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.
Like from New York, I'm Caroline Hyde and I'm Jackie Devalas in San Francisco. This is Bloomberg Technology coming up.
Government and tech leaders descend on Paris for the AI Summit, as French President Macron touts the country's very own stargates. Plus President Trump says Elon Musk's doge may have found quote irregularities at the Treasury Department, and China's BYD is putting driver assistance features into most of its cause for free and plans to integrate software from Deep Seat.
But first as.
Checking on the markets, and actually the Nasdaq pushes higher, but Tesla is one of the points.
Contributed to the downside.
We'll dig into what buyd means for Tesla competition in China. We're up one point three percent, and video lending its power to the points that move to the higher side. We'll dig into the AI track that drives in Vidia higher. But move on to a keystock. You've got to watch because as a record high and is up sixteen percent for sixteen straight days.
If we see Meta managed.
To hold onto these rather small gains on today, Jackie, it would be sixteen straight days of gains and a lot of it is fueled by AI exuberance. For Meta Sweet sixteen, we move on to what's happening over in Paris, where it's all about AI. Of course, the summit is now underway, bringing together almost one hundred countries and over one thousand private sector and civil society representatives, including Prime
Minister of India. We've also got some outment of open AI and this as France is said to announce a total of one hundred and thirteen billion dollars in investments in AI projects in the country. Bloomleg's Tom McKenzie is standing by and rather a kin to stargate here in the United States. It's about private sector commitments over in France, Tom it is.
And if the fire had not been lit under the feet of policymakers in Europe prior to Deep Seek, prior to Stargate, then it certainly has been in recent weeks. And this is Macwan's response. The President expects to take the stage behind me shortly. Currently you've got Cinder Butcher, the CEO of course of Alphabet speaking. The commitment from Macaran, as you say, around one hundred and nine billion euros one hundred and thirteen billion US dollars, is in part
an answer to stargate. It is also in part an answer to the innovation now coming out of China, and it's an attempt by France and Europe to position itself to close the gap. Macwen's been quite clear about this. This is catch up now for Europe. And here's line yesterday in an interview of a French newspaper saying we need to invest, invest, invest, along without one hundred and thirteen billion US dollars. By the way, you also have general catalyst plans for about one hundred and fifty billion
used audiers. They say they could put to play in France, in Europe alongside the likes of Mistral and Spotify and other investors to drive and build out infrastructure to layer as well on top of the industrial sector of Europe. That's where they think the opportunity is. But of course, key questions around implementation, key questions about regulation, key questions about whether or not we have the data the chips and the talents to really compete versus the US and China.
Tom, this is a huge ambitious effort on behalf of the Europeans. But I'm curious because they've usually also been those to be at the forefront of regulation, something that private companies and some investors don't really love. Have you gotten any sense from the executives you've spoken to about how that balance is playing out right now?
There that is very much a live debate to what extent Europe should regulate when it comes to AI. The EUAI Act, it is going to start to take hold this year. It will be largely fully implemented by twenty twenty six, so that is already the door is already open. The horse has already bolted on that front. To what extent they can tweet the regulation going forward is a question mark. But clearly the divide now between Europe's approach and the US approach for President Trump ripping up AI
regulation is stark. Some would say that having the guard rails in place in Europe is at least beneficial. At least there is the guard rails in place, at least there is a track to followed. You do, of course, have a disparate landscape in the US state by state that potentially could point pose problems for US AI companies trying to navigate the state by state regulation. The other question, of course, is to what extent Europe can leverage its talent that it's got here, and it does have software
engineering talent the energy as well. France is a big producer of nuclear energy. That's a benefit. We do not have semiconductors, we do not have companies like Huawei in China on video in the US, so that remains a challenge. But they're hoping that at the end of this summer they come through with a proposal, they can meet some of the challenges and really get to grips on the implementation of this.
That's Bloomberg's Tom mackenzie.
Thank you.
Here's something that came out of one of Tom's interviews at the Paris I Summit earlier today. The head of Google's Deep Mind, Demasisabus, questioning the notion that China's Deep Seek actually spent less than six million dollars to develop its aisystem.
The first thing to say is it's a very impressive model, or very impressive piece of work. I think the team is probably the best team that I've seen come out of China. So that said, I think a lot of the claims are exaggerated and a little bit misleading.
Let's bring in RBC Capital Markets Internet analyst Brad Ericson for more on the AI landscape, on the race, so called race between US and China, and Brad, when you hear of deep Sea, you know, it seems as though people are questioning the overall amount. Therefore, do you like Amazon committing to one hundred billion dollars to Microsoft more than eighty and similar coming out of Meta for example.
Yeah, I think for the moment, they have the revenue growth to support those investments. I think people are right to question the long term return on invested capital equation there, which is really where the debate has shifted. But I think for the moment, again, we're in this moment where we don't know how big this technology can become, and so I think we lean on those statements such as the risk of underinvestment is far greater than the risk of overinvestment.
I do think that's kind of the prudent way for these companies to be treating this for the time being.
And look, investors have thus far given them a pass. Look at Meta shares extraordinary sixteen day run to the higher side. Many would say that the open source commitment from Meta has been speaking to their large language model operation, the commitment to leaning into AI, even as of course they're announcing who's going to be hit by the job cuts today. Just talk to us about some of the thesis around Meta outperforming the other big tech players here.
Yeah, sixteen game hitting streak is pretty impressive.
I would say Meta has kind of the two key things going for it right now. One, they have arguably the greatest amount of revenue coverage.
It's what we call it.
So again, just when you look at the marginal revenues they're adding, they just did twenty seven billion dollars a year over year in twenty twenty four. That's plenty of extra revenue dollars to cover the depreciation that's coming into the model from the capex spen. So that's the first part. The second part is they just have a heck of a great narrative on all the ways that AI could
affect their products, both now and in the future. And so I think the market is what the market's telling you right now is they really believe in Meta's strategy and really AI giving it the most torque.
To the upside in the future for.
This model, Brad, I want to talk about open source technology now that we're on the topic of Meta, because in your note you actually highlight startups desire to want to build more N specific applications on top of already existing AI and I'm curious how that shift is better for investors. Are open source companies better position than closed source companies like open AI for example.
You know, I think there's a debate as to whether the closed source guys actually do decide to open things up because it could to the improvement of their models or the improvement of their distribution over time.
Clearly, you know Meta's open source strategy. I don't want to call it a wolf in sheep's clothing.
But there is an aspect where they're looking to leverage sort of developer intel of the world to be able to then potentially pull in those innovations in the model and then use them for internal purposes which would never see the light of day, and that benefits their products longer term. So I do think, you know that's that's definitely a part of their strategy and frankly, that's really
smart from our perspective. So that's that's I think, you know, at this point, I think that's how they are leveraging open source so well. And again, like I said earlier, the market is really buying into that thesis right now.
One of the other things that we're looking closely at today is of course the Paris AI Summit, and we've had a few of these summits over the last couple of years, ever since child GBT hit the scene. But I'm curious as an investor if anything that comes out of that really affects your long term thesis, because we are talking about big sums of money being poured into this space now from the Europeans.
Yeah.
Absolutely, It's kind of funny how this is following the history of tech right US innovates, China replicates, and Europe regulates. Is kind of the famous saying there. Yeah, I mean, certainly from a sovereign AI perspective, I think countries recognized there's a huge opportunity to modernize and make better use of the treasure tropes of data they all have. On the flip side, of course, there's a civil society impact here.
What does it do to jobs? Does it make people more productive or does it just straight up take.
Jobs from and and remove or dislocate industries in an adverse way for you know, macroeconomies.
Those are giant.
Questions right that are that are certainly being discussed there. But I think from this point, you know, our perspective, our seat, where we're covering.
These these Internet names.
They're making all the investments to push you know, the industry and frankly the world forward this with this innovation.
We have to be behind that at this.
Point, and you get the infiltration across not only the large language model makers but those that adopt general to AI. Just think of pinterest numbers last week and how generative AI is fed into that story. We get the news today from other sources that LIFT is going to be eyeing robotaxes as soon as twenty twenty six. Where are the level of Internet names that we haven't yet factored in, where GENAI is going to really be reaching forward in terms of their earnings and revenues.
Yeah, I think you know, certainly autonomous vehicles you hear about particularly around synthetic data, so that's being able to take for instance, real world miles and then model them out into other situations that you couldn't possibly capture in testing, people driving around and accidents that could come up, just as an example. So that obviously benefits we think the riot hailing names, and there's going to be others.
You mentioned.
Digital ads is kind of one of the best sectors where you can apply AI. Web design is another sector we've mentioned, So that's companies like Wix and GoDaddy.
Those companies are making it easier than ever to.
Go out and start a small business, build a website, and grow off of that.
So those are some of the ones in our coverage.
We like, Brad, we had some big tariff news coming out of Washington today. How does that weigh into how you're looking at tech names going forward, especially given all this talk about AI infrastructure, which will require perhaps some of these critical materials.
Yeah, I mean, I think we'll reserve judgment on what the ultimate outcomes of those are given what we've seen over the past a few weeks, but I think the the impact would align with what the macro impact would be, which is if you raise costs, certainly that's probably not going to be a good thing for these companies. That are building out, certainly Meta, Amazon and Google being the three kind of you know, biggest brick layers in this industry.
I'll be see Capital Markets Internet analyst Brad Erickson.
Great to get some time with you. Thanks coming up.
Those may have found irregularities and the Treasury Department that's acording to President Trump.
More on that.
Next is blue meg technology.
There could be a problem.
You've been reading about that with treasuries, and that could be an interesting problem because it could be that a lot of those things don't count with other words, but some of that stuff that we're finding is very fraudulent. Therefore, maybe we have less debt than we thought of.
Think of that.
That was President Trump speaking an Air Force One ahead of yesterday's Super Bowl attendant saying Elon Musk's doze may have found irregularities at the Treasury Department. For more, Bloomberg's Mike Shepherd joins us. Now, Mike, what do we know about what exactly Trump was referring to and what the potential implications could be on the treasury market.
Well, this is one of the great mysteries that emerged from that Air Force One conversation with reporters. He raised the prospect of irregularity somewhere in the Treasury Department, but offered no specifics, nor have we heard from the cost cutting team over at Treasury now that's being led by DOZE, by executives from this US digital service that DOZE.
Is working with.
We don't know what they're referring to as a government contracts. Is it excessive spending within the Treasury Department on certain cons and we don't know even whether it refers to the Treasury's market itself the way the President was discussing in that moment. We do know that Scott Bessant, the Treasury Secretary, sought during his interview with our colleague Salaiah mosen to are for reassurances that everything would be normal operationally at Treasury despite whatever Doge might be doing.
And how much Doge can do when well judge his way in and try and halt some of the activity is a question mark, Mike, so too, is well laying back some of the operations.
Of other units.
I'm thinking about the Consumer Financial Protection Bureau seems to me now skeletal too.
Well, that's right, Karen, that is one of the latest targets of this cost cutting initiative sweeping through the government. DOGE is going really agency by agency, and the Consumer Financial Protection Bureau was the latest target over the weekend. Essentially, they have shuddered operation. Employees have been told not to come into the office.
And of course this does.
Coincide with Elon Musk's own efforts to turn x into a one platform for all different types of services, not just social media and communication, but also turn into a place where you could do transactions financially.
Now, the Doze.
Team over at CFPP has gotten access to some sensitive data our reporting has shown, including bank examination records and other information that is closely guarded that's very sensitive. Of course, the Consumer Financial Protection Fewer does scrutinize banks and other financial services companies for the way that they interact and intersect with consumers. And this agency exists, as its name
would suggest, to protect consumers. You know, there are other financial regulators, but this is the one that is really geared to look out for the average American and it is returned since its inception in twenty eleven, roughly twenty one billion dollars in improper payments or improper charges back to consumers from these institutions.
Megs, Mike Shepard, thanks for all of that.
Meanwhile, Doge gaining a key ally we understand at the US Treasury with veteran tech employee Tom Krause, who has been appointed a special Government employee and will serve as Fiscal Assistant Secretary at the Treasury. But he's also the CEO of Cloud Software Group, which saw deep cost cuts at Citric Systems.
For Marburne, Megs, Jordan and Robertson.
And it is such a deep dive that you do with colleagues here, Jordan, just talk us through who Tom Krause is and why we're worried about the deep cuts he's done in the private sector.
Yeah, thanks for having me.
So we're learning more information about Tom Krause, you know, the Doge pick Elon Musk's Doge pick at Treasury. As you mentioned, he currently holds two jobs, one at Treasury and one as CEO of Cloud Software Group. Cloud Software Group was formerly known as Citric Systems. That company got acquired for sixteen point five billion dollars in a private
equity leveraged buyout in twenty twenty two. And what we've done is we've taken a look at a series of pretty significant cost cuts that Kraus enacted at Cloud Software Group over the last two two and a half years, and in particular the impact of those cuts on the security of Citrix products. There's a family of Citrix products called NetScaler. These are security appliances that have been at the center of a number of major hacks in recent years.
And what we looked at was to try to quantify the level and the extent of the cuts to the security department at Citrix and any impact to the security of those appliances. And what we found was that the cuts were very deep, indeed across the company, including the security department, and that this in fact may be why Kraus was selected by Musk's doge.
As for this Treasury role.
His background is cost cutting, His background is all so according to the sources we've spoken with, running incredibly profitable companies, you know, this is his specialty.
Jordan, your reporting raises an important question about employing private equity tactics, this cost cutting that you're talking about in sensitive sectors. So now kind of take that into the government in Washington, where cross now has more influence. What has the agency said the Treasury Department as well as maybe even the company he's still employed by, which I found super surprising.
Yeah, the Treasury Department says things are you know, operating normally while the DOGE team kind of conducts its work, you know, over the weekend, Krouse, you know, and other DOGE staffers were you know temporarily stopped from accessing data and using the data.
You know.
However, you know, what.
The reporting finds is that, you know, this is a strategy, a very time honored strategy among private equity companies to you know, kind of load up companies with debt, buy them with lots of debt. You have to then cut lots of costs in order to repay that debt. And you know, in terms of the potential impact in the federal government, it really depends who you talk to. You know, we've spoken to folks and we've included their commentary in
our story. You know that think that ultimately cross could be really a positive thing for the government in terms of you know, really aggressively slashing expenses and finding ways
to be more efficient. We've also spoken to other you know, industry experts, security experts, government experts who say that applying a private equity approach to government, you know, could potentially have very very serious and very unexpected consequences, especially when you're talking about cutting not just thousands of employees that were cut at Citrix, but potentially tens of thousands, hundreds of thousands or more.
That's Bloomberg, Jordan Robinson, thanks for joining us Tesla Challenger. BYD plans to integrate software from deep Seek and is also offering you driver assistance features in almost all future models at no additional cost to buyers. Let's bring in Bloomberg's Craig.
Triudell for more.
Craig, before we get into some of these new features, what do we know about how by D is incorporating this new Deep Seek technology.
Yeah, it's still a little bit unclear how exactly this is going to work, but there is some support here. And I think the combination of two names that you know, it's hard to kind of think of of two that
are any bigger in the tech and auto spaces. I mean, a deep Seek very much as sort of come lately name, whereas by D I feel like has been you know, the last few years, just the sort of talk of the auto industry in terms of how quickly they've grown and and you know just how big of ambitions they have, and you know, the one knock that they've they've maybe had against them is they don't necessarily bring a ton
of you know, driver assistance technology to the table. They're trying to address that and also do so in a way that you know, is very sort of consistent with what their approach has been just to the industry in general, which is you know, really aggressive pricing.
Really aggressive rampop in it shares. Recently, we're just showing a chart. It's up more than twenty percent over the last four training days.
Craig.
So, what is it about this cheaper autonomous option.
That is just leaving competitors for dust? It must be a big pash for Tesla.
Yeah, I mean, I think that's a big part of it.
When when you sort of measure BYD up against you know, a Tesla or some of the you know EV startups in China that you know have really kind of leaned into driver assistance technology, you know, BYD hasn't really had you know, as much to offer there, and so there's been a sort of you know a knock against the the you know BYD offerings that you know, well, if you're going to have to kind of settle right, you're you're not going to be able to have the sort
of latest and greatest in lane keeping or what have you. And so I think there was a real sort of opportunity for BYD to kind of address that in a more aggressive way. And you know, you mentioned that they
run up in the stock. All it took was a sort of you know, invite of an event, and it was sort of reminiscent of how you know, Tesla over the years has been able to get investors really excited just by you know, setting a data on the calendar for a big product event that they you know, they have something to share.
Bloomberg's Cratydell, thanks so much, Welcome back to Bloomberg Technology and Caroline Hide in New York.
And I'm Jackie Devalas in San Francisco.
Quick check on these markets, Jackie, because look, we managed to shake off the latest geopolitical acts, those towers coming on aluminium and steel coming from President Trump, not managing to contail some of the money on that we're seeing in then as that one hundred, we're up one and a quarter percent. What leads us on a points perspective, Synvidia, we managed to claw back some of that sell off from Deep Seat. We're about halfway there. It gains another
two point eight percent on the day. The biggest points contributed to the upside maybe feed across from what's happening in TSMC and Honheim TSMC, even despite that earthquake, not to production, standing by their forecast for the first quarter and the full year and committing to AI investment. That of course helps the signals for nvideo. So TWU does hon Hi at server Maker also showing resiliency there. So all about AI today and more on how indeed companies
and governments are approaching AI safely. Let's go back to Paris imst Tom McKenzie standing by the CEO of German startup at left Alpha, Caroline.
Yes, thank you, very pleased to say. I'm joined by the CEO and the founder of Alaf Alpha, one of the leading AI startups of course in Europe coming out of Germany were a focus on sovereignty and on enterprise. Jonas start with Deep Seek because I've heard from some in this room that it's a spartanek moment, echoing what we heard from of course Andres and Horowitz anothers saying it's a flash in the pan. Where do you stand on DeepC.
Well, first of all, it's basically a continuation of what we've been seen for years. Models get better, they're get more efficient open sources, Steve. So it's basically a confirmation of what we've been positioning ourselves for four years.
So for us, it's a victory lab. For our customers, it's a victory lab.
More incremental than spotning moment.
Well, yes, but I mean, of course it's a spotning moment also because China is maybe in some regions ahead of the US for the very first time. So that's certainly that you're.
Offering it to your clients. Now have you tested it? Are you data secure on this? What does the response be?
Yeah, I mean our strategy is to include open source model, right, So we've been doing this with LAMA, We've been doing this with our gate models, and we've been doing this with DEFCIG and so our customers, our large enterprises and governments. So any transformation you're doing for these customers takes years and years. So what is today the very best model
is almost irrelevant. They just need a partner that can guarantee that they can leverage the best of open source together with innovation they need for their own business.
You've told me in the past you think large language models will end up being commoditized and essentially price will drop to zero. Okay, And yet on the stage behind us, Macron, the president of France, is going to get up and announce one hundred and nine billion euros of investment to kind of match stargate in the US. Is that funding misspent both in the US and Europe? Is this misaligned?
So we need funding, We need funding for our companies, We need funding for infrastructure. But what I worry is that we are getting distracted and putting too much focus on infrastructure. Infrastructure is not going to save US. Infrastructure investment alone will not solve digital sovereignty.
What we need is value creation.
We need innovation, and then the infrastructure comes almost automatically, as like a second order factor.
You pivoted from building out large language models to building out generalized AI operating systems, very much focus on the enterprise. I know you're going to push back on the word pivot here. What does the response been, is you reflect back on that, was that the right decision?
How are your investors.
Feeling about that?
So deep seek For our investors, they were having a party because there's this precisely what we've been positioning ourselves for. We can solve the general purpose task like writing a poem for your mother's birthday, or like email replies. Our customers can solve that with the best of the best
of open source models. But one thing all the customers have in common is that the things that are really valuable they care about, like compliance workflows, government workflows, manufacturing requirements, engineering, all these open source models can't do it. So we are adding our own innovation to solve that and everybody's winning.
We'll start doing for revenues at alf alpha, what do they come in at it twenty twenty four? What's your target for this year?
So I'm not going to tell you my target, but I'm going to tell you that we just recently signed a nineteen million contract with the German Labor Agency Work Work Agency, Employment Agency. So these are the kind of contracts we're currently signing, and I think that is numbers worth getting out of the bed.
For your German starts up. Elon Musk is meddling. So some would say in German politics. What lessons do you take from that Elon Musk supporting the AfD in Germany, a right wing party.
I mean, to be fair, we've been meddling in US elections as well, right, so it's like a well, I mean, some German politicians have been very vocal about their preference.
Nothing for this. Come on, I mean, what does it? Does It underscore your design to have sovereign AI. You have a tech mogul interfering in German politics.
So I don't like it, right, but I think there's nothing really we can do about it right now. The US are still our partners, are still our friends, and we need to find a solution together with the democratic countries, together with our partners open AI.
We're talking to me earlier and they said, the two leaders in AI are the US and the CCP, the Communist Party of China. Do you agree? Does you have skin in the game here?
Really?
So that's probably correct, But this is not a Serra sum game. This is not about who is the first and everyone else is empty handled?
Right?
We Davost just two weeks ago we.
Launched for the We have a first team that launched an evolution of the transformer architecture, and I talked to several research labs, leading research labs US researchers that are now starting to look at that. So it's with all likelihood the next architecture of the next evolution of the transformer is coming out of Europe.
Do we need to be worried that we can't access and we can't build out or in semiconductors. Not only do you have a Nvidia and AMD in the US, but you also have Amazon Microsoft developing their own semiconductors around artificial intelligence. Is that a gap that we need to close and how do we do it?
Well, that's a painful gap, certainly, right. Also the fact when you look at the capex the hyperscalers have assembled, that's almost so possible to kind of reach the same levels. And it's equally unlikely that we'll build a competitor to AMD or Nvidia basically with a snip off a finger. So that's painful, but it's a reality we have to accept. And I think every money that is directed against like building a competitor to these jagernauts is wasted money.
Okay, honest, thank you very much, And DIG always a pleasure to speak to you and so and Joyalist of course, the CEO and co founder of Alah Alpha, speaking to me here at the Gonepalais just ahead of Macrant's speech. The president of course of France guys.
Back to you, Tom McKenzie, Thanks so much. It's time now for talking tech. First up, Deep Seek is ramping up hiring recruiting specialists for artificial general intelligence roles, data experts, researchers, and a legal chief. All this really signaling the Chinese AI startup to actually expand ambissions after its stunned markets for the cheaper model that was on par with Open Aiyes, plus,
Apple gets a mandate from the UK. The government has ordered Apple to build a backdoor to access global user data. It's a move that could spart one of the biggest privacy battles for the iPhone maker. In an undisclosed order in January, British authorities asked Apple to circumvent encryption news to secure user data in its cloud service, and Rivian and anounced that it will begin selling its EV trucks
that we're designed for Amazon to other customers. This is an effort to boost revenue as consumer demand for EV cars actually slows the commercial vehicles can be purchased by fleets for any size, but deliveries slated to begin in the second quarter. Meanwhile, we're watching shares of LIFT. Interesting reports coming where up some three percent could robotaxis be offered by Lift as soon as twenty twenty six. That's a report coming from TechCrunch and from an executive inside Lift.
Jackie coming up.
Joseph Makett from the Center for Strategic and International Studies joins us to talk about how energy demand could impact AI growth. This is Bloomberg.
Nock Here will replace its CEO with Intel's recently appointed head of AI, Justin Hotad. The surprise announcement comes as nock Here it's the network equipment maker watch the exit a prolonged slump in orders from phone operating Now the incoming CEO joined Bloomberg Television earlier today.
Just take a listen.
My focus coming in though, is going to be to spend time with all of our major stakeholders, and one of those will be our customers and particularly our telco customer base, to understand how we're serving them well and what we can do to continue to serve them better and drive growth the way I think about it is we have a strong and broad portfolio today, some of which is being touched by AI today. But I have also The way I think about the AI cycle is
it will be a supercycle. And much like we saw with the Internet, it spawned the mobile Internet, it spawned cloud, I believe we'll see that same thing with AI.
As the AI Summit is underway in Paris, let's also consider the future of AI safety regulations here in the US under Trump. This just as a top official at the USAI Safety Institute steps down, Let's bring in Bloomberg Sharing Gafari for more. Sharen talk us through what this agency was doing and how this departure potentially puts that progress at risk.
Yeah.
So the USAI Safety Institute is relatively new, right, a little over a year old, and its mandate was to sort of try to convene some standards around what AI companies, you know, are going to be, how they're going to be testing their AI models to make sure that they're safe, and also to test their capabilities. Now that Elizabeth Kelly, the head of that institute, is stepped down and there's no replacement announced, it's a little bit unclear what the future of this institute will look.
Like just at the time that world leaders, indeed leaders of companies gathered to talk about the future of AI, of any sort of guardrail that Europe can agree with the United States, It's a pretty pivotal movement. It feels like Jade Vance is going to be speaking on it.
That's right.
So we have, you know, the US Vice President planning to have a presence at this summit. But again, you sort of in order to kind of get all these companies together to give access to their models, to test them, to see how safe each release is, you sort of need I mean, I guess the argument for the USA Safety Institute is that they said that you need some entity to sort of actually be in the weeds doing
this work. And so who will be the leader of this institute, what their priorities will be, what kind of safety evaluations they will run, is critical beyond the kind of executive level that we see Jbvance taking leadership on.
Well said Sharin Vifari's great reporting as ever, thank you. Meanwhile, as governments and companies around the world, as we say, announced plans to build or expan AI infrastructure, another question is.
Being raised, how are they going to power it all.
Joseph miket Is with US is the director of Energy Security and Climate Change over at the Central Strategic and International Studies.
And look, it's so interesting in this moment that France.
Is going to be committing, we understand, to a gigawat of nuclear energy for this AI exuberant moment. You've got leaders of Microsoftware indeed Andy Jasse just last week saying it's all about energy that stops us building out data centers. Joseph, what are we not getting about energy needs here?
Well, you know, a data center is nothing more than a bunch of chips and racks that will take models, data, use algorithms to cook together intelligence, answer questions, help people perform economically productive activities. Energy is the flame that powers
that process. Electricity in particular, and when we look at what the leading hyperscalers and companies want to build, these are immediately going to be some of the largest electricity loads in the United States and probably the largest class of new generation capacity adding over the next five to ten years.
So there is a real question about just how much the electricity grid can actually withstand.
What do we know about.
What buy in we need at the local level, because a lot of these plans are coming from the federal level and it's not clear to me that do they actually need more permission or perhaps collaboration with municipalities states.
So it's important think about the size of the challenge.
We need to even to hit sort of medium level forecast for how much data center energy we're going to need over the next five years, we need to build more generation capacity over that next five years, then we
added over the last twenty. It's just a muscle memory that in the United States we don't really have and that involves everyone from the state and local permitting committees, from local utilities, and there is a role for the federal government in providing some backstops, some funding, perhaps some permitting reform help so that we can get the speed or we can get power to these data centers with
enough speed. The reality is we don't just have a bunch of gigawatt power plants lying around that are ready to plug data centers into.
So give modifort tends to the current administration. Apologies stacking entirely my fault give marks out of ten for the current administration.
I think we need to see some more texture around what the current administration wants to do. President Trump has said we are going to open up all the breaks in terms of permitting. We're going to look even to coal plants to rebuild these kinds of things. I think what's really important is what's going to match what the hyperscalers and the companies want to do, and what matches what's going on in the power sector.
Now.
When we look at what we expect to see over the next few years, the real additions are going to be solar, maybe some wind batteries, natural gas. I think the administration is in a position to help some of that proceed. Then we need to look at long term growth because this isn't the problem that ends in twenty thirty when the first few mega sized data centers come Aliine, if this is truly the economic revolution that the companies are investing toward.
We need to have a much more ready.
System for further growth in the twenty thirties and there I think we want to see the administration move toward nuclear power. We need to see investments in modernizing our grid. Those are going to be the key things for long term progress in the United States.
Joe, So, where does this leave climate goals, because when I hear coal power plans, it seems like it's actually taking a step back. And as you mentioned, hyperscalers are involved, what are they going to be amenable to, especially because they have net zero goals themselves. You look at climate security, this is not just a story about the environment, but it's one that has to do with national security. Talk to us about where it leaves as goals.
Of course, there's a lot of considerations that we need to think about, especially as the government gets involved trying to support these megaprojects. I suspect that we're not going to see new coal plants build the United States to power data centers. Instead, where we want to see those
hyperscalers want to see investments in clean energy. One of the things that is most helpful in that regard is that they're pretty deep pocketed and they need to spend to build power quickly, but they also want to spend and can make investments in the novel technologies that will allow us to grow this stuff over time. That could
be small modular reactors. We've already seen them restarting large nuclear actors here in the United States, and so the reality of the system today is the more we build, the easier we make it to.
Build, the cleaner and cleaner.
I think this process is going to be for us.
What's the breakdown of what that share can look like? Because we've heard about nuclear energy kind of for a long time, and I would like to kind of get a better sense of what the horizon actually looks like. Is it a five year or ten year kind of initiative?
We move really really fast.
I think we can make both grade upgrades and nuclear power a part of the solution set early in the twenty thirties, but a lot of that stuff is still a little bit far from commercialization.
It'll take us time to get there. Over the next few years.
I don't expect to see much difference in the capacity that's added from what's been added historically. The US builds a lot of solar, the US has large natural gas resources, and the investments we are seeing over the last year and the things we expect to see over the next couple of years are going to combine those two things.
Combine those two as principally with some battery storage and some investments in green energy and other places to ban to manage both the emergent power need, the need for stability. You want to run these facilities basically all the time once you've made such a large investment in the underlying chips and data, and then dealing with the climate effects by making investments in renewables and other things.
That's Joseph Market from the Center for Strategic and International Studies. Thanks for joining us.
Just take a look at shares of t Mobile right now. Stock as you'll see, is up today. The telecom provider has officially launched its service with the law Musk's Starlin Network, softwering customers texting and data usage completely off the grid at the low promotional price of zero dollars.
It's free for now.
The offer is open to anyone for a limited time, with the service eventually costing fifteen dollars a month. Tmobile announced the program on Sunday with an ad debut during the Super Bowl Jackie.
For more on the Super Bowl and how the NFL is dominating entertainment, Bloomberg's f Lucas Shaw joins us. Now Lucas viewership of almost everything on TV has plummeted, but somehow the NFL's has just gotten bigger. Help us understand this resilience because it's not every day that we're all tuning into a nail bier like last night.
Oh yeah, that was not the game that the NFL wanted.
You know.
Part of it is just live in sports, right.
You look at the television business and as the ratings for scripted shows has gone down, you know, sports has stayed pretty resilient, whether that's football, baseball, basketball. Football has been much stronger and gotten much bigger than basketball and baseball. I think part of it is that there are so few games, this idea of scarcity, right, so every game in the football season matters in the way that whatever NBA games are happening this week don't really.
And then the I don't know, the product is just really good.
You know, there was a moment six eight years ago where the NFL looked like it was a little shaky, The ratings were going down. There were a lot of different reasons for that, and it has fully rebounded.
So interesting in those times that maybe we were questioning ESPN had at the time, John Skipper, you write in your newsletter, was actually trying to persuade Bobyger to get out of the NFL, go bigger, on other sports.
But good thing Bobbiger ignored.
Him, and a lot of it is about content and about knowing the people. It's almost like you're following some sort of real life, real life celebrity culture here.
Yeah, I mean, I thought that was a really interesting story that John Skipper told me. He was the head of ESPN for a long time, really brilliant executive, and he had the same feeling I think that I did several years ago, which was he saw all the bad news around concussions and thought, Okay, this sport is unsafe. It's going to be on the decline. Turns out most Americans do not care. And to your point about celebrity, they're way more interested in Patrick Mahomes or in the
Taylor Swift of it all. I mean, what a couple of the emails I got from readers last night were that I did not give Taylor Swift enough credit for the current popularity of the NFL.
And sadly she probably wasn't that exuberant after last night. More love being shown for the halftime performances as well. But they make some big bets on when to play shows right, like Christmas Day for example.
Yeah, I mean, look, the NFL has gradually expanded its control over the schedule.
It used to be that the.
NFL was primarily a Sunday product, and then there was Monday Night Football. They added Thursday Night Football, which is now on Amazon. I think they're very happy with that because Amazon helps them reach more viewers get into streaming, and they can also prime Amazon to maybe bid on some international rights that they'll sell down the line. They
started to have more games on Saturday. I think there's an expectation that eventually, because they're playing games and other markets, that they might create sort of a new window early Sunday morning, and that's the type of thing that they might chop to a Netflix.
So yeah, NFL is now five months long.
Lucas Shaw all over NFL this week, And even if you're not watching it that much yourself, what does it for this edition of Blue Bag Technology, don't forget to check out our podcasts. You can find it on the terminal as well as online on Apples, Spotify, and iHeart.
This is vloume leg Technology