March 20, 2025 • 43 mins
Bloomberg’s Tim Stenovec discusses Nvidia’s growth plans with Bloomberg’s Ed Ludlow from Quantum Day at GTC. And Ampere Computing’s CEO joins to talk about the chip designer’s $6.5bn sale to SoftBank. Plus Campus CEO Tade Oyerinde and 8VC Managing Partner Joe Lonsdale on the future of higher education and Campus’ latest fundraise.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:02):
Bloomberg Audio Studios, podcasts, radio news from the heart of
where innovation, money and power collide in Silicon Valley and beyond.
This is Bloomberg Technology with Caroline Hyde and Ed Ludlow.
Speaker 2 (00:35):
Live from San Francisco.
Speaker 3 (00:37):
I'm Tim Stenebeck and this is Bloomberg Technology. Coming up
today is Quantum Day at Nvidia GtC, and Ed Ludlow
is there. We'll have the main takeaways from the conference
so far and discuss what to expect from today. Plus,
soft Bank is acquiring the chip designer Ampier in an
all cash transaction died at six point five billion dollars.
(00:58):
We'll sit down with Amper's CEO to learn more about
the deal. And Campus, a for profit community college, has
raised forty six million dollars from investors including Sam Altman,
Peter Tiele's Founder Fund, and Joe Lonsdale's eight VC. The
Founder and Joe Lonsdale join us on the future of
ed tech. First though, a check of the markets, and
(01:19):
we're seeing some green across the screen.
Speaker 2 (01:21):
Stocks opening higher.
Speaker 3 (01:22):
After yesterday's FED fueled rally or should say lower, but
then it was the best FED day going.
Speaker 2 (01:27):
Back to July.
Speaker 3 (01:28):
Then about a half hour into the trading session, we
got some surprisingly strong housing data which turned around the trade.
At last check, about sixty five stocks in the Nasdaq
one hundred moving higher right now, up about four tens
of one percent off its best levels of the day
so far, but still in the green. One of the
stocks that's helping to push the Nasdaq one hundred higher
is Meta Platforms, investors cheering the news that Meta will
(01:50):
roll out Meta AI across forty one European countries this week.
It's up right now by about four percent. Meta's intelligent
chat function will also be rolled out across twenty one
overseas territories and available in six European languages. The company
said in a statement. It's going to be free too
across its apps, including Facebook, Instagram, WhatsApp, and Messenger.
Speaker 4 (02:10):
Also watching.
Speaker 3 (02:10):
Shares of Intel down about seven tenths of one percent.
It did fall by nearly seven percent yesterday after a
TSMC board member dismissed report that the company has pitched
a major US shipmakers about taking stakes in it JV
to operate intels and factories. Shares we're bouncing back earlier,
but lower now, and in Vidia is higher today as
the company continues its GTCAI conference in San Jose today.
(02:33):
Of course, Quantum day and video shares hired by one
point three percent. And that's exactly where we go now
live from the DCC event, our own ed Ludlow joining
us at I want you to set the scene for us,
but I also want you to clear something up for us,
because there's lots of chatter this morning about in Nvidia
in the context of money being spent in the US.
Speaker 5 (02:52):
What's going on, Yeah, it's based on an interview that
Jensen Wong, the CEO, gave the Financial Times, and he
quoted as saying is that in Vidia will procure half
a trillion dollars worth of electronics in the next four
or five years. But what he's not saying is that
that is capital expenditures. Right, this is a company that
(03:13):
has seventy percent market share in the market for AI accelerators,
high performance GPUs that go into data centers. He spent
a lot of time yesterday. We were with him for
an hour explaining the mechanics of that. Right, when you
build a data center from scratch or you upgrade its technology,
that is a tens of billions of dollars or hundreds
of billions of dollars project. If you have seventy percent
(03:34):
market share for the brain that goes into it, Inevitably
you're going to have to pay TSMC to manufacture the chips.
You're going to work with Dell in HPE on the
server IRAQ assembly and packaging. That's what he's referencing, essentially
the cost of doing business. It's interesting because this is
what Gensen one wants us to be talking about. He
sees in video as foundational to all AI companies. In
(03:56):
other words, companies are being created because of what in
Video is doing. In the context of it calling itself
an AI infrastructure company, an AI factory company.
Speaker 3 (04:06):
Okay AI infrastructure a AI factory today though is quantum
day and in Nvidia doesn't actually make sure quantum computers
what's going on?
Speaker 4 (04:16):
Correct?
Speaker 5 (04:17):
Yeah, so exactly the same point as with AAI data centers.
In Vida does not make quantum computers. What it does
is sell its existing technology to the quantum computing industry
to help them make their own machines better. You can
use an AI supercomputer for error correction and calibration of
a quantum computer, but they're essentially two distinct field. Of course,
(04:40):
quantum computers follow quantum mechanics and are coded in cubits,
not in bits, ones and zeros. But we're all here
today because what happened in January, right, Jensen one was
asked basically at point blank, what do you think of
quantum computers, and he said they're more than a decade
away from being useful. The net result that day, January eighth,
was the the publicly traded quantum computing stocks all sank
(05:03):
thirty forty percent. And so we're all assembling today and
Jensen's going to be on stage with all of the
quantum computing CEOs who are basically his customers. He just
sells the existing gear to them, and maybe we'll get
an update on how Jensen feels on quantum computing. But
from in Vidia's perspective, it's an arm of research, and
it's where they sell existing tech GPUs principally to that industry.
(05:26):
They do not make quantum computers.
Speaker 3 (05:28):
Bloomberg's Ed Ludlow ed at Invidio GtC ed, good to
see you. We'll see you a little later too, Thanks
so much. Stay with us, though, because Bloomberg this Afternoon
has a special edition of Bloomberg Technology on quantum Computing.
Ed Lodlow will return live from in Video's GtC event
starting at four to thirty pm Eastern Time. Four on
the broader tech market and in Video, let's bring in
(05:50):
Sylvia Jablonski, Defiance ETF's CEO.
Speaker 2 (05:52):
She joins us.
Speaker 3 (05:53):
Now, Sylvia, I'm wondering how you're watching everything happening at
in Video GtC. You do argue that in Vidio is
a buy right now off of its highs. How are
you watching everything come out of San Jose this week?
Speaker 6 (06:07):
Yes, good morning, Thanks for having me here. I think
it's all very exciting. You know, what we're what we've
been seeing over the last.
Speaker 7 (06:14):
Year or two is just you know, so much news
around the growth of AI, the potential for quantum computing,
the buildout.
Speaker 6 (06:20):
Of six G.
Speaker 7 (06:21):
And you know who's the star of that show. It's
it's Nvidia. And you're talking about a stock that you know,
was trading up over above one forty pretty recently before
this this pullback that.
Speaker 6 (06:31):
We see here at these levels.
Speaker 7 (06:33):
I mean, I love the stock, I love the company
you're talking about potentially, you know, a fourth Industrial Revolution compounded.
I know, growth rate themes, of AI and quantum and
things like this of thirty five to forty percent per year.
As an investor, I'm patient with technology. It takes time
for things to play out, but I want to get
in early. And this is still you know, first innings.
A lot of people are saying that, but you know,
(06:54):
we're seeing that it's it's it's true as we get
more news from this conference.
Speaker 3 (06:59):
So do you think that in Nvidia investors right now
have it wrong? Does the market have it wrong? To
what extent do you think this stock is undervalued right now?
Speaker 7 (07:09):
Well, we can always say that the market is you know,
the market's a little bit emotional, right, So I think
that there are a lot of reactions in the market,
and when the market becomes emotional and people panic, usually
they sell the macseven or they sell kind of like
the high flying names that have done well for that year.
We've seen it happen with Tesla, we saw it happen
with Apple during COVID, you know, all these different sorts
(07:30):
of things. And then video is kind of the poster
child of the market this year.
Speaker 6 (07:33):
So when we have fear.
Speaker 7 (07:34):
And panics about tariffs and things like this, people tend
to run for the hills and sell you know, the
most profitable stock. So I just think that there's a
lot of liquidity on the sidelines. There's still this consumer
consumer sentiment that is uneasy.
Speaker 6 (07:47):
But eventually, you know, when.
Speaker 7 (07:49):
Some of the tariff things shake out, and then we
get you know, the tailwinds of lower regulation and tax policy,
things that are favorable to the market.
Speaker 6 (07:57):
These are the names that also rally first.
Speaker 7 (07:59):
Right, It's kind of like the buy the dips and
you know, sell the rips or hold on to the
rip scenario.
Speaker 3 (08:05):
Here, well, we're still down on the NAZAC one hundred
and ten percent, so officially in correction territory. You said,
when we get the regulatory clarity, when we get tax cuts,
when the tariff stuff shakes out, how are you so
certain that stuff is going to shake out?
Speaker 7 (08:24):
Well, I think, you know, all of these things take time, right,
and the only information we have is the information that
we actually get from Washington, and that information seems to
be that, you know, the tariff policy is inactive because
of these fensanyl issues, immigration issues, cybersecurity, these other types
of things that have to be sorted out. We have
information that the president plans to you know, cut taxes
(08:45):
to support your regulation and businesses. So I think to
your point, it's a fair question, right, we actually have
to see these things pan out. But even if nothing
else happens, right, you have an economy that.
Speaker 6 (08:56):
Is still growing positive, you know, positive GDP.
Speaker 7 (09:00):
It's a little bit lower, but we're still above that
two percent range. Jobs are fine, wages are fine. Corporate
earnings are estimated to be in the high single digits,
you know, even up to ten percent by some analyst estimates.
Speaker 6 (09:10):
The earning season was very good. There are still strong balance.
Speaker 7 (09:14):
Sheets, and you know what we're hearing out of corporate
America is that it's you know, there's soll cap bax.
Speaker 8 (09:20):
Right.
Speaker 6 (09:20):
I don't see a recession either way. So maybe you
don't get hyperbolic growth. But when you have these themes
like quantum computing and.
Speaker 7 (09:27):
AI that are on sale, I think it's worth taking,
potentially taking a risk for long term returns, regardless of
what happens in the next year or so with policy.
Speaker 3 (09:37):
Hey, Sylvia, before we let you go, just twenty seconds
on Broadcom another soccer bullet round, but down twenty percent
from all time highs.
Speaker 7 (09:45):
Yeah, I mean I think Broadcom is going to be
one of the leaders in AI and videos, the poster
child there, but Broadcom should be a second winner there.
Speaker 6 (09:52):
You know, they're in software, they're in VM sales.
Speaker 7 (09:54):
They've had over fifty seven percent growth in AI revenue
per the last earnings call. I just I think that
this is a name that did sell off a little bit.
It might be good to get in for the long run.
Speaker 3 (10:05):
Sylvia Jablonski of Defiance ETF's always good to see you, Sylvia,
thanks so much for joining us. Well, coming up, we're
going to talk about soft bank six point five billion
dollar acquisition of Chip Designer and Peer Computing and Peer
CEO Renee James joins us.
Speaker 8 (10:20):
Next.
Speaker 2 (10:20):
This is Bloomberg.
Speaker 3 (10:36):
Soft bank six point five billion dollar acquisition of Chip
Designer and Peer Computing is highlighting how the increasing demand
for compute is crashing up against infrastructure constraints. Renee James,
founder and CEO of and Pierre joins us to talk
more about this deal. Renee, good to see you, congratulations
on this.
Speaker 9 (10:54):
Thank you.
Speaker 2 (10:54):
I just want to know how are you feeling this morning.
Speaker 9 (10:59):
I'm feeling I think it hasn't sunk in.
Speaker 10 (11:02):
I'm of course thrilled with this outcome for you know,
my employees, my investors, and most importantly, we're a group
of inventors and innovators who are very excited about the
vision that Masa has for AI and our ability to
continue innovation as part of SoftBank.
Speaker 2 (11:23):
What is that vision?
Speaker 3 (11:24):
Because Ampier will operate as this wholly owned subsidiary of
soft Bank, of course it is a majority owner of ARM.
How do you fit into that vision? What is that vision?
Speaker 10 (11:35):
You know, Massa has talked a lot, even the Stargate
announcement that was recently done in the White House, about
AI and the the role AI will play in everybody's
lives and building super chips, and he's talked a lot
about that.
Speaker 9 (11:53):
So I think our role is to make that come
to life.
Speaker 10 (11:57):
We are the leading supplier of high performance, very power
efficient processors for data centers on the our architecture, so
it's a very synergistic for us to join into the
SoftBank family and continue working on the Silicon roadmap that
we have which includes AI acceleration, and now we'll have
(12:20):
a broader mandate.
Speaker 9 (12:21):
So I'm very excited about that.
Speaker 3 (12:24):
What happens to your existing customers and your existing product
line when this deal does close in the second half
of the year.
Speaker 9 (12:31):
Yeah, we continue as is.
Speaker 10 (12:33):
We continue with the product line that we've worked on
for the last eight years are very low power, high
efficiency microprocessors, and now we've announced that we have AI
acceleration in our products. So I think that's the future
of where we're going in the data center. We're going
to see compute and AI start to come together, especially
as inference becomes the larger part of the market, and
(12:57):
so our customers continue with us and hopefully we'll be
excited about a broader set of products from us.
Speaker 3 (13:05):
Well, Y, you recognize something really early on that there's
this need and there's going to be this need, and
indeed we're seeing it right now for super high performance
that required lower power. When you look across the landscape
right now and where we are in AI, what do
you see that perhaps other people don't see right now.
Speaker 9 (13:25):
Well, as you know, Tim, I've been doing this a
long time.
Speaker 10 (13:30):
I've been in semis a long time, and power is
always been a variable in semiconductors for how you get
performance or a limited performance. And so one of the
things that I didn't get to work on in my
long career at Intel was working on how to do
the highest performance possible in the most constrained power envelope.
(13:52):
That was a portion of the spectrum of computing that
we didn't really work on. And the reason the thesis
was we knew that power would be the biggest limited
to growth long term.
Speaker 9 (14:03):
There wouldn't be enough of it. You need increasing amounts
of compute.
Speaker 10 (14:07):
We've talked a lot about AI, especially with GtC this week,
taking nothing away from that, we're also having a massive
growth in compute. It's going alongside this massive growth in
AI AI compute.
Speaker 9 (14:20):
So those two things.
Speaker 10 (14:22):
Are just taking you know, a tremendous amount of growth
and power and AIS a ten x.
Speaker 9 (14:27):
You know, if you.
Speaker 10 (14:28):
Will a function growth in power required and I think
we knew that you could know that from the workloads,
and we decided that, you know, one of the things
we should go pioneer is is this sufficiency. Our architecture
is very efficient and we preserve that efficiency, but we
used our experience in high performance and building high performance
(14:51):
microprocessors to really, you know, get us to this level.
Speaker 3 (14:55):
You know, we've heard a lot from Jensen this week
about physical A and I'm wondering from your perspective. When
you think about the compute that will be needed in
the years to come, what does that world actually look
like for the normal person. What are the products and
services and tools that we are all going to be
using that will require this compute.
Speaker 6 (15:16):
You know, I used to think this is funny.
Speaker 10 (15:19):
You know, every wave of computing, whether it was the
wave of PCs, the waves of mobile phones and laptops,
we thought, this is it. We're going to have computers
that you know here, you have a computer in your pocket,
You're going to have a computer here at computer. I
think that in this next phase, you know, as was
discussed at GtC, we begin to really crest over into
(15:42):
integrated computing and everything.
Speaker 9 (15:44):
And it really is transparent.
Speaker 10 (15:47):
It's a background activity that happens in your life that's
assistive in different ways.
Speaker 9 (15:52):
Whether it's robotic or not.
Speaker 10 (15:54):
All of your appliances are smart now, all of your
homes have become smart, your.
Speaker 9 (15:58):
Car is smart.
Speaker 10 (15:59):
So the experience of computing that used to be isolated
to your computer or your phone or what have you,
is now integrated into your life and you have I think,
you know what we'll see. This is why I'm very
positive on semi conductors.
Speaker 9 (16:15):
Semiconductors have fueled every single.
Speaker 10 (16:17):
One of these waves of growth, and the base technology
to go into any kind of AI is basic computational semiconductors.
That's why, despite you know semis are always cyclical, we
do see this, we see these downturns. I'm very confident
that we have another growth cycle ahead of us in
(16:37):
semis and.
Speaker 3 (16:39):
Peter founder and CEO Renee James Renee, thanks so much
for joining us on what is certainly a really big day.
Speaker 2 (16:45):
I really appreciate it.
Speaker 3 (16:55):
Deep Seeks Innovation made ripples across the AI industry when
it announced that it's models performed as well or better
than its American counterparts and at a cheaper price.
Speaker 2 (17:04):
That was back in January.
Speaker 3 (17:06):
Since then, China and many other companies have been raising
to integrate that model throughout the country. Bloomberg's Daviting glaz
sat down with one AI c Kaifu Lead to discuss
their adoption of Deep Seek.
Speaker 11 (17:18):
Well, I think China had its Chatgibet moment when Deep
Seak came out. We can call it deep Seek moment.
Everyone's aware of it over the Chinese holidays, everyone's talking
about it, and the CEOs came back to work saying,
put put deep Seek to work and my company, and
what they found out was deep Seek is a fantastic model,
amazing AI, but it doesn't have the middleware and the
(17:42):
user interface that it takes to connect to corporate databases
to build applications to make it useful for HR finance
and customer service.
Speaker 4 (17:51):
So what zero one Dot.
Speaker 11 (17:52):
Day I did was we saw deep Seak has been
making great momentum, and we decided to really bet on
deep Seek and build a missing middleware and UI so
that deep Sea can be made.
Speaker 8 (18:04):
Useful in corporations.
Speaker 11 (18:06):
That's the product we announced this Monday, and we're getting
fantastic reception in China and.
Speaker 8 (18:12):
Also in Hong Kong.
Speaker 9 (18:13):
Tell us more about that launch.
Speaker 11 (18:14):
What we talked about was many of you have deep
Seak now you love to use it. In fact, one
CEO friend of mine asked his employees what do you
use it for?
Speaker 4 (18:25):
And good question and the answer was fortune telling.
Speaker 11 (18:28):
By the way, that's a great thing to try for you,
but it's not very deep into the industry the company.
You know, every company has ERP and the CRM databases,
they have employee records, they have their internal information, and
they want the model to be more a generalist. They
wanted to be knowledgeable. Bloomberg would want a finance knowledgeable model,
(18:52):
right Ping I would want an insurance knowledgeable model. So
our job is to really build that layer for that purpose,
sort of like if I gave a If I gave
you a Windows kernel that is the core operating system,
you wouldn't know what to do with it. You need
all the Windows layer, the application interfaces, so that the
(19:13):
Windows kernel can be useful. And we like to think
that zero one dot AI is providing that layer for deepseak,
which is the underlying model in technology.
Speaker 3 (19:24):
K That was Kai fu Lee. There's you know, Innovation
Ventures and also one AI. Meanwhile, with the advent of
the AI boom, many of the manufacturers involved are becoming
amopolistic and only ever growing companies such as Nvidia and
it's partners who make the semiconductors just keep getting richer
every time you use your favorite chatbot. That's the story
in today's quick Take, and Bloomberg's Peter Elstrom joins us
(19:47):
now Peter the team over a quick take, writing that
every time we use chat GBT or Claude or Lama.
We're making a handful of companies wealthier take us through it.
Speaker 12 (20:01):
That's right, or even deep Seek for that matter, they
also use this supply chain. So we took a look
at is this really unusual dominance that we've seen in
the supply chain of AI technologies. It begins with the
in Nvidia, which is probably the highest profile player here
in the supply chain, but it's also TSMC, the company
that makes the chips, s Kehaynix, which makes the high
(20:23):
bandwidth memory that is paired with invidious chips, and then ASML,
the maker of these lithography machines that are really the
necessary to be able to make the highest end chips
in the business.
Speaker 8 (20:35):
So what we've seen is this really.
Speaker 12 (20:36):
Consolidation of power in the AI supply chain with these
four companies where they wield tremendous power over how companies
are able to get these technologies and then deploy them.
That's true for all the hyperscalers, Meta, Microsoft, Open Ai, etc.
But also the companies in China have been trying to
buy these Now there are limitations on which chips Chinese
(20:58):
companies are able to buy, and deep Seek and even
Kaifu les zero.
Speaker 4 (21:02):
One Dot AI.
Speaker 12 (21:03):
But they want to be able to get those Nvidia
chips and the rest of the technologies from these companies
to be able to build the AI models that are
now going to be marketed to companies and to individuals.
Speaker 3 (21:13):
What is the moat that these companies have, Peter, Because
typically when we think about innovation and technology such as this,
we think about it from the perspective of, Okay, if
these companies are making money, a rush of competitors are
going to come in and they're going to try to
do the same thing.
Speaker 2 (21:27):
What's the moat that these companies have.
Speaker 12 (21:30):
Yeah, that's a very important question. And just to take
a step back, I'd say that when you look at
monopolies over time, especially monopolies in tech, they tend to
last for quite a while. We saw it with Windows
and Intel in the PC era. Before that, we saw
it with IBM, which got sued three different times for
and I trust allegations. But they tend to last for
a very long period of time, and they tend to
(21:52):
fade not just because of competition, but because of government
intervention too. Now, these AI dominating players, let's call monopolies
for now, four players that are effectively monopolies in their categories.
They've only been in place for a very short period
of time at this point. When it comes to Nvidia,
they have lots of competition. TSMC and ASML have quite
(22:13):
a bit less.
Speaker 3 (22:15):
Bloomberg's Peter Elstrom joining us from London today. Welcome back
to Bloomberg Technology. I'm Tim Seneveek in San Francisco. Let's
get a quick check of the markets. We do see
stocks off their best levels of the day. We did
(22:38):
see a lower open, but then we got some surprisingly
strong housing data which turned around the trade. Our last
check just about sixty five stocks in the Nasdaq one hundred.
Speaker 2 (22:47):
We're moving higher. A couple of.
Speaker 3 (22:49):
Individual equities I do want to check in on. Check
out what's going on with pdd Hired by about two
percent right now. These are shares listed in the US.
They erased that earlier decline, this coming after the company
reported fourth quarter results. Sales did misestimates for a third consecutive.
Speaker 2 (23:04):
Quarter, but earnings were better than expected.
Speaker 3 (23:06):
And look at Tesla down about eight tenths of one percent.
The company is recalling all cyber trucks produced and sold
in the US over the past fifteen months. This due
to a safety issue with steel trim pieces that can
detach from the vehicle. The company's recalling them all, but
it estimates that only about one percent of the recalled
vehicles have the defect. It can actually create a road
(23:27):
hazard and increase the risk of injury or collision. Now,
let's head back to in video GtC where Bloomberg's Ed
Ludlow is standing back.
Speaker 5 (23:35):
Hey, Ed, Yeah, there's just such a large volume of
news and data about in video out GtC. If you
look at the stock over the first four days of
this week. There's also skepticism in the market about the
understanding for demand long term. That's all many care about.
And I've got a brilliant guest to unpick it with me.
(23:56):
Flad Galibov is research director at Omdia. Stick that caught
my eye is that compute demand, particularly from agentic and reasoning,
is one hundred x today one hundred x what it
was one year ago. And to many people that doesn't
actually mean anything. But the way that it was explained
to me by in video is that they just counted
(24:17):
all the tokens.
Speaker 2 (24:18):
Right.
Speaker 5 (24:18):
In a tokenization context, you basically take a token three
bytes of data and you say, okay, what our companies
beyond the hyperscalers doing right now today it's one hundred
x more than it was a year ago.
Speaker 4 (24:31):
How do you model that?
Speaker 5 (24:32):
I mean, it's a very very difficult forward looking metric.
Speaker 8 (24:35):
So there's two ways to do that.
Speaker 13 (24:36):
So and by the way, there is a misunderstanding and
because it's complex, right, that's why people struggle. So one
part of my team tests models. So you know, what
they tried to find out is how good a model
can behave and they found out there is any models
behave better. The reason why they behave better is that
extra tokens is the extra computing. By being able to
(25:01):
in essence think, they actually end up getting a better result.
So my team got very excellent results from that. But
I have a different part of my team that actually
tries to understand exactly how many GPUs the different companies bought.
So my team in China was able.
Speaker 5 (25:16):
To you're going to bring us to deep Sea, can't you?
Speaker 8 (25:18):
Yes, I'll bring it to.
Speaker 13 (25:19):
Deep Sek because my team in China found out that
Deep Seek bought a huge amount of GPUs. So imagine
then they release we have given that information to our
clients they release their paper, and in the paper they say,
we don't use many GPUs for training. So my clients
immediately came in they said, why did you tell us
they bought so many GPUs And.
Speaker 8 (25:39):
I said, well, they did.
Speaker 13 (25:40):
I have the receipts, and we know now that they
bought them because their inference is so compute intensive, because
their infance uses, as you said, one hundred x small
tokens than a traditional knowledge model. But that's a good
thing for us. It's actually better for us to train
quickly and simply and have a better out put through
(26:00):
more tokens, through more reasoning. People are, especially the environmentally
conscious people, which we all should be, are very concerned
about that extra tokenization during the inference stage. But actually,
if you can get the right answer once with one question,
that stops you from having to prompt many times. You
(26:22):
might have got a good answer from CHGBT, but you
would have needed to ask it one hundred questions, so
you're doing the reasoning for it now. If you use
a reasoning model with you know, what they call a
gender GAI, you end up having the right answers traded.
Speaker 5 (26:37):
This is the core of Jensen Wong's argument right. If
you were at GtC in twenty twenty two, twenty twenty three,
twenty twenty four, maybe the fixation was on H one
hundred and training the next frontier model. But the world's
very different now. I think in videos really focus on
its enterprise customers. What Gensen one did outline was a
roadmap four years and four generations worth of hardware. Electronics
(26:59):
company the consumer electronics technology companies they don't do that.
They don't say here's what I'm doing this year all
the way through to twenty twenty seven. What did you
make of that?
Speaker 13 (27:08):
So I disagree that electronics companies don't share their roadmap,
I'll be honest, because I think if you look at AMD,
they share their roadmap pretty broadly.
Speaker 8 (27:18):
I think this is very transparent.
Speaker 13 (27:20):
You know, I come from Intel and Intel we've always
shared a roadmap pretty broadly.
Speaker 8 (27:24):
Well maybe not in the le nasty this have been
a bit shick.
Speaker 5 (27:26):
Well, there's a difference between sharing a roadmap and executing
on it.
Speaker 8 (27:29):
It is very big difference.
Speaker 13 (27:30):
I think what's amazing about Nvidia is an extreme laser
focus in this incredible culture, and they understand, they understand
the hardware stack. They understand their software stack, they understand
the services, they have a strategy, and they understand that
the world is getting tokenized, so they're focused on that.
Their laser focused on how do we make the most
efficient token processing engine.
Speaker 5 (27:53):
The analogy that gents one gave was Louis Vuitton bagged.
So for what it's worth, you argued, Louis Vuitton comes
out of this twenty five bag. But at the same day,
it doesn't tell you what it's going to be doing
in twenty twenty eight or twenty twenty nine. Whether you
agree with that that analogy or not remains to be seen.
What is different is you get a sense then videos
move beyond the hyperscalers the demand side of that equation.
(28:17):
What do you see?
Speaker 13 (28:18):
So I do think that let me just kind of
just touch on enterprises in the world for a second.
Enterprises need predictability and they like it so and actually
they've been looking ever since Intel stop delivering, they've been
looking for more partners to be honest, to give them
a roadmap, to explain things and to then deliver. So
(28:40):
I think it's actually the best way to address the enterprise.
Speaker 5 (28:42):
Does it protect the enterprise's ability to commit spending if
they know what employer signed the technology bar it does.
Speaker 13 (28:49):
In my discussion with Jensen, that's exactly what we got
into this protection of enterprise spend, this guarantee because the
investments these days are huge. But it also helps to
create an ecosystem. So what you need to make it
in the enterprise is you need an ecosystem. So over
seventy percent of it is purchased through partners, through channel partners,
(29:10):
but in the enterprise, if you zoom into just the enterprise,
it's virtually every transaction, So you really need to have
trust in partners. You need to have trust that you
know there'll be people who will help you to have
a hard ar ecosystem.
Speaker 8 (29:27):
And video.
Speaker 13 (29:29):
May make a GPU, but they work with the cooling
vendors on this exact specification of how the codeplate that
will cool it will look like.
Speaker 8 (29:36):
That's very impressive, right.
Speaker 13 (29:38):
So then when you go to the software layer, you
want to get the developers behind you. And on top
of developers, you want to also users that might.
Speaker 8 (29:46):
Not be experts.
Speaker 13 (29:47):
So by having both you know, a language, by having
a platform, by having models, that means that the different
level of skills you know, people can work with you,
and then you need to have a services prior enterprises.
Some enterprises like to do stuff in house. Other enterprises
like to have a partner.
Speaker 5 (30:07):
And we were short on time, and I've got to
mention quantum day. That's why we're here in this room.
And Video does not make quantum computers. Yes, we're having
quantum day. How do you approach it?
Speaker 13 (30:20):
So I think you know, big speculation of quantum computing.
Speaker 8 (30:23):
When is it coming?
Speaker 13 (30:24):
So I'll just tell you one very quick story. Arm
CPUs Right, arm CPUs are now a really big part
of the ecosystem. We use it, and Video has them,
Amazon has them, many people have them.
Speaker 8 (30:34):
But when it was.
Speaker 13 (30:35):
When the first kind of data center ARMCPU was launched
in about twenty eleven twenty thirteen. You know, I was
at Intel when we were very worried about it. But
at the time it lacked performance, so it took another
five years for performance to happen. But then it lacked
software ecosystem, it lacked programmability, it lacked libraries, and it
lacked you know, being able to use enterprise software out
(30:57):
of the box. So it took another five years. Huge
investment from Amazon. Actually, if we're honest for codes to
be rewritten to work. So we're now in the place
where Armor was in twenty eleven. So I think that
we need at least another five years for the hardware
to get to a place where it's highly reliable. But
(31:18):
then the programmability of it, how easy it can be popularized,
that's the difficulty. So in many ways there might be
true behind truth behind everyone sayings. Pad Guessinger thinks it's
going to take five years for the hardware, yes, but
I think Jensen thinks about it very practically. It takes
longer for the programma BOS.
Speaker 5 (31:36):
System, and Nvidia's argument would be their AI supercomputer separate
technology can help in the development. Flag Alobov of Ondia
really great to catch up here in San Jose.
Speaker 3 (31:46):
Turn back to UNSF no great stuff, big thank you
to add lod though out there in San Jose. Tune
in at four thirty pm Eastern today for a special
edition of Bloomberg Technology, hosted by our very own the
Live from Videos GtC.
Speaker 2 (32:01):
Now coming up, investor.
Speaker 3 (32:02):
Joe Lonsdale and Campus CEO today O Rende are going
to join us to discuss the startup Series B funding
round and changes to higher ed.
Speaker 2 (32:10):
This is Bloomberg, a tech startup.
Speaker 3 (32:25):
Campus has just raised forty three million dollars in a
Series B funding round. The company aims to give students
a more affordable path to a college degree. Bloomberg Beta,
the venture capital arm of Bloomberg LP, is one of
Campus's five largest shareholders. We should note joining us now
is campus CEO Toddy o ya Rende and one of
its investors, Joe Lonsdale. Toddy, I want to start with
(32:45):
you because you studied aerospace engineering in the UK and
at Embry Riddle in Florida. Was it your experience with
education that led you to start this company?
Speaker 14 (32:54):
Hey, Tim, thanks for having me. Hey Joe, definitely, I
mean before college, I was homeschooled until high school. My
paternal grandfather was a college dean. My paternal grandfather was
a high school principal. My mother is a college dean.
My older sister is a professor. So probably I was
brainwashed from birth to get really excited about education.
Speaker 3 (33:13):
Well, a school's reputation when it comes to academics is everything.
How do you build that reputation today from the ground up,
especially in the early years and when for profit schools
have had such a checkered past.
Speaker 14 (33:24):
Look, it's about elite education for all and so that's
what we're doing at Campus. We're sort of rethinking the
first two years of college. We're building a new kind
of two year college where students get to learned from
a the best professors from the top schools in the
country Princeton, Stanford, UCLA knock out the first years of
college with us, and there not just learning theoretical nonsense,
learning like really useful skills. And then they transferred to
(33:46):
the four year school of their dreams to complete their
bachelors with no debt. And I think that's the key.
Speaker 8 (33:51):
No debt.
Speaker 14 (33:52):
Student loan debt's about to pass two trillion dollars in
this country. We were hearing crazy stories students taking out
one hundred thousand dollars two hundred thousand dollar loans that
are graduating. They can't even get jobs. It makes no sense.
It's got to stop. But now there's actually a better way.
Speaker 3 (34:04):
Hey, well, speaking of that, I want to bring in
Joe to this conversation.
Speaker 2 (34:07):
Joe Lonsdale.
Speaker 3 (34:08):
Look, you've already invested in Campus, but this isn't your
first foray into education. You co founded the University of
Boston a few years ago. What in your view is
wrong with higher education? You went to Stanford, you seem
to be doing pretty well.
Speaker 4 (34:23):
Well.
Speaker 15 (34:23):
Of course, there's a lot of issues with the very
top of our education, which is what the University of
Austin's focus on. But you know, I'd argue that our
community college is unfortunately, are even more troubled in this country.
There are many have very low graduation rates. A lot
of them also are focused more on ideology than skills, sadly.
And so what today and campus represents to me, It
represents excellence, it represents merit. And you know, our economy
(34:44):
is changing drastically ais you guys are talking about on
other segments today. You know, it's changing everything how it's
going to work, and we need to get the right
skills and the right frameworks, you know, to millions of
young adults. And you know, I'm hoping Todd I could
scale this to a million students, kepture timers, sent the community,
call colledge market and really help all of those live
better lives and succeed more on the economy that it's coming.
Speaker 3 (35:05):
Let's go, well, Joe, what's your input on the curriculum,
because You're hiring a lot of folks, your portfolio companies
are hiring a lot of folks who have diverse backgrounds,
who have skills that are arguably not necessarily taught in
some schools and universities. What's the input that we're getting
him on the curriculum?
Speaker 15 (35:26):
You know, my push from my side is let's do
let's add some in more, some more courses in that
reflect what you need to know for AI.
Speaker 4 (35:33):
You know, there's people like Sam Altman involved.
Speaker 15 (35:35):
As well, who build open AI of course, and others
who are invested here. And the idea is, how can
we help hundreds of thousands, millions of young Americans, you know,
obtain the skills necessary to work in an economy where
AI is going to be involved in a lot more So,
that's not the immediate focus today. The immediate focus today
is on a lot of basic skills needed in today's economy.
But what's really fun is today's talking a lot and
(35:56):
thinking a lot about what else can we add in
here to really make sure we get people ready for
the twenty thirties.
Speaker 3 (36:01):
Well, Toddy, you were talking about the cost of college
getting out of control. You were sharing some pretty staggering
statistics about the trillions of dollars of student loan debt
that exists in this country. Nobody argues with that. How
do you make the economics of campus work though, when
other colleges and other even junior colleges community colleges.
Speaker 2 (36:21):
Are more expensive.
Speaker 14 (36:22):
Yeah, I think the key is, like the completion rates
actually have to go up for the economics to work.
So the traditional community college has an average completion rate
of about twenty seven percent graduation rate, and so when
you lose students when they drop out, you actually earn
less tuition revenue per student. So if you actually it's
sort of paradoxical. But if you actually keep students longer
(36:42):
because you help them graduate, then guess what, you earn
more tuition revenue, which makes the economics more healthy. And
so that's like the sort of the beautiful symmetry in
terms of what's best for the student, what's best for campus,
and what's best for our country. Driving up graduation rates
is actually how you make the economics work.
Speaker 3 (36:59):
Joe, come on in here, because I'm curious about your
view of the federal government's involvement in education. The government
is frozen, suspended, or cut more than a billion dollars
from universities In a recent week, we're talking about reports
from Columbia, Johns Hopkins ten and more. And I'm wondering,
as an entrepreneur, as somebody who's hired a lot of
folks who founded successful companies, somebody who has founded successful companies,
(37:23):
are you concerned about the American talent pipeline being cut
off as a result of these cuts?
Speaker 15 (37:29):
You know, I'm more concerned about making sure we spend
money effectively and efficiently, and so I really like what
Toddy is doing along those lines. A lot of the
policy I'm pushing, you know, coming from my side of things,
is how do we make this spend accountable. So, for example,
if you're going to do vocational education, Unfortunately, just like
our community colleges, a lot of the vocational programs, low
graduation rates, wrong skills, not helpful if you can spend
(37:49):
the money effectively, if you say these money is going
to be tied to results. For example, when you tie
the money to the salaries of students coming out.
Speaker 4 (37:56):
Of vocational schools, it doubles those results.
Speaker 15 (37:58):
Those are types of policies I think be popular on
both the left and the right, And what I love
about what today is doing is it's not really playing
the ideological games.
Speaker 4 (38:05):
There's people of all backgrounds.
Speaker 15 (38:06):
There's people involved from the left, from the right, black, white, Hispanic, everything,
and it's just about merit and excellence and getting good
results for very small spend. So I think this is
this sort of thing that is going to remain popular
with everyone, regardless of some of the other fights going on.
Speaker 3 (38:20):
Well, today, how do you watch what's happening in the
federal government because Predesident Trump today is expected to sign
an executive action that formally asked officials to take steps
to dismantle the Department of Education. According to our reporting,
what happens to your business then, because forty percent of
your students qualify for pelgrants and those are administered by
the Department of Education.
Speaker 14 (38:39):
Yeah, Look, the vast and jort of our students use
PEL grants to cover their twition and so they don't
have to pay anything out of pocket for twition expenses.
Paul Grant is not going away. Even if the Department
of Education is dismantled, some of these key programs that
are mandated by Congress are going to be split across.
You know, maybe Treasury or the IRS or other organizations.
The way I look at what's happening in Washington right
now is, hey, look, obviously everyone's looking at this and saying,
(39:02):
we need to be more accountable. As show talked about,
we definitely need to be more efficient with taxpayer dollars.
It's really early days. Secaturmic Man's been in there for
less than three weeks. I think you know we're watching
it closely, but we're going to have to let this
one play out.
Speaker 3 (39:13):
Hey, Joe, last one to you, speaking of efficiencies in
the federal government, You've been supportive of Dog this week,
though a federal judge ruled that Elon Musk's actions to
shut down USAID violated likely violated the Constitution in multiple ways.
Are you concerned that the courts are going to prevent
Elon from being able to do the cuts that you
want to see him?
Speaker 15 (39:32):
Do you know that particular ruling. I'm glad you mentioned
it because it was so ridiculous. So actually the ruling
was so misguided that it thought Congress had created USA,
which is not correct. It was actually created by executive
action the USA. It is just such a great example
of just complete waste, right they're just all sorts of
scams and fraud that we've uncovered. I think no matter
what your party background, if you look at the details,
(39:53):
you'll agree this should have been turned off.
Speaker 4 (39:54):
And there are.
Speaker 15 (39:55):
Activist judges they are going to try to slow it down.
I personally hope the Supreme Court is going to step
in and make some sound rulings here and stop activist
judges from violating the Constitution by their interference. And it
is going to be a big issue.
Speaker 3 (40:07):
So what have you spoken to Elon about this? Have
you spoken to Elon since he's been a doge.
Speaker 15 (40:13):
He's a friend and I am in touch, and he's
working really hard with a lot of smart people. They're
being very aggressive. A lot of my friends are involved
in DOGE and listen, there's there's I think. I don't
think everything they're doing is going to always be perfect,
but there's so many crazy things that have to be
turned off and have to be confronted that overall, I'm
very very happy for the work they're doing, and I
think they're kind of shocked about some of the ridiculous
(40:34):
things they're finding as well as they're publishing all right.
Speaker 3 (40:37):
Well, really appreciate both of you guys joining us. That's
Joe Lonsdale from a VC also campus CEO today. Oh
Yareen Day, thanks so much for joining us. If you
wanted to grind the world to a complete hall, you
could achieve.
Speaker 2 (40:50):
That by removing magnets.
Speaker 3 (40:52):
They're crucial to basically all tech, including EVS and the
next nuclear breakthrough fusion Energy Primer or the latest Bloomberg
original series takes a deep dive into all of this.
Speaker 16 (41:03):
It takes a lot of work to build something big.
You also gets very expensive, like the amount of money
you need to spend on something to get just the
first one can get very very expensive.
Speaker 17 (41:14):
That's exactly what's happening with ETA, a giant fusion reactor
currently under construction in France that uses super conducting magnets.
Look at this thing. It's huge and as a result,
it's projected to cost as much as sixty five billion dollars.
So to make fusion smaller, cheaper and more practical, Commonwealth's
founders needed a whole new kind of magnet.
Speaker 16 (41:38):
The question was like would that material ever happen? And
it wasn't until the early two thousands we could really
see that that material was going to happen that there
would be a new type of superconductor. And it's a
weird it's not a wire, it's a weird thing. It's
a film and it won the Nobel Prize like months
after it was discovered.
Speaker 18 (41:59):
So this is a material called HTS or high temperature superconductor.
Is it is Literally it comes down a tape. It's
probably kind of hard to see on the camera. It's
very thin. It's actually mostly copper and steel, but there's
a very very very thin layer inside of this that
is high temperature superconducting material.
Speaker 17 (42:17):
High temperature in this case is still wildly cold, but
not quite out of space cold. And that was the
breakthrough that Commonwealth needed. Magnets made with this material can
create a stronger magnetic field, so they don't have to
be so massive.
Speaker 18 (42:34):
You can shrink the size of the device by a
factor of ten. Basically allows us to make things smaller,
which makes things cheaper and makes things faster to get
fusion to a spot where we can make energy from it.
Speaker 3 (42:50):
And that was the voice of Caroline High. Tune into
the first episode of Primer tonight on a Bloomberg TV
at six pm Eastern time. That is going to do
it for Bloomberg Technology. Tune in later today at four
thirty pm Eastern one thirty pm Pacific for a special
edition of Bloomberg Technology Live from Nvidia's GtC event.
Speaker 2 (43:07):
Also check out our podcast.
Speaker 3 (43:09):
You can do that on the terminal, as well as
online at Apple, Spotify, and iHeart This is Bloomberg.
Bloomberg Audio Studios, podcasts, radio news from the heart of
where innovation, money and power collide in Silicon Valley and beyond.
This is Bloomberg Technology with Caroline Hyde and Ed Ludlow.
Speaker 2 (00:35):
Live from San Francisco.
Speaker 3 (00:37):
I'm Tim Stenebeck and this is Bloomberg Technology. Coming up
today is Quantum Day at Nvidia GtC, and Ed Ludlow
is there. We'll have the main takeaways from the conference
so far and discuss what to expect from today. Plus,
soft Bank is acquiring the chip designer Ampier in an
all cash transaction died at six point five billion dollars.
(00:58):
We'll sit down with Amper's CEO to learn more about
the deal. And Campus, a for profit community college, has
raised forty six million dollars from investors including Sam Altman,
Peter Tiele's Founder Fund, and Joe Lonsdale's eight VC. The
Founder and Joe Lonsdale join us on the future of
ed tech. First though, a check of the markets, and
(01:19):
we're seeing some green across the screen.
Speaker 2 (01:21):
Stocks opening higher.
Speaker 3 (01:22):
After yesterday's FED fueled rally or should say lower, but
then it was the best FED day going.
Speaker 2 (01:27):
Back to July.
Speaker 3 (01:28):
Then about a half hour into the trading session, we
got some surprisingly strong housing data which turned around the trade.
At last check, about sixty five stocks in the Nasdaq
one hundred moving higher right now, up about four tens
of one percent off its best levels of the day
so far, but still in the green. One of the
stocks that's helping to push the Nasdaq one hundred higher
is Meta Platforms, investors cheering the news that Meta will
(01:50):
roll out Meta AI across forty one European countries this week.
It's up right now by about four percent. Meta's intelligent
chat function will also be rolled out across twenty one
overseas territories and available in six European languages. The company
said in a statement. It's going to be free too
across its apps, including Facebook, Instagram, WhatsApp, and Messenger.
Speaker 4 (02:10):
Also watching.
Speaker 3 (02:10):
Shares of Intel down about seven tenths of one percent.
It did fall by nearly seven percent yesterday after a
TSMC board member dismissed report that the company has pitched
a major US shipmakers about taking stakes in it JV
to operate intels and factories. Shares we're bouncing back earlier,
but lower now, and in Vidia is higher today as
the company continues its GTCAI conference in San Jose today.
(02:33):
Of course, Quantum day and video shares hired by one
point three percent. And that's exactly where we go now
live from the DCC event, our own ed Ludlow joining
us at I want you to set the scene for us,
but I also want you to clear something up for us,
because there's lots of chatter this morning about in Nvidia
in the context of money being spent in the US.
Speaker 5 (02:52):
What's going on, Yeah, it's based on an interview that
Jensen Wong, the CEO, gave the Financial Times, and he
quoted as saying is that in Vidia will procure half
a trillion dollars worth of electronics in the next four
or five years. But what he's not saying is that
that is capital expenditures. Right, this is a company that
(03:13):
has seventy percent market share in the market for AI accelerators,
high performance GPUs that go into data centers. He spent
a lot of time yesterday. We were with him for
an hour explaining the mechanics of that. Right, when you
build a data center from scratch or you upgrade its technology,
that is a tens of billions of dollars or hundreds
of billions of dollars project. If you have seventy percent
(03:34):
market share for the brain that goes into it, Inevitably
you're going to have to pay TSMC to manufacture the chips.
You're going to work with Dell in HPE on the
server IRAQ assembly and packaging. That's what he's referencing, essentially
the cost of doing business. It's interesting because this is
what Gensen one wants us to be talking about. He
sees in video as foundational to all AI companies. In
(03:56):
other words, companies are being created because of what in
Video is doing. In the context of it calling itself
an AI infrastructure company, an AI factory company.
Speaker 3 (04:06):
Okay AI infrastructure a AI factory today though is quantum
day and in Nvidia doesn't actually make sure quantum computers
what's going on?
Speaker 4 (04:16):
Correct?
Speaker 5 (04:17):
Yeah, so exactly the same point as with AAI data centers.
In Vida does not make quantum computers. What it does
is sell its existing technology to the quantum computing industry
to help them make their own machines better. You can
use an AI supercomputer for error correction and calibration of
a quantum computer, but they're essentially two distinct field. Of course,
(04:40):
quantum computers follow quantum mechanics and are coded in cubits,
not in bits, ones and zeros. But we're all here
today because what happened in January, right, Jensen one was
asked basically at point blank, what do you think of
quantum computers, and he said they're more than a decade
away from being useful. The net result that day, January eighth,
was the the publicly traded quantum computing stocks all sank
(05:03):
thirty forty percent. And so we're all assembling today and
Jensen's going to be on stage with all of the
quantum computing CEOs who are basically his customers. He just
sells the existing gear to them, and maybe we'll get
an update on how Jensen feels on quantum computing. But
from in Vidia's perspective, it's an arm of research, and
it's where they sell existing tech GPUs principally to that industry.
(05:26):
They do not make quantum computers.
Speaker 3 (05:28):
Bloomberg's Ed Ludlow ed at Invidio GtC ed, good to
see you. We'll see you a little later too, Thanks
so much. Stay with us, though, because Bloomberg this Afternoon
has a special edition of Bloomberg Technology on quantum Computing.
Ed Lodlow will return live from in Video's GtC event
starting at four to thirty pm Eastern Time. Four on
the broader tech market and in Video, let's bring in
(05:50):
Sylvia Jablonski, Defiance ETF's CEO.
Speaker 2 (05:52):
She joins us.
Speaker 3 (05:53):
Now, Sylvia, I'm wondering how you're watching everything happening at
in Video GtC. You do argue that in Vidio is
a buy right now off of its highs. How are
you watching everything come out of San Jose this week?
Speaker 6 (06:07):
Yes, good morning, Thanks for having me here. I think
it's all very exciting. You know, what we're what we've
been seeing over the last.
Speaker 7 (06:14):
Year or two is just you know, so much news
around the growth of AI, the potential for quantum computing,
the buildout.
Speaker 6 (06:20):
Of six G.
Speaker 7 (06:21):
And you know who's the star of that show. It's
it's Nvidia. And you're talking about a stock that you know,
was trading up over above one forty pretty recently before
this this pullback that.
Speaker 6 (06:31):
We see here at these levels.
Speaker 7 (06:33):
I mean, I love the stock, I love the company
you're talking about potentially, you know, a fourth Industrial Revolution compounded.
I know, growth rate themes, of AI and quantum and
things like this of thirty five to forty percent per year.
As an investor, I'm patient with technology. It takes time
for things to play out, but I want to get
in early. And this is still you know, first innings.
A lot of people are saying that, but you know,
(06:54):
we're seeing that it's it's it's true as we get
more news from this conference.
Speaker 3 (06:59):
So do you think that in Nvidia investors right now
have it wrong? Does the market have it wrong? To
what extent do you think this stock is undervalued right now?
Speaker 7 (07:09):
Well, we can always say that the market is you know,
the market's a little bit emotional, right, So I think
that there are a lot of reactions in the market,
and when the market becomes emotional and people panic, usually
they sell the macseven or they sell kind of like
the high flying names that have done well for that year.
We've seen it happen with Tesla, we saw it happen
with Apple during COVID, you know, all these different sorts
(07:30):
of things. And then video is kind of the poster
child of the market this year.
Speaker 6 (07:33):
So when we have fear.
Speaker 7 (07:34):
And panics about tariffs and things like this, people tend
to run for the hills and sell you know, the
most profitable stock. So I just think that there's a
lot of liquidity on the sidelines. There's still this consumer
consumer sentiment that is uneasy.
Speaker 6 (07:47):
But eventually, you know, when.
Speaker 7 (07:49):
Some of the tariff things shake out, and then we
get you know, the tailwinds of lower regulation and tax policy,
things that are favorable to the market.
Speaker 6 (07:57):
These are the names that also rally first.
Speaker 7 (07:59):
Right, It's kind of like the buy the dips and
you know, sell the rips or hold on to the
rip scenario.
Speaker 3 (08:05):
Here, well, we're still down on the NAZAC one hundred
and ten percent, so officially in correction territory. You said,
when we get the regulatory clarity, when we get tax cuts,
when the tariff stuff shakes out, how are you so
certain that stuff is going to shake out?
Speaker 7 (08:24):
Well, I think, you know, all of these things take time, right,
and the only information we have is the information that
we actually get from Washington, and that information seems to
be that, you know, the tariff policy is inactive because
of these fensanyl issues, immigration issues, cybersecurity, these other types
of things that have to be sorted out. We have
information that the president plans to you know, cut taxes
(08:45):
to support your regulation and businesses. So I think to
your point, it's a fair question, right, we actually have
to see these things pan out. But even if nothing
else happens, right, you have an economy that.
Speaker 6 (08:56):
Is still growing positive, you know, positive GDP.
Speaker 7 (09:00):
It's a little bit lower, but we're still above that
two percent range. Jobs are fine, wages are fine. Corporate
earnings are estimated to be in the high single digits,
you know, even up to ten percent by some analyst estimates.
Speaker 6 (09:10):
The earning season was very good. There are still strong balance.
Speaker 7 (09:14):
Sheets, and you know what we're hearing out of corporate
America is that it's you know, there's soll cap bax.
Speaker 8 (09:20):
Right.
Speaker 6 (09:20):
I don't see a recession either way. So maybe you
don't get hyperbolic growth. But when you have these themes
like quantum computing and.
Speaker 7 (09:27):
AI that are on sale, I think it's worth taking,
potentially taking a risk for long term returns, regardless of
what happens in the next year or so with policy.
Speaker 3 (09:37):
Hey, Sylvia, before we let you go, just twenty seconds
on Broadcom another soccer bullet round, but down twenty percent
from all time highs.
Speaker 7 (09:45):
Yeah, I mean I think Broadcom is going to be
one of the leaders in AI and videos, the poster
child there, but Broadcom should be a second winner there.
Speaker 6 (09:52):
You know, they're in software, they're in VM sales.
Speaker 7 (09:54):
They've had over fifty seven percent growth in AI revenue
per the last earnings call. I just I think that
this is a name that did sell off a little bit.
It might be good to get in for the long run.
Speaker 3 (10:05):
Sylvia Jablonski of Defiance ETF's always good to see you, Sylvia,
thanks so much for joining us. Well, coming up, we're
going to talk about soft bank six point five billion
dollar acquisition of Chip Designer and Peer Computing and Peer
CEO Renee James joins us.
Speaker 8 (10:20):
Next.
Speaker 2 (10:20):
This is Bloomberg.
Speaker 3 (10:36):
Soft bank six point five billion dollar acquisition of Chip
Designer and Peer Computing is highlighting how the increasing demand
for compute is crashing up against infrastructure constraints. Renee James,
founder and CEO of and Pierre joins us to talk
more about this deal. Renee, good to see you, congratulations
on this.
Speaker 9 (10:54):
Thank you.
Speaker 2 (10:54):
I just want to know how are you feeling this morning.
Speaker 9 (10:59):
I'm feeling I think it hasn't sunk in.
Speaker 10 (11:02):
I'm of course thrilled with this outcome for you know,
my employees, my investors, and most importantly, we're a group
of inventors and innovators who are very excited about the
vision that Masa has for AI and our ability to
continue innovation as part of SoftBank.
Speaker 2 (11:23):
What is that vision?
Speaker 3 (11:24):
Because Ampier will operate as this wholly owned subsidiary of
soft Bank, of course it is a majority owner of ARM.
How do you fit into that vision? What is that vision?
Speaker 10 (11:35):
You know, Massa has talked a lot, even the Stargate
announcement that was recently done in the White House, about
AI and the the role AI will play in everybody's
lives and building super chips, and he's talked a lot
about that.
Speaker 9 (11:53):
So I think our role is to make that come
to life.
Speaker 10 (11:57):
We are the leading supplier of high performance, very power
efficient processors for data centers on the our architecture, so
it's a very synergistic for us to join into the
SoftBank family and continue working on the Silicon roadmap that
we have which includes AI acceleration, and now we'll have
(12:20):
a broader mandate.
Speaker 9 (12:21):
So I'm very excited about that.
Speaker 3 (12:24):
What happens to your existing customers and your existing product
line when this deal does close in the second half
of the year.
Speaker 9 (12:31):
Yeah, we continue as is.
Speaker 10 (12:33):
We continue with the product line that we've worked on
for the last eight years are very low power, high
efficiency microprocessors, and now we've announced that we have AI
acceleration in our products. So I think that's the future
of where we're going in the data center. We're going
to see compute and AI start to come together, especially
as inference becomes the larger part of the market, and
(12:57):
so our customers continue with us and hopefully we'll be
excited about a broader set of products from us.
Speaker 3 (13:05):
Well, Y, you recognize something really early on that there's
this need and there's going to be this need, and
indeed we're seeing it right now for super high performance
that required lower power. When you look across the landscape
right now and where we are in AI, what do
you see that perhaps other people don't see right now.
Speaker 9 (13:25):
Well, as you know, Tim, I've been doing this a
long time.
Speaker 10 (13:30):
I've been in semis a long time, and power is
always been a variable in semiconductors for how you get
performance or a limited performance. And so one of the
things that I didn't get to work on in my
long career at Intel was working on how to do
the highest performance possible in the most constrained power envelope.
(13:52):
That was a portion of the spectrum of computing that
we didn't really work on. And the reason the thesis
was we knew that power would be the biggest limited
to growth long term.
Speaker 9 (14:03):
There wouldn't be enough of it. You need increasing amounts
of compute.
Speaker 10 (14:07):
We've talked a lot about AI, especially with GtC this week,
taking nothing away from that, we're also having a massive
growth in compute. It's going alongside this massive growth in
AI AI compute.
Speaker 9 (14:20):
So those two things.
Speaker 10 (14:22):
Are just taking you know, a tremendous amount of growth
and power and AIS a ten x.
Speaker 9 (14:27):
You know, if you.
Speaker 10 (14:28):
Will a function growth in power required and I think
we knew that you could know that from the workloads,
and we decided that, you know, one of the things
we should go pioneer is is this sufficiency. Our architecture
is very efficient and we preserve that efficiency, but we
used our experience in high performance and building high performance
(14:51):
microprocessors to really, you know, get us to this level.
Speaker 3 (14:55):
You know, we've heard a lot from Jensen this week
about physical A and I'm wondering from your perspective. When
you think about the compute that will be needed in
the years to come, what does that world actually look
like for the normal person. What are the products and
services and tools that we are all going to be
using that will require this compute.
Speaker 6 (15:16):
You know, I used to think this is funny.
Speaker 10 (15:19):
You know, every wave of computing, whether it was the
wave of PCs, the waves of mobile phones and laptops,
we thought, this is it. We're going to have computers
that you know here, you have a computer in your pocket,
You're going to have a computer here at computer. I
think that in this next phase, you know, as was
discussed at GtC, we begin to really crest over into
(15:42):
integrated computing and everything.
Speaker 9 (15:44):
And it really is transparent.
Speaker 10 (15:47):
It's a background activity that happens in your life that's
assistive in different ways.
Speaker 9 (15:52):
Whether it's robotic or not.
Speaker 10 (15:54):
All of your appliances are smart now, all of your
homes have become smart, your.
Speaker 9 (15:58):
Car is smart.
Speaker 10 (15:59):
So the experience of computing that used to be isolated
to your computer or your phone or what have you,
is now integrated into your life and you have I think,
you know what we'll see. This is why I'm very
positive on semi conductors.
Speaker 9 (16:15):
Semiconductors have fueled every single.
Speaker 10 (16:17):
One of these waves of growth, and the base technology
to go into any kind of AI is basic computational semiconductors.
That's why, despite you know semis are always cyclical, we
do see this, we see these downturns. I'm very confident
that we have another growth cycle ahead of us in
(16:37):
semis and.
Speaker 3 (16:39):
Peter founder and CEO Renee James Renee, thanks so much
for joining us on what is certainly a really big day.
Speaker 2 (16:45):
I really appreciate it.
Speaker 3 (16:55):
Deep Seeks Innovation made ripples across the AI industry when
it announced that it's models performed as well or better
than its American counterparts and at a cheaper price.
Speaker 2 (17:04):
That was back in January.
Speaker 3 (17:06):
Since then, China and many other companies have been raising
to integrate that model throughout the country. Bloomberg's Daviting glaz
sat down with one AI c Kaifu Lead to discuss
their adoption of Deep Seek.
Speaker 11 (17:18):
Well, I think China had its Chatgibet moment when Deep
Seak came out. We can call it deep Seek moment.
Everyone's aware of it over the Chinese holidays, everyone's talking
about it, and the CEOs came back to work saying,
put put deep Seek to work and my company, and
what they found out was deep Seek is a fantastic model,
amazing AI, but it doesn't have the middleware and the
(17:42):
user interface that it takes to connect to corporate databases
to build applications to make it useful for HR finance
and customer service.
Speaker 4 (17:51):
So what zero one Dot.
Speaker 11 (17:52):
Day I did was we saw deep Seak has been
making great momentum, and we decided to really bet on
deep Seek and build a missing middleware and UI so
that deep Sea can be made.
Speaker 8 (18:04):
Useful in corporations.
Speaker 11 (18:06):
That's the product we announced this Monday, and we're getting
fantastic reception in China and.
Speaker 8 (18:12):
Also in Hong Kong.
Speaker 9 (18:13):
Tell us more about that launch.
Speaker 11 (18:14):
What we talked about was many of you have deep
Seak now you love to use it. In fact, one
CEO friend of mine asked his employees what do you
use it for?
Speaker 4 (18:25):
And good question and the answer was fortune telling.
Speaker 11 (18:28):
By the way, that's a great thing to try for you,
but it's not very deep into the industry the company.
You know, every company has ERP and the CRM databases,
they have employee records, they have their internal information, and
they want the model to be more a generalist. They
wanted to be knowledgeable. Bloomberg would want a finance knowledgeable model,
(18:52):
right Ping I would want an insurance knowledgeable model. So
our job is to really build that layer for that purpose,
sort of like if I gave a If I gave
you a Windows kernel that is the core operating system,
you wouldn't know what to do with it. You need
all the Windows layer, the application interfaces, so that the
(19:13):
Windows kernel can be useful. And we like to think
that zero one dot AI is providing that layer for deepseak,
which is the underlying model in technology.
Speaker 3 (19:24):
K That was Kai fu Lee. There's you know, Innovation
Ventures and also one AI. Meanwhile, with the advent of
the AI boom, many of the manufacturers involved are becoming
amopolistic and only ever growing companies such as Nvidia and
it's partners who make the semiconductors just keep getting richer
every time you use your favorite chatbot. That's the story
in today's quick Take, and Bloomberg's Peter Elstrom joins us
(19:47):
now Peter the team over a quick take, writing that
every time we use chat GBT or Claude or Lama.
We're making a handful of companies wealthier take us through it.
Speaker 12 (20:01):
That's right, or even deep Seek for that matter, they
also use this supply chain. So we took a look
at is this really unusual dominance that we've seen in
the supply chain of AI technologies. It begins with the
in Nvidia, which is probably the highest profile player here
in the supply chain, but it's also TSMC, the company
that makes the chips, s Kehaynix, which makes the high
(20:23):
bandwidth memory that is paired with invidious chips, and then ASML,
the maker of these lithography machines that are really the
necessary to be able to make the highest end chips
in the business.
Speaker 8 (20:35):
So what we've seen is this really.
Speaker 12 (20:36):
Consolidation of power in the AI supply chain with these
four companies where they wield tremendous power over how companies
are able to get these technologies and then deploy them.
That's true for all the hyperscalers, Meta, Microsoft, Open Ai, etc.
But also the companies in China have been trying to
buy these Now there are limitations on which chips Chinese
(20:58):
companies are able to buy, and deep Seek and even
Kaifu les zero.
Speaker 4 (21:02):
One Dot AI.
Speaker 12 (21:03):
But they want to be able to get those Nvidia
chips and the rest of the technologies from these companies
to be able to build the AI models that are
now going to be marketed to companies and to individuals.
Speaker 3 (21:13):
What is the moat that these companies have, Peter, Because
typically when we think about innovation and technology such as this,
we think about it from the perspective of, Okay, if
these companies are making money, a rush of competitors are
going to come in and they're going to try to
do the same thing.
Speaker 2 (21:27):
What's the moat that these companies have.
Speaker 12 (21:30):
Yeah, that's a very important question. And just to take
a step back, I'd say that when you look at
monopolies over time, especially monopolies in tech, they tend to
last for quite a while. We saw it with Windows
and Intel in the PC era. Before that, we saw
it with IBM, which got sued three different times for
and I trust allegations. But they tend to last for
a very long period of time, and they tend to
(21:52):
fade not just because of competition, but because of government
intervention too. Now, these AI dominating players, let's call monopolies
for now, four players that are effectively monopolies in their categories.
They've only been in place for a very short period
of time at this point. When it comes to Nvidia,
they have lots of competition. TSMC and ASML have quite
(22:13):
a bit less.
Speaker 3 (22:15):
Bloomberg's Peter Elstrom joining us from London today. Welcome back
to Bloomberg Technology. I'm Tim Seneveek in San Francisco. Let's
get a quick check of the markets. We do see
stocks off their best levels of the day. We did
(22:38):
see a lower open, but then we got some surprisingly
strong housing data which turned around the trade. Our last
check just about sixty five stocks in the Nasdaq one hundred.
Speaker 2 (22:47):
We're moving higher. A couple of.
Speaker 3 (22:49):
Individual equities I do want to check in on. Check
out what's going on with pdd Hired by about two
percent right now. These are shares listed in the US.
They erased that earlier decline, this coming after the company
reported fourth quarter results. Sales did misestimates for a third consecutive.
Speaker 2 (23:04):
Quarter, but earnings were better than expected.
Speaker 3 (23:06):
And look at Tesla down about eight tenths of one percent.
The company is recalling all cyber trucks produced and sold
in the US over the past fifteen months. This due
to a safety issue with steel trim pieces that can
detach from the vehicle. The company's recalling them all, but
it estimates that only about one percent of the recalled
vehicles have the defect. It can actually create a road
(23:27):
hazard and increase the risk of injury or collision. Now,
let's head back to in video GtC where Bloomberg's Ed
Ludlow is standing back.
Speaker 5 (23:35):
Hey, Ed, Yeah, there's just such a large volume of
news and data about in video out GtC. If you
look at the stock over the first four days of
this week. There's also skepticism in the market about the
understanding for demand long term. That's all many care about.
And I've got a brilliant guest to unpick it with me.
(23:56):
Flad Galibov is research director at Omdia. Stick that caught
my eye is that compute demand, particularly from agentic and reasoning,
is one hundred x today one hundred x what it
was one year ago. And to many people that doesn't
actually mean anything. But the way that it was explained
to me by in video is that they just counted
(24:17):
all the tokens.
Speaker 2 (24:18):
Right.
Speaker 5 (24:18):
In a tokenization context, you basically take a token three
bytes of data and you say, okay, what our companies
beyond the hyperscalers doing right now today it's one hundred
x more than it was a year ago.
Speaker 4 (24:31):
How do you model that?
Speaker 5 (24:32):
I mean, it's a very very difficult forward looking metric.
Speaker 8 (24:35):
So there's two ways to do that.
Speaker 13 (24:36):
So and by the way, there is a misunderstanding and
because it's complex, right, that's why people struggle. So one
part of my team tests models. So you know, what
they tried to find out is how good a model
can behave and they found out there is any models
behave better. The reason why they behave better is that
extra tokens is the extra computing. By being able to
(25:01):
in essence think, they actually end up getting a better result.
So my team got very excellent results from that. But
I have a different part of my team that actually
tries to understand exactly how many GPUs the different companies bought.
So my team in China was able.
Speaker 5 (25:16):
To you're going to bring us to deep Sea, can't you?
Speaker 8 (25:18):
Yes, I'll bring it to.
Speaker 13 (25:19):
Deep Sek because my team in China found out that
Deep Seek bought a huge amount of GPUs. So imagine
then they release we have given that information to our
clients they release their paper, and in the paper they say,
we don't use many GPUs for training. So my clients
immediately came in they said, why did you tell us
they bought so many GPUs And.
Speaker 8 (25:39):
I said, well, they did.
Speaker 13 (25:40):
I have the receipts, and we know now that they
bought them because their inference is so compute intensive, because
their infance uses, as you said, one hundred x small
tokens than a traditional knowledge model. But that's a good
thing for us. It's actually better for us to train
quickly and simply and have a better out put through
(26:00):
more tokens, through more reasoning. People are, especially the environmentally
conscious people, which we all should be, are very concerned
about that extra tokenization during the inference stage. But actually,
if you can get the right answer once with one question,
that stops you from having to prompt many times. You
(26:22):
might have got a good answer from CHGBT, but you
would have needed to ask it one hundred questions, so
you're doing the reasoning for it now. If you use
a reasoning model with you know, what they call a
gender GAI, you end up having the right answers traded.
Speaker 5 (26:37):
This is the core of Jensen Wong's argument right. If
you were at GtC in twenty twenty two, twenty twenty three,
twenty twenty four, maybe the fixation was on H one
hundred and training the next frontier model. But the world's
very different now. I think in videos really focus on
its enterprise customers. What Gensen one did outline was a
roadmap four years and four generations worth of hardware. Electronics
(26:59):
company the consumer electronics technology companies they don't do that.
They don't say here's what I'm doing this year all
the way through to twenty twenty seven. What did you
make of that?
Speaker 13 (27:08):
So I disagree that electronics companies don't share their roadmap,
I'll be honest, because I think if you look at AMD,
they share their roadmap pretty broadly.
Speaker 8 (27:18):
I think this is very transparent.
Speaker 13 (27:20):
You know, I come from Intel and Intel we've always
shared a roadmap pretty broadly.
Speaker 8 (27:24):
Well maybe not in the le nasty this have been
a bit shick.
Speaker 5 (27:26):
Well, there's a difference between sharing a roadmap and executing
on it.
Speaker 8 (27:29):
It is very big difference.
Speaker 13 (27:30):
I think what's amazing about Nvidia is an extreme laser
focus in this incredible culture, and they understand, they understand
the hardware stack. They understand their software stack, they understand
the services, they have a strategy, and they understand that
the world is getting tokenized, so they're focused on that.
Their laser focused on how do we make the most
efficient token processing engine.
Speaker 5 (27:53):
The analogy that gents one gave was Louis Vuitton bagged.
So for what it's worth, you argued, Louis Vuitton comes
out of this twenty five bag. But at the same day,
it doesn't tell you what it's going to be doing
in twenty twenty eight or twenty twenty nine. Whether you
agree with that that analogy or not remains to be seen.
What is different is you get a sense then videos
move beyond the hyperscalers the demand side of that equation.
(28:17):
What do you see?
Speaker 13 (28:18):
So I do think that let me just kind of
just touch on enterprises in the world for a second.
Enterprises need predictability and they like it so and actually
they've been looking ever since Intel stop delivering, they've been
looking for more partners to be honest, to give them
a roadmap, to explain things and to then deliver. So
(28:40):
I think it's actually the best way to address the enterprise.
Speaker 5 (28:42):
Does it protect the enterprise's ability to commit spending if
they know what employer signed the technology bar it does.
Speaker 13 (28:49):
In my discussion with Jensen, that's exactly what we got
into this protection of enterprise spend, this guarantee because the
investments these days are huge. But it also helps to
create an ecosystem. So what you need to make it
in the enterprise is you need an ecosystem. So over
seventy percent of it is purchased through partners, through channel partners,
(29:10):
but in the enterprise, if you zoom into just the enterprise,
it's virtually every transaction, So you really need to have
trust in partners. You need to have trust that you
know there'll be people who will help you to have
a hard ar ecosystem.
Speaker 8 (29:27):
And video.
Speaker 13 (29:29):
May make a GPU, but they work with the cooling
vendors on this exact specification of how the codeplate that
will cool it will look like.
Speaker 8 (29:36):
That's very impressive, right.
Speaker 13 (29:38):
So then when you go to the software layer, you
want to get the developers behind you. And on top
of developers, you want to also users that might.
Speaker 8 (29:46):
Not be experts.
Speaker 13 (29:47):
So by having both you know, a language, by having
a platform, by having models, that means that the different
level of skills you know, people can work with you,
and then you need to have a services prior enterprises.
Some enterprises like to do stuff in house. Other enterprises
like to have a partner.
Speaker 5 (30:07):
And we were short on time, and I've got to
mention quantum day. That's why we're here in this room.
And Video does not make quantum computers. Yes, we're having
quantum day. How do you approach it?
Speaker 13 (30:20):
So I think you know, big speculation of quantum computing.
Speaker 8 (30:23):
When is it coming?
Speaker 13 (30:24):
So I'll just tell you one very quick story. Arm
CPUs Right, arm CPUs are now a really big part
of the ecosystem. We use it, and Video has them,
Amazon has them, many people have them.
Speaker 8 (30:34):
But when it was.
Speaker 13 (30:35):
When the first kind of data center ARMCPU was launched
in about twenty eleven twenty thirteen. You know, I was
at Intel when we were very worried about it. But
at the time it lacked performance, so it took another
five years for performance to happen. But then it lacked
software ecosystem, it lacked programmability, it lacked libraries, and it
lacked you know, being able to use enterprise software out
(30:57):
of the box. So it took another five years. Huge
investment from Amazon. Actually, if we're honest for codes to
be rewritten to work. So we're now in the place
where Armor was in twenty eleven. So I think that
we need at least another five years for the hardware
to get to a place where it's highly reliable. But
(31:18):
then the programmability of it, how easy it can be popularized,
that's the difficulty. So in many ways there might be
true behind truth behind everyone sayings. Pad Guessinger thinks it's
going to take five years for the hardware, yes, but
I think Jensen thinks about it very practically. It takes
longer for the programma BOS.
Speaker 5 (31:36):
System, and Nvidia's argument would be their AI supercomputer separate
technology can help in the development. Flag Alobov of Ondia
really great to catch up here in San Jose.
Speaker 3 (31:46):
Turn back to UNSF no great stuff, big thank you
to add lod though out there in San Jose. Tune
in at four thirty pm Eastern today for a special
edition of Bloomberg Technology, hosted by our very own the
Live from Videos GtC.
Speaker 2 (32:01):
Now coming up, investor.
Speaker 3 (32:02):
Joe Lonsdale and Campus CEO today O Rende are going
to join us to discuss the startup Series B funding
round and changes to higher ed.
Speaker 2 (32:10):
This is Bloomberg, a tech startup.
Speaker 3 (32:25):
Campus has just raised forty three million dollars in a
Series B funding round. The company aims to give students
a more affordable path to a college degree. Bloomberg Beta,
the venture capital arm of Bloomberg LP, is one of
Campus's five largest shareholders. We should note joining us now
is campus CEO Toddy o ya Rende and one of
its investors, Joe Lonsdale. Toddy, I want to start with
(32:45):
you because you studied aerospace engineering in the UK and
at Embry Riddle in Florida. Was it your experience with
education that led you to start this company?
Speaker 14 (32:54):
Hey, Tim, thanks for having me. Hey Joe, definitely, I
mean before college, I was homeschooled until high school. My
paternal grandfather was a college dean. My paternal grandfather was
a high school principal. My mother is a college dean.
My older sister is a professor. So probably I was
brainwashed from birth to get really excited about education.
Speaker 3 (33:13):
Well, a school's reputation when it comes to academics is everything.
How do you build that reputation today from the ground up,
especially in the early years and when for profit schools
have had such a checkered past.
Speaker 14 (33:24):
Look, it's about elite education for all and so that's
what we're doing at Campus. We're sort of rethinking the
first two years of college. We're building a new kind
of two year college where students get to learned from
a the best professors from the top schools in the
country Princeton, Stanford, UCLA knock out the first years of
college with us, and there not just learning theoretical nonsense,
learning like really useful skills. And then they transferred to
(33:46):
the four year school of their dreams to complete their
bachelors with no debt. And I think that's the key.
Speaker 8 (33:51):
No debt.
Speaker 14 (33:52):
Student loan debt's about to pass two trillion dollars in
this country. We were hearing crazy stories students taking out
one hundred thousand dollars two hundred thousand dollar loans that
are graduating. They can't even get jobs. It makes no sense.
It's got to stop. But now there's actually a better way.
Speaker 3 (34:04):
Hey, well, speaking of that, I want to bring in
Joe to this conversation.
Speaker 2 (34:07):
Joe Lonsdale.
Speaker 3 (34:08):
Look, you've already invested in Campus, but this isn't your
first foray into education. You co founded the University of
Boston a few years ago. What in your view is
wrong with higher education? You went to Stanford, you seem
to be doing pretty well.
Speaker 4 (34:23):
Well.
Speaker 15 (34:23):
Of course, there's a lot of issues with the very
top of our education, which is what the University of
Austin's focus on. But you know, I'd argue that our
community college is unfortunately, are even more troubled in this country.
There are many have very low graduation rates. A lot
of them also are focused more on ideology than skills, sadly.
And so what today and campus represents to me, It
represents excellence, it represents merit. And you know, our economy
(34:44):
is changing drastically ais you guys are talking about on
other segments today. You know, it's changing everything how it's
going to work, and we need to get the right
skills and the right frameworks, you know, to millions of
young adults. And you know, I'm hoping Todd I could
scale this to a million students, kepture timers, sent the community,
call colledge market and really help all of those live
better lives and succeed more on the economy that it's coming.
Speaker 3 (35:05):
Let's go, well, Joe, what's your input on the curriculum,
because You're hiring a lot of folks, your portfolio companies
are hiring a lot of folks who have diverse backgrounds,
who have skills that are arguably not necessarily taught in
some schools and universities. What's the input that we're getting
him on the curriculum?
Speaker 15 (35:26):
You know, my push from my side is let's do
let's add some in more, some more courses in that
reflect what you need to know for AI.
Speaker 4 (35:33):
You know, there's people like Sam Altman involved.
Speaker 15 (35:35):
As well, who build open AI of course, and others
who are invested here. And the idea is, how can
we help hundreds of thousands, millions of young Americans, you know,
obtain the skills necessary to work in an economy where
AI is going to be involved in a lot more So,
that's not the immediate focus today. The immediate focus today
is on a lot of basic skills needed in today's economy.
But what's really fun is today's talking a lot and
(35:56):
thinking a lot about what else can we add in
here to really make sure we get people ready for
the twenty thirties.
Speaker 3 (36:01):
Well, Toddy, you were talking about the cost of college
getting out of control. You were sharing some pretty staggering
statistics about the trillions of dollars of student loan debt
that exists in this country. Nobody argues with that. How
do you make the economics of campus work though, when
other colleges and other even junior colleges community colleges.
Speaker 2 (36:21):
Are more expensive.
Speaker 14 (36:22):
Yeah, I think the key is, like the completion rates
actually have to go up for the economics to work.
So the traditional community college has an average completion rate
of about twenty seven percent graduation rate, and so when
you lose students when they drop out, you actually earn
less tuition revenue per student. So if you actually it's
sort of paradoxical. But if you actually keep students longer
(36:42):
because you help them graduate, then guess what, you earn
more tuition revenue, which makes the economics more healthy. And
so that's like the sort of the beautiful symmetry in
terms of what's best for the student, what's best for campus,
and what's best for our country. Driving up graduation rates
is actually how you make the economics work.
Speaker 3 (36:59):
Joe, come on in here, because I'm curious about your
view of the federal government's involvement in education. The government
is frozen, suspended, or cut more than a billion dollars
from universities In a recent week, we're talking about reports
from Columbia, Johns Hopkins ten and more. And I'm wondering,
as an entrepreneur, as somebody who's hired a lot of
folks who founded successful companies, somebody who has founded successful companies,
(37:23):
are you concerned about the American talent pipeline being cut
off as a result of these cuts?
Speaker 15 (37:29):
You know, I'm more concerned about making sure we spend
money effectively and efficiently, and so I really like what
Toddy is doing along those lines. A lot of the
policy I'm pushing, you know, coming from my side of things,
is how do we make this spend accountable. So, for example,
if you're going to do vocational education, Unfortunately, just like
our community colleges, a lot of the vocational programs, low
graduation rates, wrong skills, not helpful if you can spend
(37:49):
the money effectively, if you say these money is going
to be tied to results. For example, when you tie
the money to the salaries of students coming out.
Speaker 4 (37:56):
Of vocational schools, it doubles those results.
Speaker 15 (37:58):
Those are types of policies I think be popular on
both the left and the right, And what I love
about what today is doing is it's not really playing
the ideological games.
Speaker 4 (38:05):
There's people of all backgrounds.
Speaker 15 (38:06):
There's people involved from the left, from the right, black, white, Hispanic, everything,
and it's just about merit and excellence and getting good
results for very small spend. So I think this is
this sort of thing that is going to remain popular
with everyone, regardless of some of the other fights going on.
Speaker 3 (38:20):
Well, today, how do you watch what's happening in the
federal government because Predesident Trump today is expected to sign
an executive action that formally asked officials to take steps
to dismantle the Department of Education. According to our reporting,
what happens to your business then, because forty percent of
your students qualify for pelgrants and those are administered by
the Department of Education.
Speaker 14 (38:39):
Yeah, Look, the vast and jort of our students use
PEL grants to cover their twition and so they don't
have to pay anything out of pocket for twition expenses.
Paul Grant is not going away. Even if the Department
of Education is dismantled, some of these key programs that
are mandated by Congress are going to be split across.
You know, maybe Treasury or the IRS or other organizations.
The way I look at what's happening in Washington right
now is, hey, look, obviously everyone's looking at this and saying,
(39:02):
we need to be more accountable. As show talked about,
we definitely need to be more efficient with taxpayer dollars.
It's really early days. Secaturmic Man's been in there for
less than three weeks. I think you know we're watching
it closely, but we're going to have to let this
one play out.
Speaker 3 (39:13):
Hey, Joe, last one to you, speaking of efficiencies in
the federal government, You've been supportive of Dog this week,
though a federal judge ruled that Elon Musk's actions to
shut down USAID violated likely violated the Constitution in multiple ways.
Are you concerned that the courts are going to prevent
Elon from being able to do the cuts that you
want to see him?
Speaker 15 (39:32):
Do you know that particular ruling. I'm glad you mentioned
it because it was so ridiculous. So actually the ruling
was so misguided that it thought Congress had created USA,
which is not correct. It was actually created by executive
action the USA. It is just such a great example
of just complete waste, right they're just all sorts of
scams and fraud that we've uncovered. I think no matter
what your party background, if you look at the details,
(39:53):
you'll agree this should have been turned off.
Speaker 4 (39:54):
And there are.
Speaker 15 (39:55):
Activist judges they are going to try to slow it down.
I personally hope the Supreme Court is going to step
in and make some sound rulings here and stop activist
judges from violating the Constitution by their interference. And it
is going to be a big issue.
Speaker 3 (40:07):
So what have you spoken to Elon about this? Have
you spoken to Elon since he's been a doge.
Speaker 15 (40:13):
He's a friend and I am in touch, and he's
working really hard with a lot of smart people. They're
being very aggressive. A lot of my friends are involved
in DOGE and listen, there's there's I think. I don't
think everything they're doing is going to always be perfect,
but there's so many crazy things that have to be
turned off and have to be confronted that overall, I'm
very very happy for the work they're doing, and I
think they're kind of shocked about some of the ridiculous
(40:34):
things they're finding as well as they're publishing all right.
Speaker 3 (40:37):
Well, really appreciate both of you guys joining us. That's
Joe Lonsdale from a VC also campus CEO today. Oh
Yareen Day, thanks so much for joining us. If you
wanted to grind the world to a complete hall, you
could achieve.
Speaker 2 (40:50):
That by removing magnets.
Speaker 3 (40:52):
They're crucial to basically all tech, including EVS and the
next nuclear breakthrough fusion Energy Primer or the latest Bloomberg
original series takes a deep dive into all of this.
Speaker 16 (41:03):
It takes a lot of work to build something big.
You also gets very expensive, like the amount of money
you need to spend on something to get just the
first one can get very very expensive.
Speaker 17 (41:14):
That's exactly what's happening with ETA, a giant fusion reactor
currently under construction in France that uses super conducting magnets.
Look at this thing. It's huge and as a result,
it's projected to cost as much as sixty five billion dollars.
So to make fusion smaller, cheaper and more practical, Commonwealth's
founders needed a whole new kind of magnet.
Speaker 16 (41:38):
The question was like would that material ever happen? And
it wasn't until the early two thousands we could really
see that that material was going to happen that there
would be a new type of superconductor. And it's a
weird it's not a wire, it's a weird thing. It's
a film and it won the Nobel Prize like months
after it was discovered.
Speaker 18 (41:59):
So this is a material called HTS or high temperature superconductor.
Is it is Literally it comes down a tape. It's
probably kind of hard to see on the camera. It's
very thin. It's actually mostly copper and steel, but there's
a very very very thin layer inside of this that
is high temperature superconducting material.
Speaker 17 (42:17):
High temperature in this case is still wildly cold, but
not quite out of space cold. And that was the
breakthrough that Commonwealth needed. Magnets made with this material can
create a stronger magnetic field, so they don't have to
be so massive.
Speaker 18 (42:34):
You can shrink the size of the device by a
factor of ten. Basically allows us to make things smaller,
which makes things cheaper and makes things faster to get
fusion to a spot where we can make energy from it.
Speaker 3 (42:50):
And that was the voice of Caroline High. Tune into
the first episode of Primer tonight on a Bloomberg TV
at six pm Eastern time. That is going to do
it for Bloomberg Technology. Tune in later today at four
thirty pm Eastern one thirty pm Pacific for a special
edition of Bloomberg Technology Live from Nvidia's GtC event.
Speaker 2 (43:07):
Also check out our podcast.
Speaker 3 (43:09):
You can do that on the terminal, as well as
online at Apple, Spotify, and iHeart This is Bloomberg.
Popular Podcasts
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.