Bessent Says Nvidia, AMD Deals Could Be a “Model”

13 Aug 2025 · 38 min · 17 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

This Bloomberg Tech episode covers: (1) US-China chip export policy and market reaction; (2) Tencent’s latest earnings and AI/gaming/ad momentum; (3) CoreWeave’s post-earnings AI data-center scaling; (4) a Bullish crypto IPO; and (5) broader AI/markets context.

Key claim

Treasury Secretary Scott Bessent says the US will not sell “advanced” chips to China (e.g., H20-class and similar lower-tier parts only), to prevent Huawei from gaining a “digital Belt and Road” footprint. He frames it as a “unique model” that could be expanded to other industries, potentially involving revenue sharing for market access.

Notable examples

H20 export-license reversal; discussion of “unenhanced Blackwell” talks; AMD MI308 mentioned; Nvidia/AMD stock moves; CoreWeave inference demand >50%; Tencent’s Peacekeeper Elite +30% user volume and +20% ad revenue; Bullish IPO priced at $37.

Guests

Anne-Marie Horden (Bloomberg TV co-host, interviews Bessent); AllianceBernstein (speaker Lechu/Lei, investment view); Jacob Cook (WPIC Marketing & Technologies CEO); Michael Entretter (CoreWeave CEO); Mitchell Green (Lead Edge Capital founder/partner).

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Effective Work Strategies

0:00 to 0:29

Learn about working smarter within better systems rather than harder.

“The most effective people at work aren't working harder than everyone else.”

Managing Organizational Risk

0:46 to 1:23

Explore the complexities of risk management in midsize and large companies.

“It could stem from regulatory requirements or challenges tied to a specific industry or the scale of an operation.”

Managing Organizational Risk

1:28 to 2:09

Explore the complexities of risk management in midsize and large companies.

Market Insights and Trends

2:37 to 3:14

Analyze market trends and implications of Federal Reserve policies.

“But first, we check in on these markets that are very close, if not at record highs.”

Nvidia and AMD Market Dynamics

3:14 to 5:01

Discussion on the market movements of Nvidia and AMD in context of US-China relations.

“It's really hard to see the catalyst, a causal link for why Nvidia is down a percentage point and AMD is up 4.5%.”

US-Chinese Relations and Chips

5:01 to 7:18

Insights on US chip export policies and their implications for China.

“about what it means to make sure that NVIDIA or AMD have a footprint in China because they don't want Huawei to get control of that.”

AI Market Competition and Strategic Decisions

7:18 to 12:31

Examine the competitive landscape in AI and the impact of market access strategies.

“The very quick question is what happens next, right?”

AI Market Competition and Strategic Decisions

13:38 to 15:05

Examine the competitive landscape in AI and the impact of market access strategies.

“The thing about AI for business, it may not automatically fit the way your business works.”

Tencent's Strong Performance

15:37 to 20:49

Discover Tencent's impressive growth in gaming and advertising amidst market challenges.

“They say that they have enough AI chips for now.”

Geopolitical Insights on Chinese Tech

20:55 to 22:21

Explore the impact of geopolitical tensions on Chinese technology companies like Tencent.

“Relationships have been built up over decades.”
Show all 17 chapters

Bullish IPO Highlights

22:34 to 25:09

Get insights into the recent IPO of Bullish and its market implications.

“raised more than a billion dollars in an IPO pricing its shares above the marketed range and the digital asset exchange operator and also the owner of media outlet Coindesk saw 30 million shares yesterday for$37 each.”

AI Demand Dynamics

25:17 to 28:00

Learn about the shifting demand for AI training and inference within the industry.

“We're waiting for trading to start, but bullish shares indicated open somewhere$60 to$65 each after pricing at 37.”

Demand Dynamics in AI Infrastructure

28:00 to 33:36

Explore the shifts in demand for AI training and inference capacity.

“And we haven't lost demand for training new models, but we have been incredibly pressed across the industry for compute for additional inference capacity.”

Expansion Opportunities in AI and Tech

33:36 to 39:56

Learn about diversification strategies and emerging sectors integrating AI.

“This is Bloomberg Tech and you're looking at a live shot of the Printable Room.”

Integrating Healthcare Solutions

42:00 to 42:28

Learn about how technology is being used to streamline healthcare services.

“Things like making it easier to get care that looks at the whole person, from primary care doctors to mental health support and even in-home care, and then using technology to make sure they all work together.”

Recent Space Launches Overview

42:38 to 45:56

Get insights into ULA's Vulcan rocket launch and Europe's Ariane 6 mission.

“Let's get more from Bloomberg's Lauren Grasch, who leads our space coverage.”

Recent Space Launches Overview

46:01 to 46:37

Get insights into ULA's Vulcan rocket launch and Europe's Ariane 6 mission.

“As industries evolve faster than ever, companies need an environment that accelerates strategic growth, and Michigan delivers on that promise.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00The most effective people at work aren't working harder than everyone else. They're working smarter, inside better systems. Superhuman Go, from the makers of Grammarly, is the AI assistant that works inside every tool you already use, always ready, already aware of what you're working on. It's a teammate whose only job is to help you be better at yours. With Go working with you, you can show off what you do best. See what Superhuman Go can do at superhuman.com. When you're running a business, the best days are the ones where priorities stay on track. For midsize and large companies, that isn't always easy.

0:37Risk can touch multiple parts of an organization at the same time, often in ways that aren't immediately obvious. It might involve property, liability, or cyber. It could stem from regulatory requirements or challenges tied to a specific industry or the scale of an operation. At that level, managing risk becomes an ongoing discipline, not a one-time decision. At The Hartford, the focus is on helping businesses manage risk before it turns into something more disruptive. That means working with companies to identify where they're exposed, decide what matters most, and put practical standards in place so risk is managed as part of day-to-day operations.

1:13And when losses do happen, The Hartford can pair that risk control work with insurance coverage grounded in underwriting, risk engineering, and claims experience developed over time. Learn more at thehartford.com slash risk mitigation.

1:53Get wise. T's and C's apply.

1:59Bloomberg Audio Studios. Podcasts, radio, news.

2:09Bloomberg Tech is live from coast to coast with Caroline Hyde in New York and Ed Ludlow in San Francisco.

2:20This is Bloomberg Tech. Coming up, Treasury Secretary Scott Besant says the US is open to discussing the use of NVIDIA chips with China. Plus, China's Tencent shows fast growth across its gaming and ad businesses and plans for smart AI spending. And our conversation with the CEO of CoreWeave after its earnings, we discuss how the company plans to scale to meet AI demand. But first, we check in on these markets that are very close, if not at record highs. Once again, Ed, we drive higher and it's more about Federal Reserve policy. Will we get a mega cut coming later in the year? NASDAQ 100 clings to gains, but we are near a record or at a record high.

2:58So too, for Bitcoin, up 0.7%. Ether's higher too. We now have$4 trillion market cap for the entirety of the crypto market. Just about the same size as, say, Nvidia. What are you looking at? Yeah, I'm looking at Nvidia and AMD. Now I'm going to be really clear here. It's really hard to see the catalyst, a causal link for why Nvidia is down a percentage point and AMD is up 4.5%. Social media volumes for AMD have quadrupled in the last hour. People are talking about it. There's a reason why they're talking about it. It's because this is what Treasury Secretary Scott Besson told Bloomberg's Anne-Marie Horden earlier today.

3:36We would not sell any of the advanced chips. So the H20s, I don't know whether you'd say there are four, five, six levels down the chip stack. What we do not want here, Anne-Marie, is for Huawei to have a digital belt and road. So we do not want the standard to become Chinese across the world or even in China. Let's understand this administration's policies for chipmakers and speak to Bloomberg surveillance co-host Anne-Marie Horden, who just conducted that interview. The underlying assumptions changed actually this week, whether or not China even wants America's reduced power chips. But what I found so interesting about the content of what Besant was telling you is that they feel that it's not really even a national security concern.

4:25They're very clear that these are lower performance chips and they want American tech to be present there. Give us the digest of what else he said. Yeah, that was really his answer to my question of what do you say to critics that basically think this administration is putting revenue in front of national security concerns, that export licenses now look to be seemingly for sale. He said that the H20 chip, and I guess similar to AMD, the MI308, is four, five, six stacks down from that higher level chip. And they're never going to allow China to get to the higher level chip. He almost sounded a little bit like someone you speak to often, Ed, when it comes to Jensen Wang, about what it means to make sure that NVIDIA or AMD have a footprint in China because they don't want Huawei to get control of that.

5:15And the big question mark is now is that what is the appetite in Beijing for these chips, given the fact that there's been this new urgency since the administration is allowing these export licenses once again from the CCP to send out these letters to firms telling them to snub these chips and go for some of their own national champions. And this is something the Treasury Secretary told me and Jonathan that he is going to bring up with his Chinese counterpart. Anne-Marie, what's also interesting is he also told you and Jonathan that maybe this could be replicated. At the moment it's unique, he mentioned, but maybe it applies to other chip makers, other industries?

5:53He said it's a unique model, something that the president likes to do, try different things. But yeah, that's something that caught me off guard as well, or something that we've been also just exploring every day with guests, whether or not this is new rules of engagement and rules of the road for corporate America. Can other companies potentially strike these types of deals with the United States or maybe the other way around? Could the Oval Office and the president tell other companies, if you want to gain access to a market, we're going to have to get part of that profit. Because what's going on right now, and this has been a really long saga, we should note, for the H20 chips.

6:32Remember in April, there was an export ban on these chips. Then this was reversed. In the meantime, we had U.S. and Chinese counterparts to having trade negotiations. And a lot of people glean into this. And the Treasury Secretary told me in a previous interview this, H20 was a chip part of the mosaic of these trade negotiations because the U.S. needed rare earths. And now that the U.S. has lifted the export control, instead of just allowing the export licenses to come back, It seems the president struck this deal one-on-one with Jensen Wang in the Oval Office, reported the president threw out 20 percent, and then Jensen Wang said, well, how about 15 percent?

7:11And then what does this mean is a big question for other companies and corporate America down the road. The very quick question is what happens next, right? Because Treasury Secretary Besant, on top of what the president said earlier in the week, says that they'll discuss the idea of unenhanced blackwell chips, essentially, with China. He leads talks just very quick. What does happen next? Well, I think it depends on the leverage in these negotiations and what the U.S. wants to get out of China. We also had the president the other day talk about the purchasing agreement he struck in his first administration when it comes to things like soybeans.

7:47There is now a 90-day extension, which brings us to November, for these trade talks between these two counterparts. And in my previous interviews with the Treasury Secretary in July, he talked about how now there'll be more of a frequent cadence between him and his Chinese counterpart to have these discussions. All of this likely will lead up potentially to a meeting between Chairman Xi and President Trump, either on the sidelines of the APEC summit or even maybe the president going to China. Bloomberg Surveillance co-host Anne-Marie Hildon, it was a brilliant interview. Thanks for breaking it down for us.

8:20Let's delve more into the corporate implications. China, US, the AI competition. Nature is with us. Alliance Bernstein joining us now. And I'm really interested as to how Alliance Bernstein thinks about pricing this in, if at all. How are you looking at the opportunity to access the Chinese market going forward? Well, I think China has always remained one of the most strategic markets and will remain so. And particularly in the AI world. If you think about AI, how we think about the scale benefit, right, the ecosystem. U.S. has always been, for a big part, the origination of innovations, particularly technology innovations.

9:04But China has done a phenomenal job in terms of using it and then using it at a very large scale. When you think about in the AI context, having an ecosystem globally is this very important strategic advantage long term. The more data you have, the better the model becomes. And that's not just for China alone, but think about globally. Who wants to be the dominant standard? Having the operating system, that layer, is very, very strategic for most companies longer term. So that's why we're thinking, that's what Jensen said, that's what Lisa Su at AMD said. And I think that's how we're thinking about it, which is who will be the dominant AI standard?

9:46We want to have a one global ecosystem or are we going to have different forks into the future? But these so-called dumbed down, I mean, President Trump even called them obsolete chips being allowed into China. Does that ensure that they remain part of the tech stack? Well, it's too early to tell for the most part. I think necessity usually drives innovation. So when we don't export anything, then it probably is a greater incentive for China to develop its own domestic ecosystem that's completely separate. So this is a bridge. I think in my mind that we can bridge that big, drastic action that we've seen early on this year.

10:29I think China does have advantages to some extent, and it has a more newly developed power infrastructure system. It has a vast amount of data, and it's developed a lot of these open source models that they are really rapidly pushing. So it remains to be seen, but I think it's quite important and quite strategic for the U.S. companies to think about the global standard. Think about iPhone, for instance, right? Yes, iPhone or Tesla, the data resides in China. We respect that. But that said, there's still lessons that we learned from adoption. At this point, if you think about a lot of AI models, we've trained a lot of data already.

11:12But going forward, the use of AI, the use of applications, that's new sets of data that's constantly being generated. And we don't want to not have access to that. Lei, the Treasury Secretary indicated to us that this pay to play or quid pro quo of a revenue share in exchange for access to the Chinese market could be expanded. What I want to know from you is if that were to be the case with high bandwidth memory or telematics or other parts of the server design, would you then upgrade your overall outlook for, in megawatt terms or gigawatt terms, the amount of installed capacity we could see around the world?

11:55I think it's too early to tell. I think as you've seen from a policy standpoint, there has been quite a bit of fluidity in terms of the changes back and forth. But I think it's certainly promising that, you know, some of the really drastic assumptions that we had in April, some of it's being discounted today. So it remains to be seen. And I would say it's very early. And I would refrain from making any assumptions whether too aggressive or I would say it is definitely an improvement from the worst case scenario. Lechu of Alliance Bernstein, thank you very much. The people who seem to get more done than everyone else, they're not working longer hours or running on more caffeine.

12:40They've just stopped wasting time on the stuff that doesn't move work forward. Switching apps, re-explaining context, hunting for files, those aren't small inefficiencies. They're hours wasted every week. Superhuman Go gives you those hours back. From the makers of Grammarly, Go is an AI assistant that sits inside every tab and tool you already use, always available and already aware of what you're working on. Ask it to draft something, summarize a long thread, pull up a file, or prep you for a meeting. Go handles it without you ever leaving the page you're on. This is what it looks like when AI actually fits into your work, instead of adding to it.

13:23It's like having a teammate whose only job is to help you be better at yours. Go keeps up so you can move forward. With Go working with you, you can show off what you do best. Superhuman Go. Find out more at superhuman.com. That's superhuman.com. The thing about AI for business, it may not automatically fit the way your business works. At IBM, we've seen this firsthand. But by embedding AI across HR, IT, and procurement processes, we've reduced costs by millions, slash repetitive tasks, and freed thousands of hours for strategic work. Now we're helping companies get smarter by putting AI where it actually pays off, deep in the work that moves the business.

14:07Let's create smarter business, IBM. Let's talk about healthcare for a second. It doesn't always work the way people expect it to. If you've ever waited on a prescription refill or had a hard time getting the care you needed, you know the feeling. The system should just work better for everyone. That's exactly what the people at Optum are trying to do every day. They're a health care company linking patient care and pharmacy services and using data and technology to drive the whole system so care is connected, not complicated, for patients and providers. Things like making it easier to get care that looks at the whole person, from primary care doctors to mental health support and even in-home care, and then using technology to make sure they all work together.

14:48Technology designed to help doctors spend less time on busy work and more time with their patients. And those prescriptions? Optum is working to bring costs down, save patients money, and make it easier to get refills. Little by little, Optum is helping make healthcare work as one for everyone. Head to business.optum.com to see how.

15:36like yeah thanks for having me and it's pretty interesting because when you look at their capex budgets for the quarter they spent about 19 billion RMB in a quarter so that has been you would say a fraction of what US tech giants have been spending and on the call they mentioned that they will be spending on AI chips that's a priority they also be recruiting talents however they will adopt this more prudent approach so it does seem that they're not in a rush to spend as much as they can. They say that they have enough AI chips for now. They say that their stockpile is enough for training and for inferences, and that's probably why.

16:16There is some skepticism from the analyst community in the same way that they look at the American companies on how CapEx translates to top-line growth anyway, specific AI growth. But in the quarter gone, there are just parts of the business that are doing well. Where is Tencent strong, Henry? Yeah, I would say they're firing on all cylinders for this quarter. So the games has been a standout. International games grew by about 35 % in a quarter. They launched some new games, including one called Dune Awakening in international markets. But domestically, it's about their evergreen games. So we're talking about the Peacekeeper Elite, which is a six-year-old game.

16:55The company said that it's gaining traction because of a new game model. that game saw a 30 % increase in user volume for the quarter. That's about games. But on ad revenue side, it's growing very strong as well, 20 % increase for the quarter. The company said it's benefiting from AI tools being adopted over there as well. So it's basically strength across the board. Bloomberg's Henry Wren, thank you very much. Let's get more and bring in Jacob Cook, co-founder and CEO of WPIC Marketing and Technologies in e-commerce and technology consultancy focused on the Asian region. Somebody asked me a question earlier today that fits into the Tencent story so well, which is we've been so focused on the movement of chips across borders or a lack of, we haven't really talked about the movement of software, right?

17:42And when I think about Tencent and the gaming portfolio in particular, are there Chinese domestic companies that are getting international foothold across e-commerce, gaming and other digital businesses? Yeah, definitely. I mean, even Tencent is one of those with their big investments into Shopee really kind of being a top leading player in Southeast Asia in terms of e-commerce. But, yeah, I mean, Tencent is there taking a lot of the advertising dollars, you know, that are coming from that big e-commerce boom inside China as well, too. They're not so much of a player as, say, Alibaba, you know, in terms of the B2B software that they have a whole suite of.

18:16But, you know, in terms of the advertising, becoming a bigger and bigger player in that part, we saw that revenue also up by 5 % this quarter. You know, the theme from Meta's earnings, if there was some softness in advertising, it was a pullback of Chinese advertisers in the European region in particular, perhaps in the United States, because of the geopolitical environment. I'm trying to read through what's going on with Tencent, and I just don't see that macro-level headwind. Do you? No, and I think that really what you're seeing, though, is because a lot of that advertising is still based in China, that we're coming out of a year of pretty big stimulus in terms of the government and e-commerce spending.

18:55So we should naturally see that trickle down come into about now, Q3 and Q4 as well, being pretty strong in terms of the advertising dollars that they're coming in. No, they never really had that major post-COVID stimulus in China, and they waited a little while. So we're seeing good numbers now, and we would expect that they would hold because low inflation numbers in China as well. It also bodes really well for the consumers. So we're pretty excited about Tencent's earnings, and we think that probably you're going to see some pretty more good news from the e-commerce providers in China in the coming weeks as well.

19:25Yeah, we brace ourselves for Alibaba later in the week, for example, Jacob. I want to hear how much access they're having to GPUs, how much AI and generative AI might be held back by ultimately the infrastructure they need. It sounded as though, once again, Tencent trying to say, we've got all the chips we need for inference. We're not worried about it. They wouldn't go into the nitty-gritty about NVIDIA access, for example. But, Jacob, is it a worry? Well, I mean, we've all been saying this, but we've now seen, you know, three successive, like the QN model exactly from Alibaba that's now leading.

19:58Not leading, but certainly up there in terms of its ability to code. And they're going down a little bit different path at Alibaba and DeepSeq as well. I mean, all of those models are competing internationally despite the chip ban. So if Tencent comes and says that, you know, they have all the chips they need, there's no reason not to believe that. More broadly, the geopolitical context. It doesn't seem to be knocking the wind out of owning these Chinese companies' sales. I mean, they've all recovered significantly from their lows. Yeah, I think that one of the things, you know, for us that are boots on the ground in China, you know, there is the really high-level government that, you know, where the relations are not great and we see the headlines.

20:36But, you know, for boots on the ground and for the local level, it's actually been very, very strong. You know, people-to-people connection is still very good over there. And there's still a lot of businesses that are setting records, international business that are setting sales records in China. So, you know, there's the headlines, but actually on the ground, I think people have been doing business over there for quite a long time. Relationships have been built up over decades. So we're kind of expecting this to blow over, just like it kind of did with the first Trump administration, and people will get back to business.

21:05You still have a president that in a press conference is talking about a pay-to-play arrangement where the CEO of the biggest company in the world, NVIDIA, has gone to him and agreed to a revenue share, right, to get access to the Chinese market. In a tariff or a trade dispute in China, does the same thing happen? Do you have the leaders of these biggest technology companies talking about and acknowledging their government's policies in the same way and vice versa? Well, no, you don't, certainly. But, I mean, there's almost the exact opposite approach. I mean, there's been an export tax credit out of China for a long time, too.

21:44So it's almost like they're going the other way and trying to reduce prices for international consumers as opposed to the strategy of essentially taxing exports. So, no, through all these things, like we said, no one's really had a problem getting access to what they needed to get access to. and pricing seems to be okay in the Chinese market. So, yes, there's going to be some headwinds. Q1 was particularly tumultuous in the US-China business cycle, but we've been through this before, and we're probably going to get through it again. And the numbers, I think, are starting to say that we're probably getting over the hump here.

22:19We'll see how the rest of the earnings come through for these Chinese giants. Jacob Cook, it's always great to check in with you, the CEO of WPIC Marketing and Technologies.

22:33Crypto company Bullish raised more than a billion dollars in an IPO pricing its shares above the marketed range and the digital asset exchange operator and also the owner of media outlet Coindesk saw 30 million shares yesterday for$37 each. Bloomberg's Anthony Hughes joins us as we await the company to actually start trading on the NYSE. They're indicated to open$65 to$70 each, that not quite double but it's another big pop for another hot retail eIPO it feels like. Yes, we've had a number of strong IPOs recently, obviously Circle in June and then Figma a few weeks ago so this is really directionally a very similar story and we've seen very strong demand and we reported earlier today that there was north of 20 times over subscription levels for this IPO and that a lot of institutional investors got got no shares in this offering and from all accounts there's been a lot of action amongst the retail brokers who got a little bit of the stock as well.

23:26So, yes. NT, I don't know too much about bullish, but I was interested to read that across co-founders and board members, they're taking a big chunk of the shares or have a big chunk. Yes. From what we heard, when the shares are allocated here, that people who had a relationship with management did receive some of the shares. And that's not completely unusual. But often with these hot IPOs, you get management having a big say in the allocations because they want to place the shares with, I guess, people who they think will be long-term supporters of the shares or long-term supporters of the company, or they may have other reasons for giving them the stock.

24:03But certainly in this case, that seems to be something that's quite a major aspect of the way in which the shares are allocated here. Brendan Blumer, BlockOne CEO, co-founder. He's going to be the biggest holder, about 30%. ARK gets a cut, so does BlackRock. What's interesting though is to think about how Tom Farley became the CEO. Originally it was because he was leading a SPAC, a special acquisition company, that was going to buy and merge back in the day for$9 billion. This is a strong IPO but it's not 9 billion valuation. It's interesting we saw Circle have to do the same there. SPAC unwound back in a few years ago.

Read the full transcript

24:41Then they come back to the market and IPO strongly, eToro another. They are managing to come but not perhaps at the heady valuations of the past for a bullish. Yeah, although I think in the case of Circle, it quickly reclaimed the SPAC valuation and went much, much higher. So, you know, it is really interesting that a lot of these SPAC mergers back in that 2021 era that fell over for reasons that were partly to do with uncertainty around the ones that were crypto related. There was uncertainty around crypto regulation and aspects like that, but also there was just a big sort of sell off in a lot of growth names as well in that period.

25:16So, you know, it just tells you that we're sort of somewhat back to that sort of 2021 period. Not quite as extreme as that, but, you know, we're certainly in a situation, in an environment where we're looking at sort of lower rates and the market at record highs, you know, we're sort of starting to get a lot of hot money in the IPO market, but also, you know, a lot more IPOs are probably going to come in the later stages of 2025 as well. We're waiting for trading to start, but bullish shares indicated open somewhere$60 to$65 each after pricing at 37. Bloomberg's Anthony Hughes, thank you very much indeed.

25:56Welcome back to Bloomberg Tech. Check in on these markets which are flooding between gains and losses on the Nasdaq 100. We're currently just off by eight points. But we are near record highs. Once again, all eyes on the Fed policy, whether we get the big cuts. It's helping crypto. We're up 0.4%. We're coming down from the previous highs earlier in the session, but still at 120 ,000, you're at a record high in the market cap of crypto now four trillion about the same as Nvidia let's look at Nvidia let's look at the chip stocks more broadly Ed because they are in the eye of the geopolitical storm right now Nvidia off by 1.6 % after we of course digest really what a 15 % payment on all your H20s to China really means to the business how much that's going to extrapolate into China not wanting those chips AMD though up 3.7 % interesting this difference here that we're seeing in the two key names I'm looking though also at Micron off by more than 2 % Ed you've pointing this out.

26:45Now remember it has rallied the four days prior to this almost 20 % after it pre-announced some of its numbers. They look strong in DRAM but interesting reports of course that they are laying off people in China at the moment. Let's look at another mover and actually this decline has massively accelerated. Shares of CoreWeave are now down almost 19-20 % in the session. They'd been down nearer to 10 % after the company posted a disappointing earnings outlook. Margins are under pressure from this kind of rapid AI data center expansion. We spoke with the CEO, Michael Entretter, earlier today. The message was training demand sustained.

27:23Inference demand fast growing. Listen to this. We continue to see massive demand for training, right? Like that hasn't abated in any way. But what you're seeing is a continued cranking up of the demand for inference. And we see it within our portfolio of data centers because when the compute is being dispatched, if it's being used for training, you'll see a step function, right? The training job begins, it runs, it steps down when it's over. But when you're viewing the compute being used for inference, what you see is a much more gradual increase and decrease as you move through the day, as people begin to come on and off their demands for artificial intelligence.

28:11And we haven't lost demand for training new models, but we have been incredibly pressed across the industry for compute for additional inference capacity. What's the percentage split? inference training for you? The last time I looked, and it does move a lot, it was over 50 % for inference at this point. Okay, that's an interesting change because I think I ask you that quite regularly and it seems to be at that mark. The thing that I've been thinking about a lot recently is the aging GPU. Basically, Nvidia commits to that change on an annual basis. In the context of the demand picture you just outlined, but also what you're trying to do in scaling, how do you keep up with the aging GPU scenario?

29:03Yeah, so look, what we're seeing in our infrastructure is a consistent movement of the newest iterations of architecture that come out of Nvidia being used to drive the most bleeding edge training infrastructure jobs required. And what happens is, is the prior generation begins to pick up the load on inference and you see it kind of passing from one generation to the next generation to the next generation. We are not seeing meaningful declines in the H100 demand, which is the prior generation to the the Grace Blackwell that is the cutting edge right now. We're also not seeing any real decrease in demand for the Ampere series, which is even prior to the H100s.

29:58There is broad based inference demand absorbing that compute. It's being picked up into term contracts by many of the same players that are looking to support the ever growing demands of inference. Are you diversifying the business, Michael, away from just OpenAI and Microsoft? Yeah, look, it's a huge focus for our organization. We've made tremendous progress. I talked a little bit about it in the earnings last night as we take on new clients, as we basically penetrate additional layers of the enterprise space. I talked a little bit about how we're not only seeing diversification of clients, but we're also seeing green shoots in new parts of the economy that are beginning to integrate artificial intelligence.

30:51And I focused a little bit on two sectors, and it's just two sectors that we chose. The one was the VFX space, where we saw companies like Moon Valley come in and begin to integrate artificial intelligence into visual effects. You know, it's an incredible growth sector for us. We're seeing a lot of new companies begin to ramp there. We're also seeing real developments begin to occur in the life sciences portion of the market. There are, you know, I spoke about Hippocratic AI and, you know, like those are wonderful developments. It really is speaking to a resilient broadening of artificial intelligence as it penetrates the economy.

31:38Michael, we've been studying really closely how XAI went about building Colossus and what Meta plans to do with its next gen capacity. They very much kind of go it alone. And I wondered if you talk a little bit about the opportunity or maybe lack of opportunity you see to work with those two names. Yeah, so Meta's a client of ours, you know, and we work with them very well. We think there's lots of opportunities for us to expand our relationship with them and to broaden their use of our computational infrastructure. We think that, you know, the work that X is doing at their sites is incredibly impressive.

32:24You know, but we believe that ultimately when you're dealing with the scaling exercise that looks like what is going on within AI, you're always going to wind up using partners. It's just the nature of it. You need to build so much so fast so broadly. We're very confident that there's really great opportunities for us to continue to work with a broad set of companies. You know, I spoke a little bit about this on the earnings. also, which was, you know, in the last eight weeks, we've seen two of our hyperscaler clients come back and execute extensions and broaden their contractual relationship with us.

33:08It's very exciting. And it is representative of space that's broadly trying to address this systemic imbalance of which, you know, CoreWeave is a wonderful solution. We deliver, you know, best-in-class technology, best-in-class software stack to be able to allow different users across the space to really take advantage of the infrastructure that we're building. What a rich deep dive with the CallWeas CEO. This is Bloomberg Tech and you're looking at a live shot of the Printable Room. Check out the Bloomberg Tech podcast. You can find it on the terminal as well as online on Apple, Spotify and iHeart.

33:47This is Bloomberg.

33:55This is a very unique solution. Allows NVIDIA to expand into China. It can make NVIDIA chips the bellwether for Chinese technology and then the US taxpayer gets a share of that. Key word, unique. Is it unique to NVIDIA and AMD or is this a model for other companies? I think we could see it in other industries over time. I think, you know, right now this is unique. But now that we have the model and the beta test, why not expand it? Treasury Secretary Scott Besson joining Bloomberg TV earlier today. Let's talk more about what this unique deal can mean for companies in the U.S. and in China. Mitchell Green is the founder and managing partner of Lead Edge Capital.

34:41You're an investor in Chinese names of the past, the IPOs, Alibaba, ByteDance, you hold, of course. But you're also in many other areas and industries. We heard in that conversation with Scott Besson that maybe this does apply to other industries. And maybe will this be a pay-for-play more broadly? What do you make of it? You can figure out how to read the U.S. and Chinese government relations between what they say and what they all do. please keep me enlightened and inform me because we're as confused as every other viewer that you listen to I do think both of these countries' leaders fully know that they need to work together 150 % tariffs on our companies selling to them their companies selling to us is just stupidity on all facets of it we're going to work together as it comes back to technology you know, AI and the implementation of it is very global.

35:42I mean, look, I'm talking in my own book here that we were big investors on a dollar basis and ByteDance obviously very small as a percentage, just given how big the company is. But we think ByteDance is one of the foremost AI companies on the planet. Is there a worry, Mitchell, though, that if their own government is telling them that they shouldn't really be going for H20s, yes, they're called obsolete by President Trump, but they're not really able to access them as much as they'd like, Mitchell. Is that a worry for the future of ByteDance as the AI powerhouse alongside some of the other Chinese rivals?

36:14I would not bet against Chinese ingenuity in designing solutions and systems around, you know, you saw what happened in DeepSeek. I mean, who knows how much they spent, but obviously they trained a model and used it in independent tests that was a fraction of a fraction of a fraction of what it cost some of these US people to do without who knows if they had access to these chips or not. I wouldn't bet against the Chinese semiconductor industry. Now, it's not, I'm not sure, we're not semiconductor experts, but I think if NVIDIA chips are not allowed over there or certain chips aren't allowed in there, I wouldn't bet against these companies and these entrepreneurs from figuring out ingenious solutions around designing them.

37:00Right. The underlying assumption the Treasury Secretary's meeting is that China wants America's chips, irrespective of how much revenue those companies pays the Treasury or wherever it goes anyway. I actually wanted to ask you about software. I had a funky week here in London, and various European and British software names kind of went down, and Monday.com was at the heart of it. That's the story where it's like no code, traditional SaaS, has bad print, and the market's like, Like, oh, AI makes all of this kind of legacy software stuff not very good anymore. Right. Could you explain that to me?

37:34You kind of know this field. Yeah, so it's funny. At a higher level, a lot of these software companies are actually in no man's land. You have half the investors out there screaming for growth. You have the other half of the investors out there screaming for profits and want to value them on multiples of profits. And it's like this tug of war. The companies aren't growing as fast now because they're really, really freaking big. And the much smaller companies in software are still growing fast because they're just not nearly as big. When you're$2 billion of revenue, it's a lot harder to grow 40 % than it is if you're like$200 million of revenue, right?

38:15I can tell you we are investors in – our primary business is investing in application software companies. And I will tell you that almost every one of our companies is using AI to become more productive. We have a company up in Toronto, Gravity, that may have, I don't know, 10 to 15 software engineers. They can now use it to have the equivalent of 45 software engineers and just be more productive. But they're not firing all their software engineers. They need people to work with this code. our belief is that if you are software that's a system of record and you have data, we actually think it's really, really valuable.

38:56I don't have to tell the viewers out there that if you

39:02legacy companies a lot of legacy companies are going to massively benefit from AI just like they did with the internet. The biggest beneficiary of the internet may have been like, you know, literally, iPhone, people like Amazon and Facebook and Google and Microsoft. It was around before mobile. Like, these were the guys that benefited. You know who's going to win? Snowflake, Salesforce, Workday, Datadog, private companies like Gafana, public companies like Toast. All these companies are going to use them to be that much more productive. We think 90 to 95 % of the AI application software companies getting built today are zeros.

39:44Horrible economics, negative gross margins, burning money like crazy. Okay, so in a year's time, let's check back and we'll keep track of that one. Mitchell Green, founder and managing partner of Lead Edge Capital. Great to have you back on the show. The people who seem to get more done than everyone else, they're not working longer hours or running on more caffeine. They've just stopped wasting time on the stuff that doesn't move work forward. Switching apps, re-explaining context, hunting for files. Those aren't small inefficiencies. They're hours wasted every week. Superhuman Go gives you those hours back.

40:19From the makers of Grammarly, Go is an AI assistant that sits inside every tab and tool you already use. Always available and already aware of what you're working on. Ask it to draft something, summarize a long thread, pull up a file, or prep you for a meeting. Go handles it without you ever leaving the page you're on. This is what it looks like when AI actually fits into your work, instead of adding to it. It's like having a teammate whose only job is to help you be better at yours. Go keeps up so you can move forward. With Go working with you, you can show off what you do best. Superhuman Go.

40:57Find out more at superhuman.com. That's superhuman.com. So there's a lot of noise about AI, but time's too tight for more promises. So let's talk about results. At IBM, we work with our employees to integrate technology right into the systems they need. Now, a global workforce of 300 ,000 can use AI to fill their HR questions, resolving 94 % of common questions. Not noise. Proof of how we can help companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create smarter business. IBM. Let's talk about healthcare for a second. It doesn't always work the way people expect it to.

41:36If you've ever waited on a prescription refill or had a hard time getting the care you needed, you know the feeling. The system should just work better for everyone. That's exactly what the people at Optum are trying to do every day. They're a healthcare company linking patient care and pharmacy services and using data and technology to drive the whole system so care is connected, not complicated. for patients and providers. Things like making it easier to get care that looks at the whole person, from primary care doctors to mental health support and even in-home care, and then using technology to make sure they all work together.

42:11Technology designed to help doctors spend less time on busy work and more time with their patients. And those prescriptions? Optum is working to bring costs down, save patients money, and make it easier to get refills. Little by little, Optum is helping make healthcare work as one for everyone. Head to business.optum.com to see how. Two launches blasting off in the last 24 hours, ULA's Vulcan rocket on its first national security mission, and Europe's Ariane 6, completing just its third mission ever. Let's get more from Bloomberg's Lauren Grasch, who leads our space coverage. I'm going to start with the ULA launch.

42:48Tell me about it. Why is it significant? Well, this was the third launch for the Vulcan rocket, but as you mentioned earlier, It was their first national security mission, which is really what Vulcan was designed to do. ULA is one of an elite group of launch providers that can launch the Defense Department's most sensitive national security satellites. But this was their first time doing so with this particular rocket. They had to receive certification for it. So first they had to launch two rockets or two missions with Vulcan to prove to the Space Force that it could handle launching national security satellites.

43:25And it actually suffered a tiny issue on its second flight. A strap-on booster had a small explosion. The rocket still made it to orbit, but it delayed the certification process. However, it did receive that certification in March and then ultimately was able to do this launch. so that'll pave the way for even more national security missions moving forward. And turning our attention to Europe and Ariane 6, how much is this just countries, entire areas of the globe trying to wean themselves less on dependency from SpaceX and have their own competitors? Yeah, I think there was a lot of symmetry last night because these were the third flights of both of these rockets.

44:06And both of these vehicles are in a similar class to that of the SpaceX Falcon 9 rocket. which has essentially had a de facto monopoly on the launch market for the last few years, at least with this particular class of rockets. So the launches of these two vehicles kind of do symbolize, perhaps there are going to be more options for other satellite providers in the future looking for this class of vehicle. Of course, it's going to take some time. As I said, these are the third flights. They've both struggled to ramp up in terms of their launch cadence, whereas the Falcon 9 is launching every few days.

44:42so it's going to take a while for them to reach up but at least there are more and more rockets of this caliber coming online that satellite operators can turn to and perhaps reduce that logjam Bloomberg's Lauren Grush great reporting as always thank you very much indeed now that does it for this edition of Bloomberg Tech what a lot of earnings we've still digested and got to come yeah and a critical conversation with the Treasury Secretary recap on the podcast you know where to find it on the Bloomberg Terminal as well as online on Apple, Spotify and on iHeart. From London and New York City, this is Bloomberg Tech.

45:25When you're running a business, the best days are the ones where priorities stay on track. For midsize and large companies, risk can affect multiple parts of the organization at once, from property and liability to cyber and regulatory challenges. At that level, managing risk becomes an ongoing discipline. At the Hartford, the focus is on helping businesses manage risk before it turns into something more disruptive. And when losses do happen, that work is paired with insurance coverage shaped by years of underwriting, risk engineering, and claims experience. Learn more at thehartford.com slash risk mitigation.

46:00Policies provided by Hartford Fire Insurance Company and its property and casualty affiliates, Hartford, Connecticut. As industries evolve faster than ever, companies need an environment that accelerates strategic growth, and Michigan delivers on that promise. From emerging startups to global enterprises, Michigan offers what executives value most, a resilient, innovative ecosystem, diverse communities that attract top talent, and a quality of life that supports work-life balance. With our unified Team Michigan approach, businesses scale faster and compete at the highest level. Michigan, pure opportunity.

46:33Seize your opportunity at michiganbusiness.org. wise is the smart way to manage the currencies you need around the globe when you send money abroad using your bank you could get hit with hidden fees and exchange rate markups there's a better way try wise wise uses the exchange rate you'd usually find on google with no unwelcome surprises plus most transfers happen in under 20 seconds which means your money arrives in less time than you've been listening to me it's simple and free to sign up when you download the wise app be smart get wise t's and c's apply Thank you.

From the publisher

Bloomberg’s Caroline Hyde and Ed Ludlow discuss Treasury Secretary Scott Bessent’s comments that the US deal allowing Nvidia and AMD to sell some AI chips to China could be a “model” for others. Plus, Tencent shares rise on a revenue beat and plans for AI. And CoreWeave CEO Michael Intrator discusses how the company plans to scale to meet AI demand.

See omnystudio.com/listener for privacy information.

More from Bloomberg Tech

All 342 episodes
Bessent Says Nvidia, AMD Deals Could Be a “Model”Bloomberg Tech · 38 min
Listen in VO