Latest / The Starting Block / The US Midterms - And What It Means For Crypto

Transcript

The US Midterms - And What It Means For Crypto · The Starting Block

0:10
Welcome back to The Starting Block.
0:12
Thank you so much for joining us
0:13
wherever you are today.
0:15
My name is Gareth Jenkinson.
0:16
If you're joining us for the first
0:17
time, welcome to the show.
0:19
We've been going a little over three
0:20
months now.
0:21
Obviously started our journey here
0:24
with The Starting Block back
0:25
in Abu Dhabi with a live interview
0:27
with CZ.
0:28
And we haven't stopped since then.
0:30
Three fantastic months of
0:31
broadcasting.
0:32
We'll be continuing that today.
0:34
Really looking forward to.
0:35
A big show.
0:36
We were offline yesterday because
0:39
Bitcoin Amsterdam is in town here in
0:40
the Netherlands where I live.
0:42
So I was on stage.
0:43
Really enjoyed that.
0:44
And if you're looking for any
0:46
highlights or recaps of that, you
0:48
can go and find that on my social
0:49
media handles. But today we're going
0:50
to be jumping into another regular
0:53
show.
0:54
Really looking forward to unpacking
0:57
the top stories you need to know in
0:58
the last 24 hours in the
1:00
block briefing, which coming up
1:01
first on the show and then we're
1:03
moving on to the hot seat.
1:05
A little bit later where I'll be
1:07
introducing our guest of the day.
1:09
But if you have not been reading
1:10
headlines in the last 24 hours, let
1:12
me get you up to speed with the
1:14
top stories from the block's
1:16
fantastic editorial news team.
1:17
Starting off with our first headline
1:19
of the today, the political map just
1:21
shifted under crypto's feet.
1:23
Democrats sweat both chambers of
1:25
Congress and the industry scrambling
1:27
to rewrite its Washington playbook.
1:29
Advocacy Group stand with crypto
1:30
counts 19 pro crypto senators.
1:33
And 226 pro crypto
1:35
house members elected, but plenty
1:37
of skeptics made it in too.
1:38
The biggest last year, Ohio, where
1:40
crypto super PACs poured tens
1:42
of millions into beating Sherrod
1:44
Brown and lost.
1:45
All told crypto PACs spent a
1:47
staggering $170
1:49
million this cycle,
1:51
and analysts at TD Cowan now
1:53
say the Clarity Act is likely
1:54
stalled for the next two years.
1:56
Of course, that is just the start of
1:58
this story arc and we'll be
2:00
following that very closely here at
2:02
the block.
2:04
Then digging deeper into that Ohio
2:05
result, the digital chamber is
2:07
calling Sherrod Brown Senate win the
2:09
most consequential unfavorable
2:11
result for crypto this cycle.
2:14
Brown, a long time crypto skeptic
2:16
beat Republican John Husted.
2:18
And that's despite crypto
2:20
PAC spending more than $63
2:22
million to stop him.
2:24
With Democrats now controlling both
2:26
chambers, the group warns committee
2:27
chairs will wield subpoena power
2:29
over regulators and exchanges alike.
2:31
And the name to watch here is of
2:32
course, Elizabeth Warren, the front
2:34
runner to chair Senate banking.
2:36
And the chamber says the single
2:38
biggest obstacle to the
2:39
Clarity Act.
2:40
It's a message to the industry, stop
2:42
neglecting moderate Democrats.
2:45
Then call it resilience.
2:46
Bitcoin has clawed back past
2:48
$80,000, shrugging off
2:50
both the midterm shakeup and
2:52
the record Binance hack of last
2:54
week. On Wednesday afternoon,
2:55
Bitcoin was up three and a half
2:57
percent.
2:58
To around $80,500
3:00
while Ether jumped nearly 6%
3:02
to 2433.
3:05
Solana and XRP each added about
3:07
two and a half percent.
3:09
Roughly $469 million
3:10
in leverage bets got flushed
3:12
out in that move and the equities
3:14
followed too. Strategy and Coinbase
3:17
were up 6%
3:18
and Block Inc soaring 15
3:20
on earnings.
3:22
Bitwise's Matt Hugen says the muted
3:24
reaction to a $6 billion hack shows
3:26
a market.
3:27
That's finally maturing.
3:29
Then a major French bank has
3:30
confirmed that it was hacked.
3:32
Credit Mutuel Arkea says
3:34
a cyber attack triggered fraudulent
3:35
transfers that were funneled into
3:37
crypto through specialized
3:38
platforms.
3:39
The bank insists client assets are
3:41
untouched and the incident is
3:43
contained, calling it a delay in
3:45
certain transfers.
3:46
But on-chain investigator Zach XBT,
3:48
who flagged the exploits on November
3:49
2nd, pegs the losses at more
3:51
than $133 million
3:53
and warns that number could climb.
3:56
The bank won't confirm an exact
3:57
figure, but it validated the order
3:59
of magnitude and says some
4:01
of the stolen funds have already
4:02
been blocked.
4:04
Then Eric Trump's Bitcoin miners
4:06
back in the black.
4:07
American Bitcoin swung to $126
4:10
million profit in the third quarter,
4:12
a sharp turnaround from a $57
4:14
million loss the quarter before.
4:16
The secret here, not mining, but a
4:18
$197 million gain on
4:21
its digital asset holdings.
4:22
Because the mining side actually got
4:24
tougher, revenue slipped more than
4:25
5%, production dropped around
4:28
891 Bitcoin,
4:30
and costs per coin jumped to
4:32
roughly $42,000.
4:34
That company now holds 8,427
4:37
Bitcoin.
4:38
Still, the stock is down a brutal
4:40
76%
4:42
on the year.
4:43
Then our final story, another state
4:45
is taking aim at Kalshi.
4:47
Arkansas Attorney General Tim
4:49
Griffin has sued the prediction
4:50
market platform, I'm calling it.
4:52
Illegal gambling dressed
4:54
up as trading.
4:55
Griffin says letting users wager on
4:57
whether the Razorbacks win or lose
4:58
is squarely prohibited by state law.
5:01
And he's going further arguing
5:02
Kalshi's platform carries real
5:04
risks, excessive betting,
5:06
financial losses, even depression
5:08
and mental health harm.
5:09
Kalshih's defense as always, its
5:11
events contract are CFTC regulated
5:13
derivatives beyond the reach of
5:15
state gambling rules, but the
5:16
pressure is mounting.
5:18
Utah filed criminal charges just
5:20
last week.
5:21
And Arkansas now joins a growing
5:22
line of states squaring off
5:25
against the company.
5:26
And that is it for the block
5:27
briefing this afternoon or morning,
5:29
depending on where you're joining us
5:30
today.
5:31
We are gonna be jumping straight
5:33
into the hot seat.
5:35
It's a great pleasure to welcome
5:36
Tina Baker-Taylor to the show.
5:39
She's not a stranger to the industry
5:40
either. She's worked at pretty much
5:42
all the biggest companies across the
5:43
industry. Tina, it's great to have
5:45
you with us.
5:46
Thank you so much for joining us at
5:47
5 a.m. In the morning.
5:48
Yeah. Hey, thanks, Gareth.
5:50
It's a pleasure to be here with you.
5:52
And let's start off with the big
5:54
news.
5:55
Your personal move, obviously, into
5:57
another company that you've just
5:58
founded. You've got Based AI and
6:01
Kohai AI.
6:02
If anyone out there is not familiar
6:04
with any of this, can you give us
6:05
the TLDR on the companies,
6:07
what you're building,
6:09
for anyone interested in this
6:11
booming AI sector that we're all
6:13
watching very closely.
6:14
Sure.
6:16
So Based AI is a
6:18
portfolio company, essentially,
6:20
and Cohi is our first product.
6:24
Based AI and Cohai both are focused
6:26
on really putting
6:28
open source and open
6:30
weight AI tools
6:32
and models in the
6:34
hands of everyday people.
6:36
So similar to my last business,
6:38
Venice AI.
6:39
Which is very focused on
6:41
consumers and being able to give
6:43
consumers private sovereign AI
6:46
co-high is focused on doing
6:48
that for businesses.
6:50
So today, predominantly small
6:51
businesses under 200 people,
6:54
but very rapidly
6:56
growing. So we have enterprise
6:57
partners that are lined up to start
6:59
testing next month.
7:02
So it's all happened very quickly.
7:04
We built this product in the last
7:06
six months.
7:07
We were in stealth for quite a long
7:09
time. But the product is
7:11
live today after being in a
7:13
closed beta for several months.
7:15
And quite honestly, it's just very,
7:16
very cool.
7:17
In essence, Cohi allows you
7:19
to collaborate in
7:22
a way with AI that doesn't exist
7:24
today.
7:25
So it's a multiplayer workspace and
7:27
it is really built for teamwork.
7:29
So previous platforms
7:32
that have been launched
7:34
and out there today really
7:37
kind of limit how you're able
7:39
to work across your team and
7:41
how you work with your agentic
7:43
workforce within that platform.
7:45
It's really difficult to
7:47
share files and collaborate
7:49
together.
7:50
You have to download things, you
7:51
have to send them on Slack
7:54
and CoHi kind of removes all
7:56
of that. So it integrates all of
7:57
those channels.
7:59
It integrates with anything
8:01
that you're working on.
8:02
So if you're in working in Canva,
8:04
for example, or you're work in
8:05
GitHub.
8:07
These AI workmates can
8:09
work with you there and
8:11
they can also collaborate with your
8:13
real life teammates.
8:15
So it's a very, very powerful tool
8:18
and crucially it is built
8:20
all on open source
8:22
and it is
8:24
private and sovereign.
8:26
So unlike some of the other tools
8:28
that exist that rely on Frontier
8:30
Labs or not collecting
8:32
your data.
8:33
And we're not trading models with
8:35
it, and we are not sharing it with
8:36
advertisers or anyone else.
8:38
Tiden, can you talk to us a little
8:39
about the inception of this project
8:41
and what kind of drove you in
8:43
this direction to move away from
8:45
Venice and go in this direction,
8:47
obviously with the undercurrent that
8:49
a lot of people are interested in
8:51
running their own private AIs,
8:54
especially to the points that you've
8:55
just mentioned, not wanting to give
8:57
up their data or
8:59
just help train models of
9:02
these big AI conglomerates
9:04
that are ruling everything.
9:06
Yeah, absolutely.
9:07
So, you know, if anybody's used
9:09
Venice, it's incredibly powerful
9:11
and it has
9:13
really evolved over the last couple
9:15
of years.
9:17
Its core thesis is that
9:19
your thoughts are your own
9:21
and that you shouldn't be
9:24
guided down a path because,
9:26
you know, some safety committee
9:28
or some board of directors has
9:30
decided what you're thinking about
9:32
that might not be.
9:34
You know, copacetic or taboo
9:36
in some way.
9:37
So the idea that you
9:39
can engage directly
9:41
with the machine intelligence
9:43
and not be
9:45
compromising both,
9:47
you know your personal
9:49
information, but equally in
9:51
this case your business
9:53
IP was, is
9:55
critically important. That's the
9:57
foundation that that Venice was
9:58
built on really privacy and
10:00
a lack of censorship.
10:01
Venice's
10:03
a very unique design relies
10:05
significantly on being
10:07
locally hosted, right?
10:09
And so that makes it a little bit
10:11
difficult to be able to collaborate
10:13
in a business environment, right?
10:16
And so there was a
10:18
significant amount of appetite that
10:19
I received from small businesses
10:21
while I was at Venice that
10:23
wanted to be to use
10:25
the platform.
10:26
But in a way...
10:28
That had some
10:30
kind of guarantees, right?
10:32
Was it SOC 2 compliant?
10:34
And would we be
10:36
able to guarantee specific
10:38
things about data sharing?
10:41
And Venice can do those things,
10:43
but they're really focused on
10:44
consumers.
10:45
And so that kind of multiplayer
10:47
multi-seat experience
10:50
was somewhat challenging to do
10:51
within a locally hosted environment.
10:54
And so it's also
10:56
a very different audience that needs
10:57
very different tools.
10:58
And Venice is very kind of
11:00
self-guided, being a consumer
11:02
platform.
11:03
And so, CoHi is
11:05
built to solve some of
11:07
those challenges for
11:09
business use.
11:11
And my background was more
11:13
B2B focused than consumer.
11:15
I'd worked for a number of exchanges
11:17
and obviously focused on B2C,
11:19
but I really had a deep
11:21
understanding around how businesses
11:22
were trying to integrate these
11:24
tools.
11:25
And I think small businesses are
11:27
probably the least served in
11:29
the AI market, right?
11:30
So we hear about token maxing and
11:32
some of these firms spending a
11:34
billion dollars a month on their AI
11:37
and medium and small businesses just
11:39
don't have that money to spend.
11:40
And for them to be able to opt out
11:42
of sharing their data, they need to
11:44
be on these really, really expensive
11:46
enterprise plans that they just
11:47
can't afford.
11:48
And so they're in this position
11:50
where they can't necessarily afford.
11:52
To integrate this technology into
11:53
their business and they're scared
11:55
about being left behind.
11:57
So I think solving a
11:59
problem that nobody really is
12:01
focused on right now.
12:02
And that open source element
12:05
just provides us with a lot of
12:06
flexibility to be able to
12:09
tailor the services that we offer.
12:11
It allows people to bring in their
12:13
own agents, for example, to
12:15
compliment the suite of teammates
12:16
that are already provided within the
12:18
Clohi platform.
12:20
And then it allows us to, in
12:22
a really unique way,
12:24
provide this company brain to
12:26
the business that's agnostic to
12:28
them. And over time, this
12:30
company brand learns quite a lot
12:31
about you, but it's
12:33
protected within this enclave that
12:35
is your own.
12:36
And as time goes on, you're
12:37
essentially kind of training your
12:39
own model that is very
12:40
specific to you and your company.
12:43
So again, it's an
12:45
incredibly powerful tool.
12:47
You can try it for free at kohai.co.
12:50
Beautiful.
12:51
And immediately, the only thing that
12:52
jumps to my mind is like, what are
12:54
the barriers here and what might
12:56
slow people down?
12:57
Is it a question of hardware?
12:59
If it's open source,
13:01
kind of self-hosted AI, does that
13:02
mean you need to have a fairly
13:03
powerful computer at home or your
13:06
work needs to be providing their
13:07
workforce with a decent MacBook
13:09
or something like that to be able to
13:10
run the models themselves?
13:13
Yeah, I mean, there's certainly
13:15
models you can run by yourself at
13:16
home, a MacBook might be stretching
13:19
it. You might need to get yourself a
13:20
slightly more powerful Mac mini.
13:22
But that's really
13:24
the key challenge that we have today
13:26
with open source.
13:28
And that is kind of the
13:30
core thesis that underpins based AI
13:32
and our
13:34
intention to really
13:36
kind of expand the platform
13:38
for open AI in
13:40
its, you know, holistic sense to
13:42
thrive.
13:43
So the primary challenge today is
13:45
exactly that. Anybody can download a
13:47
model and anybody can run them.
13:48
And I highly encourage people to do
13:49
that. If you've got an agent and
13:51
you've only used a small model that
13:53
you can run yourself, totally do it.
13:54
You should absolutely do it, but
13:57
for businesses, they're going to
13:59
need models that are larger
14:02
that indeed require being
14:04
able to access that inference.
14:06
So, you know, they're certainly not
14:07
gonna run the inference themselves.
14:08
They're not gonna go out and buy GPUs
14:09
and manage those.
14:11
But for businesses a little bit
14:12
bigger, even 200 employees,
14:14
they may have an inference demand
14:15
that could potentially warrant that.
14:17
That requires a lot of technical
14:19
know-how. And if you're running an
14:20
accountancy firm or a
14:22
medical co-op of
14:25
holistic doctors, for example,
14:27
or are one of our
14:29
users today, they're
14:31
doctors.
14:32
They're not tech people.
14:34
And to hire tech people to run
14:36
the integration and
14:38
to keep the system up the
14:40
inference demand, it's a lot,
14:42
right? And it's cost.
14:44
So I think that people are looking
14:45
for a plug and play solutions where
14:47
they can get access to
14:49
frontier-like services
14:51
without the bill, open source
14:53
is considerably cheaper,
14:56
without the spying and
14:58
the data harvesting and
15:00
without having to undertake all of
15:02
that overhead and
15:05
technical debt themselves.
15:06
And so really we
15:08
see a future where open source does
15:10
become easier to use.
15:12
But today it requires quite
15:14
a lot to put all the pieces
15:15
together. And I think we have this
15:17
moment with, um, I call it the open
15:19
call moment where people realize
15:21
that they could get these agents up
15:22
and running on their own.
15:23
Um, I went through several weekends
15:25
of, of getting my clock,
15:27
Chloe on the go.
15:29
Um.
15:30
And by the end of, I
15:31
don't know, two or three weeks, I
15:32
mean, I hated her.
15:33
She was pissing me off so badly
15:35
because I had done everything right.
15:37
And yet Chloe would still not
15:39
always do it. I wanted her to do, or
15:41
she would take a little vacation and
15:42
not tell anybody.
15:43
Right. Um, and so it
15:45
was a really powerful moment that
15:46
got people, um,
15:48
excited about open source and
15:51
started to understand what the
15:52
possibilities were. But it was still
15:53
really hard even for people that
15:54
knew what they were doing.
15:55
Right. And I don't know that that
15:57
account myself as one of those
15:58
people, but I know maybe more than
16:00
some, right.
16:01
And I still struggled.
16:03
And then for me,
16:05
the big kind of lightbulb moment.
16:07
Was when Peter was aqua hired by
16:09
open AI. And I just thought, you
16:11
know what? If we don't start to
16:13
create an environment where
16:14
open source can start to build with
16:16
open source, these
16:18
incredible tools that are being
16:20
developed in open source are gonna
16:21
get co-opted by the frontier labs
16:23
and then they're gonna go away.
16:25
And so it felt a little bit like
16:27
the open source community is still
16:28
swimming up upstream.
16:31
So I just though, yeah, we need to
16:32
start building out these tools rapid
16:34
fire so that these things don't
16:36
happen.
16:37
Yeah, I was just going to say, I
16:38
think that that was my biggest
16:39
concern is that people
16:41
today like convenience.
16:45
You know, we both worked in the
16:46
crypto industry for a very long
16:48
time. And that's kind of what it
16:49
came down to as well as like
16:50
onboarding people into using
16:52
Bitcoin and other cryptocurrencies
16:53
for payments, etc.
16:55
Was just still like a stretch too
16:57
far in terms of how easy it
16:59
is to just whip out your phone and
17:00
use Apple Pay. And I think it's the
17:01
same thing with agents.
17:03
Once once we kind of cross the
17:05
threshold and.
17:06
OpenClaude became a thing, but
17:07
then, you know, not even two or
17:09
three months later, I think, you
17:11
know, um, uh,
17:13
Claude came about, you know, and
17:14
then that just like, everyone was
17:15
like, holy crap, well, I'll just pay
17:16
them a hundred dollars and then I've
17:18
got the most powerful tool out
17:20
there, but there's a lot of
17:21
trade-offs that people don't
17:22
realize. And I, I guess kind of
17:24
leads me to my next question is how
17:25
do we get more people, um
17:28
familiar and okay, and excited about
17:30
using open source AI tools
17:32
and making it easier, because I
17:33
do think that there is a barrier to
17:35
entry here that is stopping a lot
17:37
of people from even considering this
17:39
when it could be way more
17:41
beneficial to them than they
17:42
realize.
17:44
Yeah, absolutely.
17:45
So I think that Claude moment showed
17:47
us a couple of things if anybody's
17:49
paying attention. Super easy.
17:50
We know people like convenience.
17:54
And Anthropic started bringing in
17:55
all of these, these partner
17:57
companies, and then they started
17:58
developing products that cannibalize
18:00
their partner companies.
18:01
Right.
18:03
And we saw that happen one after
18:04
another, after another. And so then
18:05
companies started to think, wait a
18:06
minute, I'm putting all of my
18:08
information in here.
18:09
And then Anthropic is just turning
18:10
around and launching a competitive
18:12
product. That doesn't feel good,
18:13
right?
18:15
So, so that's one, two,
18:16
um, I said
18:18
this once and now it's kind of stuck
18:20
to me. So I guess I'll just lean
18:21
into it.
18:22
Um, I have this theory that,
18:24
you know, you don't need a Ferrari
18:25
to go to the grocery store.
18:27
And for me, that's
18:29
what businesses are doing when
18:30
they're using these frontier labs
18:31
for everything that they do, right.
18:33
You don't needs some of, you.
18:35
Don't need fable five to
18:37
be able to manage
18:39
your calendar
18:40
and to be able to organize,
18:43
you know, an offsite for your
18:45
team. And a lot of these very
18:47
administrative and kind of pedantic
18:49
things that businesses need to do
18:50
payroll.
18:53
Um, some of the things that
18:54
businesses have talked to us about
18:55
that they think are like this, you
18:56
don't critically important tasks
18:58
seem very simple, right?
18:59
And they are very simple to do.
19:01
And they're quite simple to
19:02
automate. You can have,
19:04
I think at the moment we've got a
19:05
thousand connectors within where
19:07
you can connect your obviously
19:09
your Gmail, your G Suite,
19:12
you can connect Figma,
19:15
for example, and it works within
19:17
your Figma. Right.
19:17
And it's just clicks of a button.
19:19
So that starts to make these things
19:21
easier.
19:22
My big experience in crypto, similar
19:25
to you, Gareth, was it
19:27
was hard. And sometimes I still
19:29
think it's hard.
19:30
I mean, again, I know what I'm doing
19:32
and I have made mistakes
19:34
sending things on wrong network,
19:37
right? Um, tokens never
19:39
to be seen or heard it from again,
19:40
because I use the wrong chain.
19:42
Um, and, you know, double
19:44
checking and sending a
19:46
test transaction and you don't
19:47
really, because we all make these
19:48
mistakes and we
19:50
know what we're doing, so it's,
19:52
it's a difficult ask
19:54
to get somebody to
19:57
stop using Apple pay
19:59
when all of this other kind
20:01
of mental debt comes
20:03
with using these tools.
20:05
So...
20:06
Our intention is to make this
20:08
as simple or simpler than
20:11
using Claude.
20:12
So for me, and it is really a
20:14
Claude-like experience, and some
20:16
things look the same, and some
20:17
things look very different, and they
20:18
operate very differently, because we
20:20
focus very much on the UI
20:22
being intuitive.
20:24
One of the challenges that I've had
20:25
is when you go into Claude, you'll
20:27
start a project, but then you've got
20:29
a task going somewhere else, and
20:30
these two things don't talk to each
20:32
other, and then you're not quite
20:33
sure where it was.
20:34
And then you're trying to find it
20:35
and you go, oh shit, it's actually
20:36
not a task, it is a project.
20:39
So, and then, you can't make them
20:40
like work together.
20:42
So, in Cohi, that doesn't
20:43
happen, right?
20:45
And you can find all your tasks
20:47
and you can integrate all
20:49
of your chats and puts them in your
20:51
task or your folders for you.
20:53
You don't have to go looking for
20:54
things. So, we've really tried to
20:56
take a step back and look at how
20:58
people work today and
21:00
mimic that so that they don't to
21:02
learn anything new.
21:03
That just carry on working
21:05
the way that they work with their
21:07
existing team and potentially these
21:09
new AI teammates in
21:11
channels or in
21:13
a platform that looks just like
21:15
Slack.
21:16
Yeah, beautiful.
21:17
I love it. I love the vision and I'm
21:19
excited that someone of your
21:21
caliber is out there building this
21:22
because there's definitely a demand
21:24
for it. Tina, I wanted to chat to
21:26
you a little bit about headlines
21:28
making the news this week, obviously
21:29
the US midterms.
21:31
You've been in the crypto industry a
21:32
long time. I mean, you've held some
21:35
fantastic positions and very much
21:38
policy facing as well.
21:39
I think probably most prominently
21:41
being VP of policy and
21:43
regulation at Circle, obviously
21:45
EMEA, but.
21:46
I'm sure that you have, I
21:48
don't want to say a hot take, but
21:50
I'm very curious to get your kind of
21:52
gage on the results
21:53
of the midterms and what it might
21:55
mean for the crypto industry because
21:56
it kind of seems like there's a
21:57
mixed bag and the
21:59
markets didn't quite react the way
22:01
that a lot of people thought they
22:02
were, you know, Bitcoin actually not
22:04
even, you know like dropping
22:05
drastically or anything like that.
22:07
What are your initial thoughts?
22:10
Yeah.
22:12
So I was thinking
22:14
last night, I was watching CNN or
22:16
something, and everybody was
22:20
very excited about the
22:21
blue wave.
22:22
But I remember just a few years ago,
22:24
there was a red wave, right?
22:26
And so these things are obviously
22:27
cyclical.
22:29
And I think that
22:31
especially in crypto, we need
22:33
to remember that things are
22:34
cyclical, I think
22:36
we have this tendency
22:38
to be very focused gone.
22:41
The moment, right?
22:43
And not necessarily a
22:46
horizon scan for what might happen.
22:48
So for me, I think Laura Shin
22:50
probably said it best yesterday,
22:52
you know, we need to be mindful
22:53
about supporting things like Melania
22:55
Coin and World Liberty Financial,
22:58
right.
22:59
When we
23:01
have
23:04
a path in of us.
23:06
I think we still need to kind
23:08
of check ourselves and our
23:09
integrity and call out things
23:12
that don't feel good and don't make
23:13
sense and highlight them
23:15
for what they are instead of kind of
23:17
thinking that the path is open.
23:18
Because I do think it's possible now
23:20
that we're going to have
23:22
some pushback.
23:25
It's really unfortunate
23:28
that the bills that were supposed
23:29
to pass this session didn't.
23:33
And for the most part,
23:35
I think that those bills didn't
23:36
pass. Because of those integrity
23:38
issues, right?
23:39
It was all the
23:41
fine points were really focused
23:43
on the ethics
23:45
parts of the bill
23:47
and not being able
23:49
to get comfortable with
23:51
people in positions of power being
23:53
able take advantage of a system,
23:55
right.
23:57
And so I
23:59
think that there will be a little
24:00
bit of a reckoning.
24:01
I don't know that it will
24:03
necessarily be as violent
24:05
as we're all thinking.
24:05
I mean Definitely,
24:07
Sherrod Brown is going to be
24:09
problematic and now he's
24:11
pissed off and he's
24:13
going to be pissed off with crypto.
24:14
So, you know, I think we're going to
24:16
have to take our lumps there.
24:18
But I just think, you now,
24:20
we are, I used to say a
24:22
long time ago, it feels like another
24:23
lifetime that crypto
24:25
was an infant and now
24:27
it was a toddler and people expect
24:29
toddlers to be able to sit still
24:31
and not poop their
24:33
pants.
24:34
Um, and so, you know, 11 or
24:36
12 years later or into this,
24:38
I kind of think to myself, gosh,
24:40
you know are, are we, when are
24:42
we getting it out of that toddler
24:43
stage? And things are definitely
24:44
better.
24:45
Um, the tools are more,
24:47
uh, sophisticated and
24:49
so are the majority of the companies
24:50
that are in this space, but we just
24:52
have to remember that we have to
24:53
check each other because the
24:55
companies that aren't are what we
24:57
will be judged by.
24:57
Yeah I think I
25:00
resonate very strongly with that.
25:01
We had Anthony Scaramucci on the
25:02
show last week, and I asked him a
25:04
similar kind of question.
25:05
You know, how bad,
25:07
you know, like the
25:10
very apparent.
25:11
He's such an optimist. What did he
25:12
say?
25:13
Well, I mean, I kind of asked him
25:15
for his thoughts on, you know like
25:16
the dichotomy between
25:18
how much the Trump
25:20
administration positively
25:22
impacted the cryptocurrency
25:24
industry, while
25:26
a certain group of individuals
25:27
around him.
25:29
Made a load of money and very
25:32
clearly, you know, did as
25:34
much as they could to extract value.
25:36
I mean, the meme coins aside,
25:38
there are very many business
25:39
interests that they had going on.
25:42
You know, to me, it was like, this
25:43
is going to come back and bite us.
25:44
And I think that that is what we
25:46
have to deal with now.
25:47
And I was also at Bitcoin Amsterdam,
25:50
the Bitcoin for corporations kind
25:51
of in the lead up to the main
25:53
conference today, and I was chatting
25:54
to a few people and I would very
25:55
much say like.
25:56
It's probably going to be knives out
25:58
for quite a few people in the
25:59
industry.
26:00
But at the same time, I don't think
26:01
that you can undo a lot of
26:03
the good work that's going on and
26:05
the stuff that is important and laid
26:07
a good foundation probably will
26:09
be untouched.
26:10
But there's going to be a witch hunt
26:12
and you better, you
26:14
know, like for the for the people
26:15
that weren't, you better hope
26:17
that they, you know, have have
26:18
squeaky bums and they they made sure
26:20
that everything was above board
26:21
because, yeah, things swinging
26:23
around abouts and.
26:25
That's my only concern right now.
26:27
But yeah, again, like the reaction
26:29
of the markets kind of tells you
26:30
that most people are like,
26:33
even, you know, even the blue blue
26:34
wave probably wants the best
26:36
for the crypto industry, but
26:38
that remains to be seen.
26:40
Well, hopefully the blue wave and
26:41
everybody else wants the best for
26:43
their people.
26:44
I think generally, um,
26:46
if you look at our
26:49
relationship with our governments
26:50
globally, I mean, I recently
26:52
relocated back to the, to the U
26:54
S after spending 20 plus years
26:56
in Europe.
26:57
Um, and if
26:59
you look around, um the
27:01
people's relationships with their
27:03
governments right now is fractured.
27:06
It's, it's not great across
27:08
the world.
27:09
Um, and so,
27:11
I mean, I really would like
27:13
to see the people
27:14
kind of hold their line, um,
27:17
and hold their, their policymakers
27:19
and their governments to account
27:21
in the way that they're
27:23
operating for them, right?
27:25
Right. It doesn't, it doesn't feel
27:26
like governments are working for
27:28
us. And maybe they, they never were,
27:30
but I mean it's, it's blatant
27:32
now. So the reality
27:34
is, um and stand with crypto has
27:36
been great at highlighting these
27:37
statistics.
27:39
More people own crypto now than
27:41
ever before.
27:42
More people use crypto than ever
27:44
before. If you look at Venice, for
27:46
example, the majority
27:48
of the Venice token is staked
27:50
on the platform because people use
27:52
it to create DM and
27:55
pay for their inference.
27:56
From a utility token perspective,
27:59
it truly is. And this is very
28:00
important to all of us.
28:01
When we launched it, we weren't
28:02
going to launch a token just to have
28:04
a token.
28:04
It needed to do something.
28:05
It needs to be in service of
28:07
something. And it's a very powerful
28:09
tool.
28:10
And so I think that the...
28:16
Where crypto is today,
28:18
kind of in the, not just the
28:20
financial ecosystem, but at the
28:22
zeitgeist of how we're
28:24
living and the
28:26
true kind of digitization of our
28:28
lives. I mean, AI is obviously
28:30
impacting how we think and how we
28:32
work and how are operating.
28:35
And to have the
28:37
ability for machines to transact
28:39
with each other, it's going to be
28:40
critical, right?
28:41
So I firmly believe that...
28:45
Digital money is going to,
28:47
or the ability to tokenize
28:49
anything, whether it's information
28:50
or financial, so
28:52
that these machines can interact
28:54
with us and us with them and them
28:55
with each other is really
28:57
going to be a light of like crypto's
28:59
moment, right? It's all going to
29:00
come into its own.
29:01
And I do think one of the things
29:02
that make crypto difficult for
29:04
us to deal with makes it very easy
29:06
for the machines to deal, right.
29:08
So it's not going away.
29:09
And I think the policymakers
29:10
understand that. I'm a little
29:12
concerned.
29:13
That we might kind of shift back
29:14
toward the European mentality.
29:17
When you look at kind of some of the
29:19
issues that we saw within
29:21
the MECA regulation
29:23
without having kind of clarity done,
29:26
that makes me a little bit nervous.
29:28
But again, I think that if
29:30
we hold our policymakers to
29:32
account and things like,
29:34
I'll just give a shout out to
29:35
Binance. I mean, I was the general
29:37
manager at Binance in the UK.
29:40
So I know that company well.
29:42
Um, and it didn't surprise me that
29:44
after that hack, the level of
29:46
transparency, um, the communication,
29:49
um, and, and the way that
29:50
they handled that whole thing,
29:53
um, that felt very by now
29:55
to me, they're very, very committed
29:57
to their users.
29:58
Um, but that is an excellent example
30:00
of things will go wrong.
30:02
I mean, that is life, right?
30:04
And how you handle it and how you
30:05
show up, um after the
30:07
fact and how do you mitigate.
30:09
To go forward and the relationship
30:11
that you have with your users
30:13
and the trust that you build with
30:14
your user and your counterparties
30:17
is critically important.
30:19
So I think that those things kind
30:21
of demonstrate that the industry has
30:23
significantly matured and how we're
30:25
operating, but equally
30:27
it identifies some risks, right?
30:31
And just AI is going
30:33
to be a risk, I think, and we
30:35
need to plan for that.
30:36
Yeah, I mean, that was gonna be my
30:38
final question to you.
30:39
I think the midterms might also have
30:40
some influence on the AI industry.
30:43
What are your thoughts on that?
30:44
Cause I've seen some, you know,
30:46
some Democrats kind of saying like,
30:47
we need to put a pause or we need to
30:48
put stop on AI right now.
30:50
And it's kind of like, I don't think
30:51
it works like that, but there's
30:52
certainly gonna be some pressure.
30:54
Well, we also needed to put a pause
30:56
on Bitcoin, right?
30:58
And that didn't happen.
31:00
So yeah, that's
31:02
silly.
31:03
The horses definitely left the
31:05
stable. There's no, you know,
31:07
putting it back in.
31:10
I think from my perspective there
31:13
is a feeling,
31:16
so much like with crypto, we were
31:18
trying to get people to change
31:21
their relationship with money.
31:23
Which was very ingrained, right?
31:26
And with AI, I've kind of said
31:28
from the beginning, look, this is
31:29
net new.
31:30
So let's start as we intend to
31:32
continue. And the more people we can
31:33
kind of bring in and get them
31:35
comfortable and familiar with these
31:36
tools and reduce
31:38
that learning curve later, the
31:40
better off we'll be.
31:41
And the reason I think that is
31:43
not just crypto, but if you look at
31:45
the way that social media
31:47
kind of a little bit
31:49
and then all at once completely
31:51
impacted our lives.
31:52
And I have some personal views on
31:54
ways that it has impacted humanity
31:56
that aren't healthy
31:58
or productive, right?
32:00
And so if
32:03
you're looking at
32:05
how social media was
32:07
able to really
32:09
get entrenched without
32:11
any regulatory oversight, it really
32:13
kind of goes back to section 230,
32:14
right, this ability to, it's a
32:16
free speech thing, right.
32:18
And I'm absolutely a
32:20
staunch free speech advocate.
32:22
Um, I think
32:24
that the frontier
32:26
labs saying we want to be
32:28
regulated, but not like that.
32:30
We want to, we, we want it, you
32:32
know, as you do, right.
32:34
Um, so everybody needs to be,
32:35
regulated except for me, um,
32:38
uh, is, is potentially problematic
32:40
and opening or operating in
32:42
the open source space, I think we've
32:44
had these moments where we're like,
32:45
okay, we're totally out of scope and
32:47
then, okay.
32:48
Now they want to kill open source
32:50
because it can't be controlled
32:51
because somebody at Frontier Lab
32:53
says, really, that's the danger.
32:54
Look over there, right?
32:56
Well, that is clearly not the case.
32:58
So I think, for
33:01
me, I believe that open source,
33:03
because it is more transparent,
33:06
because you have the ability to
33:08
train a model, have a cutoff
33:10
date, and it
33:12
can sit ring-fenced in its
33:14
own enclave,
33:17
will always be in a position that's
33:19
a little bit more defensible.
33:20
Than maybe the Frontier Labs and
33:23
the ability to manage
33:25
data in a more secure way
33:27
exists with open source
33:29
than it
33:31
would do with a, you know,
33:33
behind the curtain way we're
33:35
managing that data and nobody really
33:36
knows what the wizard is doing.
33:39
I'm hoping that
33:41
open source kind of stays, continues
33:43
to stay off the radar for
33:45
now.
33:47
I also think that you know, this new
33:49
government is gonna have a lot to
33:50
do. I mean, you would like
33:52
to think that they've got bigger
33:54
fish to fry right now with
33:56
things like the war and gas prices,
33:57
et cetera.
33:59
But yeah, I think we'll just have to
34:00
wait and see. But once again, I it's
34:02
important for people to be
34:04
communicating what they're doing and
34:06
engaging where it makes sense to do
34:08
so. And I always say
34:10
you're better served by having a
34:12
seat at the table and having the
34:13
discussion than by ignoring
34:16
that. A potential conversation
34:18
needs to be had.
34:19
Yeah, definitely.
34:21
Tina, this has been a fantastic
34:22
conversation.
34:23
And I'm really appreciative that you
34:24
were able to join us so early in the
34:26
morning. That's OK.
34:28
Yeah, I know.
34:28
Only for you, Garrett.
34:30
Well, next time, I will make sure
34:32
to ask, where are you in the world
34:34
so that we do it at a reasonable
34:36
hour.
34:38
Don't worry.
34:39
You touched on, Kohai, my final
34:41
question to you with Based AI.
34:43
Are there other products that you'll
34:45
be launching and looking to build
34:46
out in the future, or is it very
34:47
much just focused on Kohai right
34:49
now?
34:50
Well, so Qohai is absolutely
34:52
our primary focus over
34:54
certainly in the next couple of
34:55
weeks. But as I said, we are the
34:57
enterprise version of
35:00
Qohi, which will be very
35:02
similar, but again,
35:04
kind of different in ways around
35:06
structure, the ability
35:08
to host things on prem.
35:10
We will be able to support that if
35:12
companies want to go that route
35:14
with Ford deployed engineers and
35:16
all of the stuff that goes with
35:17
that.
35:18
Equally
35:20
Probably, I don't want to over
35:22
promise, but definitely in the next
35:23
couple of months,
35:25
based APIs will be available.
35:27
So all of the models that you're
35:29
able to access and
35:31
the harnesses, which is kind of the
35:33
key thing. Like the models are one
35:34
thing. You can go to a model and
35:37
download that to yourself.
35:38
But unless you're building a very
35:40
sophisticated harness that
35:41
understands how to extract the value
35:43
out of that model that's been tuned
35:44
for a very specific task, which is
35:46
what we've done.
35:48
Those will be accessible via the
35:50
API as well, with the same
35:52
kind of enterprise level
35:55
assurances that will come with that.
35:57
So those are a couple of things on
35:59
the immediate roadmap.
36:01
But yes, we're looking at, you know,
36:03
potentially fine tuning models for
36:06
people who are working with partners
36:07
to be able to do that so that people
36:09
can work within Cohi
36:11
or the enterprise version.
36:12
And if they want to,
36:14
in a year.
36:15
Um, take their company brain and,
36:17
you know, their fine tuned model
36:19
with them, um,
36:21
being able to be able to, to offer
36:23
that level of sovereignty to
36:24
somebody and a license agreement.
36:26
So, um they'll all be iterations
36:28
of, of, you know, kind of the,
36:30
the central theme that we're working
36:32
on, which is really kind of building
36:33
tools and services for people using
36:35
open weight models, um
36:37
that protect their sovereignty.
36:39
Beautiful.
36:40
Tina Baker-Taylor, thank you so much
36:41
for joining us.
36:42
If you're looking for more
36:43
information on based AI or
36:45
CoHi, the handles to
36:47
look out for, at based AI underscore
36:49
co on x and at
36:51
cohi underscore coon x
36:53
as well. You can also find Tina
36:55
on x at Tina Taylor.
36:58
Very, very easy to find.
36:59
Tina, thank so much.
37:00
Please enjoy the rest of your
37:01
morning. I hope the coffee
37:03
is hot.
37:04
We will catch you again soon.
37:06
Feel free to drop off the stream
37:07
while I wrap things up
37:09
here. On the show.
37:11
Thanks so much for joining us on the
37:12
starting block today.
37:13
It is a Thursday, so we'll be back
37:15
with our final show of the week
37:16
tomorrow.
37:17
I'll be teasing our next
37:19
guest on social media.
37:21
If you've missed any of the content
37:22
we've been putting out in recent
37:24
weeks, you can go and find those on
37:25
the block's YouTube channel.
37:27
It's the easiest place to find them.
37:28
Just click on the Live tab there,
37:30
and you can back through all the
37:31
shows that we've been doing.
37:32
We've been traveling around the
37:33
world a lot. We've be to New
37:35
York two times in the last month.
37:38
We were at the NASDAQ, we were also
37:39
at Avalanche Summit, we were in
37:41
Singapore for Token 2049,
37:44
we were Lugano for
37:46
Bitfinex Securities, the Future
37:48
of Capital Markets Summit, as well
37:50
as Plan B Forum.
37:51
So plenty to find there.
37:54
And a quick shout out to my
37:56
colleague, Calvin Sparks, for his
37:58
interview with Murad.
37:59
If you missed that, that's also on
38:00
our YouTube channel, as well
38:02
our X account.
38:03
So if you're on YouTube right now
38:05
watching the show, please subscribe
38:06
to our channel.
38:07
And if you're joining us on X,
38:09
wherever you are, please like and
38:10
retweet so more people can catch up
38:12
on this conversation.
38:14
And if don't already know, we
38:15
typically broadcast at 8.30 a.m.,
38:17
that's half an hour before the
38:19
opening bell in the U.S.
38:20
It's 2.30 p.m.
38:22
Central European time if you are
38:23
here in Europe like me.
38:25
We'll be back tomorrow.
38:26
Until then, take care of yourself.