevent it

you're going to you know save you know

ultimately a lot a lot more in the long

run. And if you look, so we've done some

modeling. There's this kind of open

question like how do you think about the

longevity market? It's huge. It's

obviously huge, but we actually try to

put a number on it where if you look

across the US population today,

you can kind of bucket it into like uh I

die too early or even though I live long

enough, I'm still getting like weaker

and less viral over time. uh and you

basically lose if you assume 120 is

basically the top end of how long you

can live which is debatable but I think

there's pretty good evidence of that but

just leave that in place you know we

lose around 8 billion life years to

dying too early and an additional two

billion life years to like getting

weaker over time and then if you extend

everybody's life and then keep them very

muscular the whole time you get another

two billion so it's like 13 billion life

years on the current population that's

missing you know medical systems

underwrite at like $100,000 per life

year. Oh, but 800 million or 800 million

of those life years are lost to

accidents. So, I can't save those. Can't

stop you. The robo taxis will have to

predict prevent the car accidents. Net

net. This is like a $1.3 quadrillion

dollar market. So, it's the biggest one

I've ever mentioned.

>> It's absolutely massive. And and I agree

with you to to the point like, you know,

120 I think is is, you know, the kind of

current estimates of what, you know, a

human body could ultimately live to. I

mean, it's like a car. You know, a car

can only have, you know, so many miles

put on it. And this is kind of, you

know, the interesting thing we're we're

we're thinking about is like there's

like the the preventative piece that

gets you to that that 120. So, you're

avoiding, you know, kind of the the four

horsemen. You don't get cancer. You

don't have diabetes. You're not

overweight. you know, you get to that

120, but how do you actually, you know,

regenerate, let's say, your heart to,

you know, enable you to, you know, beat

another 40 years longer? How do you

ensure that your lungs have the

capacity? I mean, there's things that

ultimately give out just like an engine

in a car gives out. similar to like ABS

2011, how can we, you know, restimulate

the stem cell growth within the follicle

kind of elsewhere in the body to enable

us to ultimately live past that 120

years and kind of get into this more of

this regenerative longevity.

>> Yeah. And you kind of skated over it,

but the 2011 asset, it could yes, grow

more hair, but it also could like stop

graying of hair and reverse graying,

right? So, I mean, if I'm going to live

that long, I want to have brown hair the

whole way. I don't want to have to, you

know, do the little dye in there or

anything.

>> Oh, yeah. No, I mean, it's it's

incredible the pigmentation that that

comes back. Uh, and it all is because of

you you restimulating that that stem

cell. Uh, and that drives, you know, not

only the follicle growth, but also the

the melatin and ultimately the the

pigmentation. And uh

>> and so the pigmentation is the same hair

color that the person would have grown

up with. Yes,

>> exactly. So you basically go from your

gray colored hair to your naturally

occurring hair color.

>> So Sean, is it going to come out in time

to make you look less grizzled in your

beard then?

>> Yeah,

>> he looks distinguished. Just stop. Yeah,

my my my dad had a a sick burn on me the

other day. He's like, "Wow, Sean, you're

you're grain a lot faster than than I

ever uh I ever did." I was like

>> I was like, "Damn." Uh I obviously

didn't get your biology. Uh

>> well, no, it's just he didn't get your

market.

>> The biotech market is like recipe for

premature gray.

>> Yeah. But, uh, 2011, it'll be fun to

see, uh, everyone go back to their

naturally, uh, colored hair and and

actually have hair. And, you know, and

and it's really interesting. I mean, you

know, why people pay for Ozic, uh, out

of pocket is, you know, because it's

it's an identity piece. Uh, you know,

weight and and how we look is, you know,

fundamental to who we are. And same with

hair. Hair is a part of of of our

identity. and and being able to have

that that hair, you know, ultimately

makes us feel more more confident, you

know, dare I say more sexy, like we, you

know, it's just, you know, who who we

are and people are willing to pay out of

pocket to ultimately get their their

identity back. And I think like hair and

weight are, you know, a part of that.

>> Sean, so just uh out of curiosity, where

do you think if this did scale into a

much broader

regenerative

medicine and therefore scaled enjoyed uh

the economies of scale. Where could you

see the new hair color being priced or

the old hair color coming back being

priced?

>> Yeah.

>> Yeah. No, absolutely. I mean, like the

the way we're looking at pricing of of

this,

you know, once it's approved is uh some

somewhere between, you know, what you

would pay for a a Botox uh to kind of a

hair transplant. And the great part is

at least what we're seeing in the NHP,

the non-human primate study, is that

once you turn the follicle on, it stays

in that state for anywhere from 2 to 6

years based on on your biology. And so

we see it as actually being pulse

therapy. So you're you're on the drug

for six months and then you go off drug

for you know let's say two to you know

you know one to three years and then you

go back on on the drug and so we see it

as kind of uh you know pulse therapy and

you know a very durable drug. And so,

you know, if you're you're able to, you

know, get get your hair, you know, back

for, you know, a couple of years for,

you know, less than a hair transplant,

uh, you know, we we see that as as a

huge win for for the ultimate consumer.

>> Okay. How how much is a hair transplant

run? Yeah.

>> We're trying to get to the number.

>> Yes. Anywhere from, you know,5 to to

$10,000. So, 10,000 is kind of the lower

end of of a hair transplant.

>> Okay. Got that? Yeah. So, it's kind of

like you price. It's like if I do this

for a decade, maybe I'm getting up to

$10,000, but I'm paying in $2,000

installments or $3,000 installments,

that kind of thing.

>> Yeah. And that's the cool thing is that

like you could actually have it kind of

be almost like a subscription if if you

wanted um over that period of time.

There's a lot of I think interesting

kind of business models that you could

ultimately do with this type of product.

>> Yeah. It's like I subscribe for hair. I

subscribe for muscle.

>> Exactly. Yeah.

Okay. Well, Sean, it has been a pleasure

and always good to hear about kind of

the direction that AI and biotech is

going. You're at the leading edge and we

really appreciate the work that you're

doing.

>> Yeah, you're epitomizing

the most profound application of uh AI

and it's really exciting. Thank you,

Sean.

>> Yeah, thank you so much uh Kathy and uh

Brad. It's uh always fun to uh come on

on the podcast here

>> from Arc Invest. This has been FYI, the

four-year innovation podcast. Please

like, subscribe, rate us five stars on

the platform of your choice. One of

those three star middle of the road

ratings. Poravore. Uh and follow me on X

at Winton Arc. send any comments there

and I'll like try to incorporate them

into the podcast. I hope you enjoyed the

conversation with Sean and Kathy. I do

think that

drug discovery driven by AI is going to

work. I think that we will say, "Oh my

gosh, this is a great application of

these new AI models is to, you know, put

together kind of labin in the loop

proprietary data to drive better

outcomes in drug discovery.

Um I think ABSI is well positioned to

capitalize on that. It has been a

challenging couple years in biotech

generally and I think that's

disproportionately hurt um the AI drug

discovery companies that have been you

know earlier in their journey of getting

to commercialization. Um and so as with

you know if you look at the history of

drug development and kind of other novel

target areas

typically the market doubts it until it

gets to commercialization and then

suddenly realizes it has to pay full

freight.

What's interesting about AI drug

discovery is paying full freight just be

for the individual commercialized asset,

but it should ramify into an assessment

of the actual value of the platforms

where hey, they've used this model to

get this drug to market less expensively

and to commercialize.

How is somebody else going to rebuild

that model to compete with them? they'll

have to go through the same pain and

agony that Absi went through. And so

there is a momentum that could build for

these AI drug discovery platforms that

allows them to grow, you know, quite

large in relatively short order. And we

will see. I mean, drug discovery is a

process that is,

you know, driven by unexpected failures,

bad data readouts, things that go

sideways. So it'll be an interesting spa

space to watch over the next, you know,

12 and 24 months. Hope you enjoyed the

program. Again,

send us some comments, some likes, some

subscriptions. Thanks for joining FYI.

ARK believes that the information

presented is accurate and was obtained

from sources that Ark believes to be

reliable. However, ARK does not

guarantee the accuracy or completeness

of any information, and such information

may be subject to change without notice

from ARK. Historical results are not

indications of future results. Certain

of the statements contained in this

podcast may be statements of future

expectations and other forward-looking

statements that are based on ARK's

current views and assumptions and

involve known and unknown risks and

uncertainties that could cause actual

results, performance, or events to

differ materially from those expressed

or implied in such statements.

Thank you for listening to this episode

of FYI, the For Your Innovation podcast.

If you enjoyed this episode, please hit

that like button and make sure to leave

a comment down below letting us know

what you thought. And if you haven't

already, please make sure to subscribe

to the show and follow us at Arc Invest

on your favorite social platforms so you

stay up to date whenever we drop a new

episode.