Lux Capital – Investing in the Future: AI Strategies Unveiled

Invest Insight
9 followers
Hello, everyone. Welcome to MUFG In Conversation,
our series dedicated to exploring timely topics
with key industry leaders in the innovation ecosystem.
My name is Priya Rajan.
I'm Managing Director Responsible for National VC Coverage
at the MUFG Tech Banking Group.
Today, I'm excited to welcome Bilal Zubairi.
He is General Partner at Lux Capital.
The fund has over $5 billion in assets under management
and invests in emerging science and technology companies.
For nearly two decades, Bilal has been investing
in founders that solve big, important problems
across industries, enterprise software,
deep tech, healthcare, national security,
climate, and many more.
And across stages, see to grow.
Prior to Lux, he was an investor at GC,
and prior to that, he was a founder himself.
Excited to welcome Bilal,
and thank you for joining us today.
So excited to be here.
Thanks for inviting me.
Awesome.
Today's topic, we're going to be diving into,
investing into the future, the world of AI.
Before we dive into the topic, Bilal,
maybe you can do a quick intro about yourself
and about the fund as well.
Great.
So I've been at DC now for 15 plus years.
I was at General Catalyst,
and now at Lux Capital for the last 10, 11 years.
And before that, I was a founder,
built a company in advanced materials
and environmental technologies.
And before that, I did my PhD at MIT.
And before that, you know, I did my undergrad.
And before that, I came from Pakistan.
So immigrant entrepreneur turned into an investor,
investing in things that I think are important
to build for the world.
Our investment thesis across sectors and stages
rests on solving important problems for the world.
We like to call it matter that matters.
And that happens to range from investing
in biological sciences, to physical sciences,
to computer sciences, and then some.
Awesome.
And, you know, talking about AI,
you've been an early advocate of the power of AI
and its future potential.
I mentioned earlier, I've seen your panel from way back
when in early 2016, you were talking about this topic.
And since then, we have seen hyper growth here,
and especially recent years.
How do you look for the most useful insights around AI
and how best to creatively think about possibilities
at the company level?
So AI is not a new field, as we all know.
It's been around actually for decades,
but it has gone through several different lives
and every life has seen remarkable innovations
come out of it.
And we're going through a rapid one right now,
that became public about a year ago,
was known to the insiders of the industry for a few years,
that there was this emerging field around transformers
and large language models and diffusion models and stuff
that was basically technological improvements
that were happening,
that were giving outcomes that were unprecedented.
And a lot of the excitement you see around AI is that.
An hour ago, I met with a founder who said,
when he first started playing with GPT,
he felt it was magic.
And I think it is magic and amazing things happening there.
But the way we think about AI is not only to invest
where the technology is,
we see where people are going and spending their time.
You know, we like to follow incredible founders,
entrepreneurs, technologists, where they're spending time.
And that would guide us to where we should be spending time.
In addition to that, if you think of our investments in AI,
they sort of fall into four big buckets of problems
around the world.
First one of them is defending our borders
and environmental security.
Okay, this is a global need.
And if you look at companies like Anduril
and Saildrone and others,
they're using AI in various ways alongside other systems
to provide the solutions to these emerging
and very large problems, global security problems.
Then there's the democratization of creativity, right?
This is allowing everyone to become a creator,
taking away those barriers that existed
of not knowing how to use certain software and so on.
And if you look at companies like Hugging Face and Runway,
they use AI to allow everybody to have access to models,
to have consumer interfaces, to create beautiful images,
videos, movies, all kinds of things.
They're literally film festivals being held
where the movies are entirely AI generated.
It's the same thing that happened
when you and I got cell phones with cameras, right?
We all became master photographers
and then we started applying filters to it
and we all look beautiful on Instagram.
And I think that next wave of creativity is being unleashed.
The third one is adding efficiency to the human experience.
This has been the task of technology for obviously,
since the dawn of time, the same thing is happening now.
So if you look at companies like Applied Intuition,
bringing autonomous systems into mobility solutions,
or look at how to improve customer service solutions
in a very opposite end of the spectrum
in terms of the complexity of the system that they build,
but they're solving our everyday workflows
and make it more efficient.
And then the last bit is reducing human suffering
and improving our health, right?
This has everything to do with early detection
and diagnosis where AI is obviously very helpful
to then treatment and therapies
and development of novel therapies and novel products.
So we were early investors, you know,
you talked about 2016, around that time
is when we invested in Recursion,
which is now a public company,
which was one of the early first companies, in my opinion,
to using AI for pure drug discovery.
And now, you know, a person left Recursion
to start a company called NVIDIA,
which is now a drug discovery company.
We're investors in Icon and Calliope
and a bunch of others.
So if you look across all of these,
what's happening is that people are saying,
there's this incredible tool set
that we can use in so many different ways.
Let's use that to unleash our creativity,
whether it's in, you know, physical sciences,
biological sciences, or whatever,
and do things that previously were not possible.
And use them to not just be gimmicky
and be like, oh, I could do this that nobody else could,
but actually solve problems.
Because if you solve these important problems,
you can build great businesses around them,
lasting businesses that for funds like us, right,
we're trying to return billions and billions of dollars
to our LPs, we have to build businesses
that are large and global and solve important problems,
that becomes possible.
Yeah, and doing good at the end of the day as well,
and a lot of times,
especially in some of the use cases that you mentioned.
On that note, just a follow-on question,
what are like some of the opportunities related to AI
that people, and threats as well,
like both sides, opportunities and threats,
that people are not talking about,
or maybe even like another way,
thinking about it in a wrong way?
Look, there's a lot of stuff happening
on the infrastructure side,
of building infrastructure needed,
what model, what data, how to store that data,
the compute that's necessary,
all of that is very important,
so that entire stack is getting built up,
and there's new things that are emerging there,
people are not talking about or discussing,
but it is important, right,
and it relates to some of the threats as well.
I mean, the fact that we all talk about NVIDIA
as being one of the companies
that everybody knows about and uses,
it's NVIDIA today is like AWS was five, seven years ago,
right, and every company's using it,
but it feels to me like a little bit of a risk
at the same time.
You know, the fact that we don't have
domestic chip manufacturing capacity that is required,
and the fact that it sits in South China Sea,
where we don't exactly have the friendliest of relations
right now with China,
that remains a key risk for ourselves.
AI is also not something that if we have it,
nobody else does.
You know, the rest of the world is rapidly developing,
you know, in this field as well,
so then the question becomes,
to the rest of the world,
who becomes the exporters of AI-based solutions?
Not everybody needs to develop everything on their own,
but certainly there will be a fight
to be the dominant player in that space.
We know that our top tech companies
have no way of accessing the Chinese market, right?
We also know that some of the Chinese products
are sold globally, right?
If you go outside of the US,
a lot of the people regularly use Chinese-made cell phones,
right?
And some of the, you know,
a lot of data is getting captured
using those cell phones, drones, and other systems,
and that data is not always landing in hands
where they have the highest level of ethical controls
on how that data gets utilized.
So we have a lot of these things to think about.
My take on all of this becomes twofold.
One, you have to look at emerging technologies like this,
frontier technologies like this,
through the lens of national security.
And you have to understand what is happening.
How do we make sure that we move fast
and continue to maintain our dominance and our leadership?
And at the same time, how do we make sure
that at least our resources are not being used
by potentially our enemies to further themselves?
Second is to think of this
through the lens of building responsible companies.
I've been harping on about this for a while now.
Silicon Valley is no longer
just building dating websites and games.
This is not an entertainment tool only.
Even though some of the things emerge
initially as entertainment and games,
eventually they turn into things
that move people's livelihoods, national economies,
governments sometimes topple
based on what happens on the internet.
So we have to, as a community of builders and executives
and board members and so on, take responsibility.
We have to build these companies responsibly
from the grounds up on not only how the technologies
are used and secured, but also how we apply
the right levels of ethical controls on them
and manage how these systems are being used.
And it's doubly important because unfortunately,
government doesn't know about AI themselves.
People who are in the government do not understand it
and their knee-jerk reaction is likely going to be
restrictive and not a positive one for the industry
or for the technology or even for our own benefit
and good, by the way.
So I believe we have to do it ourselves,
show leadership and show that we can do all of this
without being biased, i.e. biased into our models,
without being unethical, i.e. using data
without the permissions that we should be getting
in advance and creating profits along the way
where everybody else benefits from that as well.
Great, great points.
I'm gonna just double click on the ethics part.
I mean, that's such an important topic.
A lot of leaders in AI recently went to Washington
and there's constantly,
there's like the evolving regulatory framework,
hopefully for the good.
You know, how do you advise your portfolio companies
to proactively navigate and ensure compliance
with those changing regulatory framework
and any advice around folks
that they should surround themselves with,
like advisors that should be on board
whether helping them through their,
thinking them through AI strategy,
the risk and execution?
There are a few different things to discuss here.
Number one, we're one of the largest investors
and certainly largest shareholders
in some of the most important companies in open source AI
and open source models.
We're strong supporters of open source
because we think that's where the fastest innovation
is happening.
Open source models are rapidly catching up
with some of the largest, most expensive,
closed source models,
incredible amount of innovation happening there.
So first of all, we believe that
that is the right way to do things.
Open source does not mean
that you don't have ethical consideration
or anything like that.
That's crazy.
That's never been true
and that's not gonna be true going forward.
So we think that the second thing to think about
is how do you actually think about ethics
and responsibility of all the players involved along the way
if the models are biased against people
or their backgrounds or ethnicity or whatever,
that's not okay and that's not okay socially, morally,
but also may not be okay commercially and financially.
These models get utilized in industries
where there are compliance requirements.
You cannot bias your models
or bias your results against a certain population.
So we do encourage our portfolio companies
to think about it very seriously,
to think about what is your ethics policy?
If you're going to be building systems
that could be used in weapons of war,
is there an ethics advisory board?
Should there be one?
Who should be on it?
What controls they have or don't have
and so on and so forth.
Now, none of this is meant to be,
just put a restriction on it and regulate it.
I think it's more to educate and inform
so you know and as we like to say,
failure comes from failure to imagine failure
and this is to avoid that.
The second thing is that a lot,
sorry, the third thing is that
a lot of our companies tend,
when you're solving important problems,
you tend to also in areas that are regulated.
If you're in financial industry,
if you're in medical industry,
if you are in defense tech,
there are regulations associated with all of this stuff.
So you have to have systems that you can,
both conform to compliance,
but more importantly, proactively inform
where the risks and vulnerabilities
in that system might be that could be important.
So even for commercial success of these companies,
we advise them early on to bring on executives
that are knowledgeable about all of this
to understand what are the regulations
that you might have to comply with?
What are the certifications you might need in a company?
A little bit like every SaaS company now knows
about SOC 1, SOC 2 compliance, right?
Is it not something you argue with?
The same thing out of your defense company,
you understand ITAR, you understand FedRAMP.
The same thing applies or FDA processes
in a pharmaceutical company.
So the same things will apply in more areas.
So just understand what those are.
And then the last bit is international markets.
Most of these companies, American,
part of the beauty of our technology companies
is that they become global company
and the rest of the world knows what Facebook is
and Twitter is and Uber is and Tesla is.
So the rest of the world is not necessarily going to follow
exactly what the US government does either.
So we are seeing good, bad, ugly.
We're starting to see certain frameworks emerge globally
around regulation of AI.
I mean, TikTok is banned in India, right?
It is because of the surveillance that the app was doing.
At the same time, AI regulations in Europe
tend to have recently emerged,
tend to have their own little flavor.
And I don't agree with all of the stuff
that's being done on the regulatory front.
That said, companies need to know
how to navigate those borders.
And having the right advisors such as you guys
and the investors play a key part
in kind of keeping them in check
as they're kind of growing and scaling
and navigating through these changing times.
You know, you invest across,
you see companies through stages, idea to growth.
Today, as we're talking about,
every company either is an AI company
or is a path to become an AI company.
When you're evaluating the company's claim of using AI,
what do you look for in like an attributes or metrics
and how do you pressure test it?
So, number one, we're trying to build companies
that are multi-generational, longstanding companies
and are trying to, you know, quick companies.
So they're not like, you know,
we are not interested in companies that are like,
there's a moment happening right now.
Let's build something and sell it to somebody
that A, number one, has never worked for us
or maybe never worked.
Certainly not at the kind of scale that is needed
for funds of our size.
So rather than focus on the hype,
we focus on the solution that you're building.
We try to understand the defensive moat
that you're building around your business.
They come in very shapes and sizes.
It's not always IP or this and that,
but, you know, what is the defensive moat
that you're building around it?
What is the motivation of the team to be building
what they're building?
What are they trying to accomplish?
If it's just to, you know,
have private capital fund their research, okay,
that's good to know.
May not be the right thing for us.
But there are many people who are doing something like this.
But there are other people who are like, you know,
hey, I need to build this because somebody needs this,
you know, Salesforce has this gap
and they're gonna buy a company.
That's again, that's not what we do.
But fortunately, smartest people
in the technology industry right now
are seeing this as an opportunity
to go after some of the hardest problems
that were simply intractable before.
They're trying to do things
that were just very hard to do
with the regression tables and lookup tables, right?
So they're, you know, using this to say,
how can we now go through a ---- ton of data
that is available in a lot of different sectors
and use that effectively to find solutions?
You know, and I think that's, you know,
we focus on that a lot.
When it comes to, you know,
the valuations of these companies and so on,
a lot of it is, you know,
some of it is driven by where the market is right now.
And frankly, we've walked away with a lot of comfort
from a lot of really good companies
just saying, we don't see,
we're at the end of the day, capital allocators.
We give you capital,
and then we hope to have returns back.
If we cannot see that meeting the bar
that we have set for ourselves,
you may be a great company,
but may not be the right fit for us.
So we've done that.
But at the same time,
we also look at it from the outcomes perspective,
so steps to outcomes.
You know, sometimes I find seed investors really,
seed founders frustrated
that you invested in that company, but not in us.
And we say, well, there's a lot of reasons
that go into it.
But one of those is that when we squinted our eyes
and looked at what they were building,
we could see a very large outcome.
And then you put all kinds of discounts back
and bring it back and you say,
okay, this is how we will fund it today.
But when I looked at your technology,
it was like interesting,
but not big enough to move the needle
for a fund of our size.
So I think those things start to matter quite a bit.
And, you know, good founders
are looking for good partners, not just capital.
We have, every space goes through a hype cycle
when capital becomes, you know, the cheapest commodity.
That is not where we play.
When that starts happening in any space,
frankly, Lux Capital walks out of there.
Our space is where people need partners.
And, you know, with the fast changing pace
of these companies,
especially the innovation that's happening,
you know, you've invested,
you're close to investing in a company,
you know, you kind of checked all the boxes
on the ability potential.
How do you suggest that they stay ahead of the curve?
I mean, there are so many kind of known
and unknown factors continuing to hit them.
And what would your advice be
to how they can stay ahead of what's coming next?
Yeah, so without getting into the known unknowns
and the unknowns and so on,
I would say two things.
One is part of the job that I take on
when I take on board roles in these companies
is to not so much be a Sharper,
but more like seeing around corners, right?
Founder who starts a company, in my opinion,
has earned the right to make their own mistakes, right?
They're navigating a new field,
they're doing things nobody's done before,
or at least they're doing them differently
than anybody else who may have gone into that space before.
So clearly they will face things that I have never faced
and they will make mistakes along the way.
But at least my job is to make sure
that I don't make the mistakes that others have made
that I have seen or I could see.
So help them see what's around the corner
so that at least prepared for that.
And advocating for them,
for the good solutions along the way.
The second thing around here is that,
the companies go through at least three distinct phases
in their growth.
First is when it's an idea stage company.
This is where you have a lot of excitement around
this problem I'm gonna solve
or this new technology that has come up
or this thing I've invented or whatever, right?
And that's an idea and you put a good team around it
and people like us get excited about it
and wanna give money to this.
Then you become a product stage company.
This is where you have the product market fit.
And a lot of people think of product sometimes
as like the substantiation or just creation of an idea
into something that works.
I think of product also including customers
and customers that are willing to pay you something
that creates a positive unit economics
so you can start building a business around it, right?
That's when you are really a product company.
And that's where a lot of the companies should be
when they raise a series A, B, C, something like that, right?
Good product companies then become good execution companies.
The companies were just founders who are like so focused
on building a great business.
It's not just about playing the startup.
It's not about coming up with ideas.
It's not about just building a product
and finding some customers.
It's that whole cycle and then some.
It has everything to do with being able to raise
the right kind of financing and doing it profitably
so you don't need a lot of money,
having great customer service,
having great partners along the way,
creating a machine that can then in return
take new ideas through new products
and then it becomes a virtuous cycle.
So the goal of companies should be
to eventually become those execution machines
where you can be multi-product,
even multi-sectoral companies when you become platforms,
right?
When we think of companies in technology sector today
like the Databricks,
which is one of our portfolio companies
or Stripe, which I wish was a portfolio company.
That's what these companies are.
They're platforms.
You know, if you think of it,
even in sectors that are away from these,
like Android in the defense sector
or applied intuition in the automotive sector,
these companies have become execution machines.
They have tons of ideas.
They've taken through the whole thing.
They've delivered it to the market
and continue to grow and layer and layer and layer.
And that's when they start to become
what we call multi-generational companies
and have founders understand
that that's the journey that they're on.
So to not get caught up early on in the hype cycles
or in optimizing for local minima,
that is an important thing that investors can help with.
And thank you.
That's great insight.
And the last question to wrap this up,
the execution stage of the companies that you've mentioned,
you know, especially the private ones looking to go public.
And I know the IPO kind of stalled a little bit for some
and hopefully something opens up more
in the next year or two.
Where do you see the exit environment,
especially for this sector evolve,
both as an M&A opportunity and also for IPO?
Yeah, so, you know, look,
market has decided for the time being
that large M&A is a little bit on a hold,
just as much as, you know, with few exceptions,
but large M&A on a hold as much as IPOs.
And part of that has to do
with all the antitrust concerns and so on.
And part of it is that, you know,
what these startups need right now
is a lot of money towards compute and capacity.
And these companies can provide that
so they can essentially turn their balance sheet
into their, you know, cashflow statement.
And so, but, you know, look,
there's going to be M&A in this space
because companies will inevitably
make, build or buy decisions.
A lot of focus always goes to the largest companies
like the Google, Microsoft, Amazons of the world.
But there's a whole host of companies
that will need to figure out
what to do down below that, right?
If you're a Salesforce, if you're an Oracle,
and then, you know, the whole list beyond that,
you have to be starting to think about, you know,
what kind of AI capabilities I need in-house
and how do I expand my platform
so AI is not just a tag along to what I'm doing.
So we are, we see, you know,
starting in a year or two from now
when these companies that are now
just doing 1 million to 10 million in revenues,
but they start becoming 30 to 50 million
and be gone.
Then they start to look like real businesses
where somebody will say,
hey, this is an interesting segment
that I need to be in myself
and that's where the M&A will happen.
The most important thing in all of this,
IPOs aside, because there's a whole preparatory work
that happens in IPOs.
I'm on two public boards that, you know,
company policies they invested in
and then they went public.
So that's a separate conversation.
But for M&A, M&A doesn't just happen randomly.
Now, sometimes it does.
Mosaic ML was bought by Databricks,
one of our portfolio companies,
for over a billion dollars.
And, you know, they publicly talked about
how they met at a dinner
and then fell in love with each other
and said we want to work together.
But most of the times,
you have to spend time with the people
who could be your potential acquirers,
really build that relationship
and understanding of what the needs are
so that they actually see you
not just as a technology
or not just as a revenue line,
but they see you as an executive
that they want to partner with
to be able to build the next phase
of their own company.
And for us, who's looking for large M&A,
we're not talking about tuck-in acquisitions
and acqui-hires, which happen all the time.
You know, that becomes very important.
Almost a strategic view of a company
that is 10, 20, 100 times your size,
a strategic view into how you can
transform their business.
That, when you're able to articulate that
and share that and talk about that,
that's when that company has to start
looking at you more than dollars and cents
and more of a, this is a strategic bet,
that we are willing to make.
And I think we try to help our founders
think through that.
Like every partner, every channel partner,
every customer, you know,
is there that kind of a strategic angle
for you or is it just a tool?
Amazing, amazing.
Great, great advice.
Thanks again for your time, Bilal,
and thank you for sharing such incredible insights.
And thank you all for joining us
and looking forward to seeing you soon.
Thank you, Bilal.
Thank you so much.
Thank you.
Bye.
