The Six Five Pod | EP 268: The Fed, Inflation, and AI: Decoding Market Signals

On this episode of The Six Five Pod, hosts Patrick Moorhead and Daniel Newman cut through the noise, discussing renewed access to China for companies like NVIDIA & AMD, AI agents on AWS, the significant Synopsys & Ansys deal, and more. The handpicked topics for this week are:

1. AI and Technology Industry Updates: NVIDIA and AMD receive approval to ship chips to China. Analysis of the negotiation process and market implications.

‍

2. Export Restrictions and National Security: A debate on the necessity of GPU export restrictions and looking at the balance between national security and market access.

‍

3. Government AI Investments: DOD's $200 million contracts with AI companies. Analysis of the significance and potential impact of these investments. Discussion of Grok's potential use in government applications.

‍

4. Economic Indicators and Market Trends: Analysis of recent inflation data and its impact on the economy. Discussion of tariffs and their effects on consumer prices.

‍

5. Synopsys-Ansys Merger: A Game-Changer for AI Design: How this deal creates an end-to-end solution for physical AI and edge computing. The new combined company is positioned to lead in next-gen chip-to-system design.

‍

6. Federal Reserve and Political Influence: Speculation on the potential firing of Federal Reserve Chair Jerome Powell.  Examination of the Federal Reserve's independence and political pressures.

‍

7. TSMC's Market Position and AI Boom: Overview of TSMC's recent financial performance. Analysis of TSMC's role in the AI infrastructure boom.

‍

For a deeper dive into each topic, please click on the links above. Be sure to subscribe to The Six Five Pod so you never miss an episode.

‍

Or listen to the audio here:

Disclaimer: The Six Five Pod is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.

‍

Daniel Newman: If we had a little like to queue up for victory laps though. This would be my moment. Like just. Yeah, we need to create that. Producers note that when one of us has a victory lap, there needs to be a Kramer button, you know, and he's like, bye bye, bye bye bye. Like a victory lap. Because I said from the very beginning this was all just a big old negotiation that was going on.

Patrick Moorhead: All right. Six five Pod is back. Man, I am so excited. I get so jazzed up with that music, man. I'm just, like, singing. I'm dancing. Maybe it has something to do with a good workout week. Shoulders feeling great. Back to bench pressing. I'm still doubling up on the peptides, but we'll see, man. Dan, great stuff. Dan, you are not in your office this week. You're definitely not in a bunker. Where are you?

‍

Daniel Newman: It's a very, very small bunker in an undisclosed location that I try to get to at least once a year, that I will never reveal its exact coordinates on the map. You may know, but I'd ask you to keep it to yourself.

‍

Patrick Moorhead: Absolutely, I do. And, you know, it's great to see you, Dan, because, you know, even when you're on vacation, even when you're sitting in the pool, you're not actually relaxing. Pretty much get a message from you. You're working on the funnel. You're working on some new product project, some new thing. You really do inspire me, Dan.

‍

Daniel Newman: Well, relaxing is. There's got to be a certain set of. Of criteria that are happening in the world around me for me to be able to relax. And very few of those things have been happening lately. So all the variables that I would need to sort of allow me to be, like, Zen.

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: You know, you know, they just. It's been a unzen. But it does make me happy. I mean, look, the reason I go on vacation is my family and loved ones is because it's great for them. I like to watch them enjoy some time away because sometimes, you know, between the stress of school and I don't know what else, but there's all those things, you know, they need a break, and so I'm really glad that they get one. But for me, heads down until everything's perfect, which will never be.

‍

Patrick Moorhead: Well, it's Kobayashi Maru. And for all you Star Trek fans out there, you know what that is, and you can appreciate that mine is very clinical, right? If. If my athletics says I'm in my 20s, in  stress, which at least half the day I need to be into, otherwise it turns into a stress ball, then, you know, I don't need to change anything. And you know, what I did with all this good health, Dan. I just ended up doing more work. But anyways, let's dive in folks. And as you know, we've got a reformatted show, we've got the decode, we're going to hit the top news, tech news, we're gonna hit the flip where Dan and I pretend to debate different sides of the topic. Even though we might be in agreement. Dan and I normally are and that's going to be the topic today and then we are going to jump into bulls and bears, what's going on in the markets, a lot of money being made. So let's dive into the decode. Dan, this is good stuff. You and I were doing a ton of broadcasts, a ton of tweets. I mean, I don't know, like 12 tweets, between the two of us. But, this week Nvidia and AMD got the green light to ship their chips to China. Now leave that opinion out. Just give us the breakdown here because we are going to debate the topic of national security and GPUs later.

‍

Daniel Newman: Yeah, the flipple brings us certainly an interesting spin on this one. If we had a little like to queue up for victory laps though. This would be my moment. Like just. Yeah, we need to create that. Producers note that when one of us has a victory lap, there needs to be a Kramer button, you know, and he's like, bye bye, bye bye bye. Like a victory lap. Because I said from the very beginning this was all just a big old negotiation that was going on and I think that's pretty much what ended up happening. But the bottom line is Nvidia, AMD and potentially others like Intel Gowdy are getting a special license. So by the way, all the panic ins that said the four and a half billion dollars of H20 was going to be dumped into the ocean and used as fish food, well, guess what, they're going to be able to sell those and then they're going to be able to launch a B30 which is in some ways with the new memory requirements even more deprecated. But having said that, China wants H20s, B30s and MI308Xs and they're going to get them. And this is a, we think maybe seven and a half billion a quarter for Nvidia, not nearly that much for AMD, but it's a very important market to open up for AMD. And that basically after all the people that said Huawei were going to become the global chip leader and Ascend was going to replace Nvidia and AMD was toast because they lost one access to one of the biggest markets. It just didn't turn out to be the case. And moreover, we're going to see these things sell in. And, and it's actually, it actually, in my opinion, is perfect. It's, you know, giving three and four year old technology to China is exactly the way I would, I would play this, by the way, that may be my opinion, it may not be, but don't really decide my opinion till after the flip.

‍

Patrick Moorhead: Now it's a great breakdown, Dan, and just to, just to fill in some of the gaps here. Well, first of all, you were a lot more bullish this was going to happen than me because in D.C. right now, right, you got the China hawks that truly believe like we should be sending nothing of importance to China. And then you've got the capitalists who are like, let's drill baby, drill. And then you've got kind of the nuanced folks in the center which is like, it's complicated. And the reality is this, this is complicated. But here are some of the facts. The age 20 fit perfectly into the BIS guidelines which you know, I wrote a blog about around or a research note, God, months ago on exactly what it meant because there was a rumor of, of these chips. So it literally hit the green spot, which means you don't need a license. But we were holding it up, Daniel, like you said, for negotiable terms. And guess what? China's like, well, okay, we're not gonna, we're not gonna give you rare earth metals. And then we're like, okay, we're not gonna give you EDA tools to develop any chips, right? How do you like this? So everybody flexes their muscles. And then cooler heads did prevail here. And yeah, there's some questions about the revenue number. Here's some other fun facts for you: there are enough Huawei Ascends both versions to provide to both the Chinese military and then the Chinese non military industrial complex out there. It's just that people want Nvidia, right? And if you look at the stacks of not only stacks of the software, but also the integration of the hardware, there's a lot of work if you want to shift as we've seen with the hyperscalers, they have built the buttress to use AMD or Nvidia and in some cases even intel, but they're still buying 95% Nvidia, right? So it just shows that people will. The definition of the best tech is complicated.

‍

Daniel Newman: Let's be very clear though, like the AMD chips and memory and their ability on inference actually in some cases we've seen outperform some of the leading Nvidia chips just purely on metrics. And even Nvidia, which has certain performance advantages, is still incredibly hard to convince these companies and enterprises and clouds to buy it in volume. It's been very hard, you know. Yeah, Huawei is not, it's not as efficient. It doesn't have the same capabilities. Yes, you can matrix together three or four times the number of chips to get the same performance. Their software doesn't work as well as Cuda does. Developers don't want to touch it. Even the Chinese developers prefer Nvidia. So I'm saying that if they can't get AMD which has a compelling platform and a compelling stack not at Nvidia level yet, Huawei is still pretty far behind. I think that was probably one of those and DeepSeek were probably two of the worst takes I've ever seen. And the fact that some people still continue to make the take is probably even worse to me is that they've like literally internalized this thing and they're just continuing to, to chew the cud and spit it all over people. It's terrible.

‍

Patrick Moorhead: I love that I'm trying to envision that right now. It's kind of getting grossed out. By the way, kudos to Huawei for a great name cloud matrix because that's exactly what it is. Even though it's 1/3 the performance and 3x the power and heat of Nvidia. What they've done is done a point to point optical solution that connects these XPUs.

‍

Daniel Newman: It's a better name than the Cuda lithography library. The Kulito from, from Nvidia. That was one of my favorites.

‍

Patrick Moorhead: Yeah, they're worth a trillion dollars though. So they get to do stuff like you know, hey, let's move to the next topic. It's a related topic, right? Everybody was fat, dumb and happy. I think Nvidia went up 4 or 5%. Nvidia AMD went up a bit as well. Intel didn't by the way. But then there's this, oh this UAE deal, right. National security concerns hit. I saw this thing coming a mile away and you know the big, you know the big article in the Wall Street Journal actually came out after I wrote my piece on Forbes where I talked about the Middle East and what it means. But it's very simple, right? Should we be shipping the Highest performance Nvidia AMD and Intel Silicon to the Middle East for UAE, Saudi Arabia with folks by the way, the US based CSPs like Google, AWS, Azure and Oracle, but also the Dells, the HPE’S, Cisco's and Lenovo's, should they be allowed to have leading edge? There is kind of an irony right now, not a lot of people know is actually the US CSP service Chinese customers from the United States. Fun fact people, a lot of people don't know that right now. But yeah, this is the debate on the table and I think you have the hardcore hawks going just harder than ever and their fear, which I do understand the fear, I do have empathy is that the military could create a front, a leading edge frontier model using these services inside of the UAE and Saudi Arabia.

‍

Daniel Newman: Yeah, I mean this sort of has some of the same odor of the China stuff. I mean look, the relationship between the US and the Middle East is complicated. Now. We've seen increased diplomacy with the, you know, parts of the Middle East like UAE, Saudi Arabia, they've been compliant and supportive in some ways of some of the ongoing war between Israel and the proxies. They actually enabled some of that airspace to be utilized to help Israel's defense. And we of course know the US Israel relationship is very tight. This is also part of the world and you and I talked about this ad nauseam. But that is very interesting in diversifying itself away from just oil. It seems that the future is going to be AI. It's making these big investments. I, I mean I could see some holdups on moral high ground from some, you know, congressional leaders and, and that this could be delayed for a period of time. But I don't see this actually sticking, Pat. I think that ultimately this gets done, these dollars flow, the investment gets made. Better to build it right and build it now and, and know to your point, and you know, your article you wrote on Forbes about China and KYC is I think that there's going to be, there is and should be some similarities here because we've got a history over time by the way of building positive diplomatic relations with parts of the Middle East that only then turn into our enemies over time. So you know, like we, we, we. I think there is a real opportunity for us to get past that. But that is how, you know, again, you often say read history. This in its own way is a certain version of arming parts of the Middle East. It's just arming in the AI era as opposed to sending guns and nukes or sending technology that will allow them to move their own development faster.

Patrick Moorhead: Yeah, Dan, you never know. I mean, we were funding Osama bin Laden to fight the Soviets in Afghanistan and that, that blew up in our face. We were, we armed Pakistan as well. And I ironically learned that last night to fight the Soviets in Afghanistan as well. So yeah, you just never, you just never know who your next friend is going to be. So hey, let's dive into the next topic. And, and this is kind of an amalgamation. A lot of stuff going on in, in, in AI. You know, DOD is investing $200 million contracts with AI companies. You know, Grok, Grok in the Pentagon. Oh my gosh. A lot of stuff is going on with Xai too.

‍

Daniel Newman: Yeah, I mean this, these are 200 million or so dollar awards. I think it was anthropic. OpenAI Xai, Google. I know this got a headline Pat, and I'm not trying to poo poo our own docket, but like 200 million in our government is like a rounding error. You know, you turn over a rando walking down the DOD hallway and like 120 million will fall out of their pocket. 

‍

Patrick Moorhead:It's like a POC. 

‍

Daniel Newman: Yeah. I mean, look, it is absolutely in the best interest that the government has access to these systems, that there is a hardened, partitioned, secure, you know, sort of approach to being able to make these tools accessible. And of course, the technology and the research that's being done that is able to optimize a business or that's able to optimize a consumer search journey or the creation of content or video or media certainly could be beneficial to our, our government agencies. And by the way, this isn't, you know, this could be GSA, this could be defense, this could be our treasury. I mean, if you think about what some of the things that we're trying to do with DOGE and some of the things that they're trying to do with finding efficiencies and some of the things that they're trying to do to improve the way our government is run and things are trying to do to find the way our files are kept, AI and agentic tools have a massive potential to help us accomplish this. So putting 200 million, making an investment, opening up some doors, partnering up with the leaders in this particular space, gaining access to research tools, library and, and even just knowledge is all invaluable to the U.S. government. And then of course, my take is that they are going to probably accelerate and increase investment substantially with some that they find are able to create the most value and others that they don't. Maybe they will fade into the black. And of course, I think there will be other players that will be continually brought in. I mean, the government's constantly funding things, whether it's something like this or it's DARPA that is always about making sure that newest technology is getting oxygen so that if the government can benefit from it, it is able to do so. So I'm not gonna say it's nothing, but in some ways it's like, I expect this. I think 200 million is, you know, very inconsequential. We've gotten to the point where these large numbers we just hear, we just blow them off. Oh, it's a trillion. Oh, well, not much. Just. Just a trillion dollars. That's all.

‍

Patrick Moorhead: Yeah. Dan, you nailed it on this. All right. I was going to say the same thing. This is a big nothing burger here. And I think it's really just a. Was to get this out ahead of the event that Trump is headlining next week in D.C. on AI winning AI in the United States. And, you know, we saw an event in Pennsylvania, right, which is one of the contested states politically, where there were gigantic investments made there in energy, with AI, with data centers, pretty much the AI ecosystem in there. So, yeah, I think this is politics and I think. I think you. You nailed it. You have any thoughts about the Pentagon looking into. Using. Is using Grok as part of that contract? Is that, you know, after Mecca Hitler went off and, you know, went crazy or. Or is that just completely overblown?

‍

Daniel Newman: It was a great story. I mean, look, I think, you know, security, I think prop injection risks, I think, you know, the ability for these things to go off the rails is all part of their sort of innovation. What did you call the diffusion of innovation? The curve. And so the more you are trying to push the limits, the more likely you are going to be at risk. You know, I don't think it's going to. I don't think it's a perma stop. I do think it might be one of those things that in those final negotiations, they're asking a question like, hey, you guys think you could stop these things from continuously turning into historical figures that provide a, you know, great discomfort to a segment of the population. But, you know, we saw it happen to Microsoft years ago. You know, we saw the absolute trash that was Google's first iteration of this stuff could have given up then I mean, Grok 4 works pretty well. It just has this weakness there. Apparently also has deep seated, racist tendencies that are created by the way by the content on the web. Because we have a very polarized world and a whole lot of very angry people out there feeding data to these systems.

‍

Patrick Moorhead: Yeah, good fill in there, Dan. So what I do good, dude, look at you keeping up with everything while on, kind of vacation. So it's great to see you keeping up. Keep it up, baby.

‍

Daniel Newman: Doom scrolling every night.

‍

Patrick Moorhead: Yeah, doom scrolling's great. My stress actually goes down. You know, they say it stresses people out. Like I get in the teens on the stress meter.

‍

Daniel Newman: Other people's misery. You haven't been to any Coldplay concerts, have you, Pat?

‍

Patrick Moorhead: You know, I haven't. I don't go to concerts. They typically start after my bedtime or at least the headliners get on. And even at the events that I attend, I don't go because I'm typically eating dinner with you somewhere in the Palazzo, in our hidden restaurant we meet at.

‍

Daniel Newman: Well, I just gotta say, if the kiss cam ever came our way. You and I were in that type of embrace. It's nothing, you know, we're just gonna.

‍

Patrick Moorhead: No, it is, it is. I mean, I would definitely hug. We would definitely hug because I don't think we have anything to hide. I mean, my gosh, we go to events, you know, we work out together, we eat breakfast together, lunch, dinner together. Maybe we do some other things together. But it's, it's good, good fun, good.

‍

Daniel Newman: Friends elaborate anymore though, right?

‍

Patrick Moorhead: No, no, no, I'm not going. I stopped. I stop. I stopped myself.

‍

Daniel Newman: I appreciate that. Yeah.

‍

Patrick Moorhead: Important stuff. Hey, so there was a big event this week in New York City. Our team had three analysts there, I think. Dan, you, you had some analysts there as well. It was AWS's New York summit keynote. And it's interesting if you look at the history of events with re invent. It used to be that they had one event a year, right? It was AWS re invent November, December. We've been to a ton. You've, I'm sure, read about a lot of these out here. And then they went twice a year which was kind of, you know, it was to reinforce what, what they're doing. And then Gen AI hit and it's like pandemonium across all vendors. And this is kind of this mid-year kicker that AWS does. And absolutely no surprise, it was all about generative AI. But it was specifically I would say about agents. That was the headline now prior to the event and I would take this as hey, this was their most important announcement was something called a Kiro and it's an IDE integrated development environment tool to build, test and deploy AI agents and by finding new ways to link code, the natural language type of rollouts. Jason Anderson and my team did a huge write up on this before the event. You can check it out in the show notes. The other big announcement there was Amazon Bedrock Agent Core, right. And it's a seven service infrastructure for actually deploying the AI agents, right? You had KIRO on the front end and you deploy them with Agent Core and it includes for all you nerds out there. Actually my son who's an intern in Nvidia would understand all of this. AI agents including runtime memory, identity, code interpreters, browser tools which I can't wait to use, gateways and observability to really put the whole package together. And finally Updates on Nova 10,000 customers using that. And again it wasn't specific whether that's production or that is POCs. But Nova, which are AWS's homegrown models, are alive and well and I thought this event overall was a good benchmark and a good milestone for AWS who has appeared to be playing catch up in generative AI since it came out. That event you and I were at two and a half years ago with OpenAI and Microsoft. The reality is that most machine learning workloads for enterprises are on AWS and it's a pretty natural if you have the tech to move that to gen AI. And I am very impressed with the progress the company has made, particularly when they're the only company who has agents and gen AI that are only for businesses. Microsoft has a consumer arm and I do recognize, Yeah, I know Amazon.com is consumer, but AWS primarily builds for other companies. Google obviously 85% consumer in there and 15% enterprise. But good progress. AWS.

‍

Daniel Newman: Yeah, I think there's been a lot of catch up that needs to be done and I think some of that catch up is still being done. The largest infrastructure of Cloudera 1 that pivots from Cloudera 1 to Cloudera 2 has been somewhat in question and you look at it by the volume of AI consumption, the volume of infrastructure, the volume of the build out of AI infrastructure on a global basis. They've been ahead for a long time and they've been early to infrastructure development, Trainium, you know, Graviton and what they've been able to do there. But there's some definite work to be done for them. But it seems like they are making progress and I think their big key is to keep the customers they have and get them to pivot. Do not let customers defect, do not let customers. You know, I think Google and Oracle are a bigger threat right now to AWS than actually Azure. You know, Azure has its own challenges. You know they were, they're a little overdependent on OpenAI for a lot of their AI features. So I'm interested in hearing more from them. We'll talk about Azure and Microsoft I'm sure as earnings come around the corner here. But you also saw Pat, Open AI went all in on, on basically doing a co pilot offering this week. I know it's not necessarily directly on the docket but God, that relationship is way more strained than I think people realize. Like, and by the way it is way more strained. But just know that, as Donald Trump likes to tell Vladimir Zelinsky, I have all the cards, we have all the cards here. Like Satya had all the cards here because you know they basically right now Sam is trying everything he can but he literally is being suffocated by Microsoft. So sometimes you gotta be careful who you choose to cuddle up with at the Coldplay concert early on in your, in your relationship building era.

‍

Patrick Moorhead: So yeah, forget about Coldplay sucks. You know, it's dangerous to go there. Yeah, let's finish up this decode, Synopsys. Got the Anysys deal. We tried to schedule some time with Sassine. He's globetrotting over in Asia right now. But yeah, they got it across the line, didn't they?

‍

Daniel Newman: Yeah, I talked to him a couple times. I'm just kidding. I talked to him once. It was a big win for them. I mean unlike even like the VMware deal which was kind of a non essential, this was very much essential tech that was totally held up in the middle of this trade war. You know, as you said earlier astutely Pat in the comments about the Nvidia stuff like leaving access to EDA and the most advanced EDA tools which includes you know, Synopsys, Cadence, Siemens, those were the big three here in the US is critical for China to be able to develop next generation chips. Also Synopsys is the number two player in IP and a lot of people don't realize that, but they are. And of course Ansys is number one, you know, in terms of the simulation engineering stuff. So what they basically brought together, this is what I really love about the deal is of course it makes the companies bigger. It was super accretive. It was leading in systems, leading in silicon. Now you're basically got the whole stack in terms of development Pat. But I love this for the physical AI era when you actually start to embed AI inside of the edge where its robots and autonomous vehicles and connected devices intelligence, this becomes really smart because now you've got silicon to system. You got the ability for basically them to be the leader, the choke point, the dependent upon. In some ways this makes them a little bit the TSMC of the engineer system design era as to what they are for the foundry fabrication of chips. So I like the deal. I think it's going to expand their TAM more than most people recognize. I think some people think it's about 1.5 times more. I think it's quite a bit bigger. I think if we are actually starting to be able to wrap our hands around heads around how big the humanoid physical AI TAM is going to be. This really positions coming well but also a lot of credit to Sassine Ghazi, their CEO who literally hand walked this deal through, did not relent because there was a very real possibility that this deal would get killed and it would have really hurt both companies, but especially Synopsys who had lost access to that market for a period of time and who had obviously made this one of their biggest sort of rocks in terms of moving the investors forward and, and building expanded TAM for the company.

‍

Patrick Moorhead: Yeah, great commentary there Dan. You know, I don't always get it, I don't always get a deal immediately. I'm a thinker. But this one was evident on day one. It literally took me 15 seconds to figure this out back in early 2024, why this was smart. And then they did an investor conference back in March that I think both of us had attended where they finally started showing some slideware. And you know, I'm the next product person, I tell everybody that and I look at it from, a core value proposition. And the core value proposition is that it enables a systems level soup to nuts design from chips to the actual infrastructure. And whether that, that infrastructure is a car, whether it's a rack in a, in a server, you know, there's nothing like Nvidia, to see what levels of integration, any systems level problem that Nvidia is solving. And for customers like AWS and Tesla, this new combined company is going to be offering incredible value. Now the key here in all of this, right, because we do go through periods of aggregation and then areas of specialization is that they can't be slow when they do this, right? When Apple went fully integrated, they were very fast. Now with AI, they're very slow. And I don't know if that's because they feel like they have to do everything and all the other pieces, the interconnects are slowing down. But when I do talk to Sassine, I will make the commentary that you need to show speed, right? You can't have slowdowns between going from electronics to the electromechanical part of the business. So congratulations. It took a long time and oh, by the way, this was an absolute China negotiation. Okay, we'll give you access to Synopsys, but you need to okay, this acquisition, China, they all happen the same in the same relative time. And here we are. All right, folks, let's move to the portion of the show where I usually win, where we actually take. We put something very provocative out there. And regardless of whether Dan and I actually believe in what we're arguing, it's all about the sake of the argument and the hope is that you can get a deeper understanding of both sides of the topic.

‍

Daniel Newman: So I think we, I think we have to maybe start polling the audience and keeping score to keep this interesting because I know these self declarations of your dominance, you know, found their way across the interwebs. But I just, you know, I tend to not agree with your assessments.

‍

Patrick Moorhead: Do you really want to crush this, this soft little ego of mine? I mean, like.

‍

Daniel Newman: Oh, no, no, no, no. I just thought, you know, maybe we could try to be accurate for once instead of just self aggrandizing. I don't know. I'm just kidding.

‍

Patrick Moorhead: I don't know, man. The all in pod says that it's the number one pod, but it's actually not.

Daniel Newman: To its credit though it's come a long way. It kind of manifested that idea.

‍

Patrick Moorhead: It has, it was right, right after Rogan, by the way, last week.

‍

Daniel Newman: So I'm just saying, like when we're. When it's like Rogan all in, six five, like in that order.

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: You and I are gonna be pretty happy with ourselves.

‍

Patrick Moorhead: We will. You know, until then, I want to get our 100,000 subscriber trophy off of YouTube and give it a kiss, maybe take a drink out of it. Like F1 races do something like that? Exactly. All right guys, let's dive in. Here's the topic for this week. Are the new export restrictions on GPUs and AI silicon enough or should we just open up the floodgates? All right, Daniel, the guy that you know loves the CCP here, Dan, you have to argue. Opening up the floodgates, baby, let it all in.

‍

Daniel Newman: Look, if we had competent leadership in Washington, I might suggest that there'd be some level of controls that should be put into place. But here's the bottom line is these devices, the DeepSeek story was all about, oh, we did three paper clips and one 15 year old amd opteron server and we built DeepSeek. No, that's not what happened. What they did was they had hundreds of millions if not billions of dollars worth of Nvidia chips. And by the way, they weren't all the China Nvidia chips, they were other Nvidia chips in there. It's been said, it's been proven, it's been shown, but not proven enough that anyone's doing anything about it because again, we don't actually do anything. It's all a big negotiation, it's all a big threat. But what it actually does do is it slows down revenue, it slows down transparency. And if we just opened up, we would know exactly where everything was. And if the idea is all about knowing your customer, this is a better version of that because now we know where things go. It's a completely open market. And by the way, it also provides the incentive that everyone agrees upon to keep innovating faster. They use our technology, they become addicted to it, they become a nested need, they need it even more so than ever before. It slows down the advancement of Huawei, it slows down their potential need to build their own lithography. And, while we may be opening up and giving access, we in some ways are also controlling the innovation speed because we've slowed down and turned off the machine of China and their deep desires to eventually eradicate our market leadership by building better and more capable products on their own. But this really does come down to only two things. It comes down to 1 competence and 2 speed. Competence, we just aren't going to manage it. We aren't going to pay attention. We're going to continue to allow these multi ship gray market yes people go, oh, they're giant racks, thousands of pounds. That isn't actually always true. The size of an individual GPU is much smaller and they know how to stick these things into racks and wire them together. So, the fact that people say it has to be a one rack and a shipment is not the case. H100s, H200s. That is not how this works. They're getting into the country anyways. We're not managing it, we aren't going to stop it anytime soon. But at least Nvidia could get the credit and the transparency could be there. And then like I said, it's all about the speed. We don't want to give our people the safety net of protectionism. We want to force them to understand that at any time China could get to the next level development of their Ascend chips, cloud matrix or even have a breakthrough for 5 and 3, 2 and below 2 nanometer 3D packaging assembly. If they get there and they do this, they're off to the races. If we don't open up the gates and give them everything they want, we're missing out on revenue and we're missing the push that we need to make sure we lead the world's economy and we lead the world's technology development. So we can keep pretending that we're protecting ourselves, but all we're doing is hurting free markets and hurting innovation. Do better.

‍

Patrick Moorhead: Dan, that was really good. Bestie on your vacation too.

‍

Daniel Newman: Yeah.

‍

Patrick Moorhead: So I am going to debate that we do need some restrictions. You know, restrictions are necessary and then, you know, you also have to debate, you know, can we actually I felt like you did a great job debating can we ever limit this. Right. You talked about the size of the unit of a GPU. But I think we do need some restrictions. I think selling as many H20s or B whatever B20s  is a fine thing. Send as many of those as you can to China. But when it comes to the higher level Reuben types of chips, I just think that we need to have some sort of restriction in there. I'm going to call them intelligent restrictions which are knowing your customer. So cloud computing is about 15 years old and we've gotten really smart at how to observe what is going on in the infrastructure. It's called observability. And we can see what is being done from a workload basis. And you can pretty much bet that if we have, if you see 50,000 GPUs being used at once somebody is training a frontier model and if that frontier model can't be distinctly, you know, brought into, identified as I'll call it the Uber Eats of China, which I think is, is okay. Then I think it should be allowed to sail through. We've got DNS, which is the lowest level of tracking out there. That should also show where are these, are these bits going? And then I think maybe a final layer. And my gosh, you saw this with center centrifuges, which is we should be able to truly track even if somebody's doing some DNS spoofing on where this is going. And here's a common sense question I have for everybody out there. Do we really think that the Chinese military is going to be creating models using Nvidia GPUs leading edge GPUs out of Saudi Arabia or out of the United States? That is an absolute fantasy. Right? The PLA is going to be doing all of its training and is doing all of its training and has been doing all of its training with non Nvidia chips. And yeah, I get it. One person's weather simulation is another person's nuclear simulation, but I think we're beyond that. And by the way, none of this is new. When I was at AMD, I was restricted from selling certain level flops, right? We've gone from flops to tops. Trademark Daniel Newman. So none of this is new. We've had restrictions, my gosh, since what, 2002, 2003. I remember having to restrict systems going to Iran as well when I was at AMD. So with intelligent restrictions using modern technology, I believe the risk is very small. One thing though, I want to know is I want to see public declarations of KYC process from companies like G42, and I want companies like G42 to have some skin in the game. If we find that some of those tops have gone to areas that they shouldn't be and there should be penalties. I have full confidence in the AWS’s , Google’s, in a cloud and, you know, and even Oracle cloud in an environment like this. But I'm unfamiliar with G42. They're an unknown entity in my world and I'd like to see a little bit more disclosure and openness from them. Hey, maybe have somebody from G42 on the podcast to, to talk about this.

‍

Daniel Newman: We could do that. We could do that. Maybe someday.

‍

Daniel Newman: By the way,  just, you know, since we do this at the end and hopefully they won't share this part. This is only for the people that listen to the whole pod. But like, I know I was great and I won the debate, but like, I actually think it's a terrible idea to give access to all the technology to China. I think that what we're doing is, you know, protectionism absolutely extends our opportunity to lead the world and gives us more and more time. And the problems that we're keeping them, like things like ASML and Leading edge process and access to our best GPUs is absolutely hamstringing their ability to compete. So I, you know, I just,  I, my own argument was so good, but it's just a terrible idea. So don't do it. Like, don't do better, do worse. Keep it up. Thanks.

‍

Patrick Moorhead: No, thanks Dan for that.

‍

Daniel Newman: It's important I got that out there.

‍

Patrick Moorhead: Yeah. And if you see the disclaimer in the lower left hand corner. Yeah, Dan's a good sport. He's fighting for something he doesn't believe in. I do that sometimes as well. So thank you for participating. Let's move into the final segment. We've got to rip through this, get you back on the beach with the family. This is Bulls and Bears where we dive into the latest earnings reports, trends, the movers, the shakers, the Fed chairs getting fired, etc. All right, the taco truck is here. Tariffs on, tariffs off. Are they real? Are they imaginary? And this week there were a lot of discussions about 30% tariffs on the EU and Mexico. Even though we already have a trade agreement with Mexico.

‍

Daniel Newman: Well, you know, some of it, we have a, it's, you know, it's not as complete.27 billion dollar surplus last month, you know what paid that surplus in our, in our budget tariffs, you know, how much our inflation was up. We'll talk about that, but not a whole lot. You know, economists, I don't know what these folks do, but what they do consistently is they're wrong. That's what I've noticed. They're consistently wrong. You know, don't ever listen to University of Michigan, don't ever listen to Atlanta Fed. I mean, these people, like, you know, they've blown it so badly. Here's the thing is like, you'd have a, you'd have an iota of credibility left if you just acknowledge the fact that this stuff is really hard to predict. And you know, and took your political bend out of it, everyone did. You know, I wrote, you know, kind of some things I thought would happen and a few things I got really right and a few things I got really, really wrong about what tariffs would do and what Doge would do. I mean, Doge was a great idea and it is, I don't think much at all. I think maybe 9 billion or so of the savings will actually end up sticking after, you know, that's it's all about the filibuster. We can't really do anything without 60 votes in the Senate to cut spending. But the only thing I know for sure is our growth is going to come from A GDP expansion and B prices will continue to rise. There is just no situation in which we will be able to cut enough that we're going to get back and we can't tax enough, so we have to grow enough. That is the way out of this whole thing. And we are the best economy in the world. You can argue this, but it's. There really isn't much of an argument about it. We are the best consumers in the world. And so if you're building anything anywhere else in the world, you want access to US Markets because even though people here often complain that they don't have a high enough standard of living, we have among the best standards of living in the world and of course the most people with money to spend on things, whether it's cars, whether it's clothes, whether it's jewelry, whether it's electronics. We are the place to do business. So all those things put together and you'd say that the thesis of the administration might have been right, is that one, a lot of people like Ferrari originally came out, said they were going to charge 25% tariff on their cars. And what they're ending up doing is I think something like 10% and they're going to eat the rest. And I think that's a lot of what ended up happening out there. People are like, yeah, we're going to raise prices a little bit on things, but we also realize that we really want access to the United States. Mexico needs access to the U.S. The EU needs access to the U.S. and, and by the way, there is still a lot of balancing done. The one question mark and I, I even got to acknowledge that I don't know is that we don't, we've been all hyper focused on goods, physical goods we do export our services everywhere. And I know that Canada tried to play a game with us and I think the EU also did a bit of that where they might start taxing our tech companies. But I think if they really try to pull this off, it's going to be an absolute freaking war. And I think the thing is that those parts of the world don't have anything because they've basically allowed us to be the builder of all of this technology forever. So what are they going to do? No Google No Amazon, you know, no Apple. Like all these things that people Netflix have become dependent on, that we've created. It's like you're going to tax more and pass that cost on to your customer and your citizens? I don't think so. So I think in the end the tariffs are basically becoming a nothing burger. And I think that we're going to absorb them and I actually think they're doing the job. I mean, we'll see over time. I'm not all about them, but in terms of fixing our revenue problem, it's doing a heck of a lot more than anything else I've seen done over the last two decades.

‍

Patrick Moorhead: Yeah, the administration, I like the overall what they're trying to do, which is let's get the best deal that we can, that, we deserve. I mean, we are so much in debt as a country and we use that debt to buy all this stuff and we buy all this stuff overseas, which is enriching other countries. So I think that we need to get more people outside of the US buying our products when they are good, not when they are crap or low quality. I remember the complaints that the US auto manufacturers had back in the 70s, but the Japanese cars were twice the quality. I'm glad we didn't dig too deep in there. What the Japanese companies ended up doing is building a ton of manufacturing plants over here in the U.S. so we wound up in a pretty darn good place there. Let's move to the next one. Right. You talked about, Dan, tariffs, you know, not making an impact. Well, what does that mean to inflation? So let me get the raw figures out there. So two things, right. PPI is wholesale. It's what manufacturers, the cost that they're experiencing. And then you have the PPI which is what are consumers paying for goods? So PPI or wholesale was unchanged. Though I find this fascinating because if we're paying more, if companies who are manufacturing stuff are paying more for goods overseas through tariffs, you would expect this to be a lot, a lot higher. There were some puts and takes. Services were actually down, which, which, which I find fascinating. The prices were up a bit by 0.3, 0.3%. So core PPI, which was excluding food, energy and, and services was flat from flat from a year ago. Anyways, I find that fascinating given that I thought all of these tariffs would, you know. So Daniel, your, your point on people eating it, final goods and not extending it. Now I'm questioning whether their costs even went up or is this based on inventory that they had otherwise PPI would be up anyways. CPI rose 0.3% after a small increase, in May, CPI did increase up to, you know, 2.7% again. This is growth up from 2.4% the previous year. And this is the highest pace since February. Where did it come from? Medical care. I don't think inflation has anything. Sorry, I don't think, I don't know why those would be going up other than we're getting less healthy household furnishings. That's interesting. A lot of that's imported. But wait a second, we didn't see a PPI increase. Why is going up motor vehicle insurance? That sounds like people having too many claims and wanting to raise prices. But all in all right, it was going to be a disaster and prices were going to go up and to the right. Remember the Biden economy days build back better. You know, they wanted $3 trillion. Inflation was at 7%. Those were some ugly freaking days. Don't even get me started. Back in the 70s we were at 13% with Jimmy Carter. There was an oil embargo, there were gas lines. I feel like we're in a pretty good place right now on inflation.

‍

Daniel Newman: Yeah, I, I would say that there was no violent move, which I think is what people were looking for. Now the one question is we're going to hit this next wave of earnings. We're starting this week. We had the banks, a couple of the techniques, TSMC, we'll talk about that, you know, has all this inflation started to work its way in, you know, last quarter? The excuse is it's coming, it's just going to take more time because there was existing inventory to sell off and you know, these companies didn't want to shock the market, you know, and I've heard it in pockets in certain industries. I mean in the tech industry, I don't think it's had a super large impact, you know, and also that kind of taco trade, there were a lot of adjustments made to tariffs that allowed some of the things that are really important to keep parts of the economy moving. We do know that the Trump administration does value markets and markets ripping. He comments on this regularly. And overall though, I think inflation is moving a little bit, but it's mostly in line. And by the way, it's way higher than they're saying it is now and it was way higher before, meaning like when they were showing it at 2 and a half or 3% in the binary, I feel like it was like 30%. I've never. I just absolutely don't think that the readouts are worthy of that. We've been getting for the longest amount of time. And we saw it in the grocery store, we saw it at the gas pump, we saw it on insurance we were buying. We saw it on the cost of hiring people. The costs were way higher. This has been going on for a long time. If anything, they've come down over the last, you know, six to 12 months, especially under the new administration.

Patrick Moorhead: Good stuff, Dan. You're always on top of this. You and future equities and your AI15. I think it's brilliant. I love it. But let's go into this could all come just tumbling down if, you know, we don't have an autonomous Fed. And is Trump going to fire Powell?

‍

Daniel Newman: I don't think Trump can fire Powell. I think the problem is. This is heavily debated, by the way, can he even make the decision to fire? I think it will. He certainly can't. Now we're hearing that there's criminal inquiry type of things being passed on to the DOJ because of his two and a half billion dollar retrofit of, the Fed office or the Fed headquarters, which is like the Taj Mahal, I guess it should be, if this is like the epicenter of, of capitalism. At the same time, you know, Trump's points about interest, you know, you want to stimulate the economy. If we want to be cool, growth is our focus. You need to make access to credit markets. You need to enable people to buy homes. You need people to get equity in their homes. Lower rates allow asset appreciation, they allow stock appreciation. So the fueling of the bubble that we want to create, that we need to, you know, to grow faster than inflation, needs to be driven by somewhat lower rates because frankly, we've gotten used to rates and assets have not come down. Even through this period of tightening, homes are still very expensive. You know, you kind of talk to people past someone your age. What'd you pay for your first home? What did I pay for my first home? You look at these kids now coming out of school with $200,000 of, of student loan debt, and then you say a starter home is $500,000. Wages have not increased enough. So we are in this really interesting juxtaposition where the only way we can basically get people there is by inflating the crap out of assets, but doing it at a rate that's just a little bit slower than how fast we grow the economy that's the play that requires lower interest. Powell is incredibly hawkish about rates right now. He doesn't want to cut rates. He continues to use the tariffs as sort of the reason he doesn't want to cut rates. But I think anyone that's watching from the outside can see there is clearly a political conflict here. It's basically Powell and Trump are like Satya and Sam are to us in our industry right now. I'm gonna pause there because I think. I think I hit it on the head, but I don't think he can fire him. I think it's going to be an act of Congress. I don't know if they're going to get there, but I think the threat is something that. It's a. It's a card Trump is playing to try to stir up the markets and, and create enthusiasm about his willingness to do anything and everything to get growth.

‍

Patrick Moorhead: Yeah, The Federal Reserve act of 1913 says that a chair can only be removed for which is, you know, misconduct, malfeasance, maybe something to do with renovations to buildings. So I hope we're not getting into this political BS. I can't stand it. I don't like it. I think it's nasty, but I don't know what the Fed has to see in inflation. I think it's the fear of tariffs. That's why I think that they're waiting is this fear that they lower rates and some super tariffs kick in and prices go up. So they don't want to let that cat out of the bag. I mean, my God, we, we saw what we did with spending and how that drove inflation, and it ended up being a bit of a disaster. All right, let's go to the last topic here. Dan, you love TSMC. Talking about TSMC. I'm kind of bored of TSMC because the news has been the same for the first, second, you know,, seven quarters. And it is basic. Basically mimics everything Nvidia and all the hyperscalers are. Are already saying there's literally no new news.

‍

Daniel Newman: All right, well, let's be very, very clear. One is love and the importance of being able to help people understand the breadcrumbs are not the same. I do not love anything except for you, Pat. But in all seriousness. Look, TSMC reports early in the quarter, not late. They report their revenue every month. So they give more data. It's different because they trade in a different exchange and they have different requirements for reporting their, you know, profit surge on their absolute monopolistic pricing power that they have right now over every single wafer and every, you know, in all of their capacity. And right now they're able to basically name their price, pick their partners and do anything they want. Makes them an incredible company. Is that sustainable? Well, as long as Intel continues to not be able to bring outside customers in at scale or win outside customers. So maybe at 14a we'll have an inflection, but right now it's just the. You know, Pat, you and I get people at times that'd be like, oh, AI is a bubble, it's going to burst. It's like, okay, but if you look at the data point right here, it's probably not going to burst at the infrastructure level anytime soon. Like maybe there's going to be a point where we've built all these ghost cities and nobody's using all this stuff. But the building of the ghost cities are going to continue for some time ahead. TSMC continues to give us proof points by the strength of the revenue. But it's all in AI. Let's just be very clear. This is a have and have not moment. It is not legacy chips, it is not cloud CPU era chips. I mean yes, they do those, but where all the growth, where all the profit is coming, it's all related to the AI boom. So when I see good results from TSMC, it just sort of sets the tone for me of what we're going to see from Nvidia, what we're going to see from AMD, what we're going to hear from the hyperscalers about cloud, the Capex affirmations, the continued investments, the strength of companies like Avago, Broadcom, Micron. The breadcrumbs here are so rich, so delicious that you could actually make a steak out of them.

‍

Patrick Moorhead: Dan, what a great way to close this show. You just did a slam dunk on that. I get bored a lot easier. I'm not making any money in these trades. I know you're not either, particularly with, with Nvidia. But man, this frothy, frothy market always makes me nervous and it made me nervous back in the 2000s when I saw it basically crashing my face. You know, my company valuation went from 3 billion to a hundred million. And you know, my spreadsheet, I think I made $12 when the crash hit with Alta Vista and it really sucked. It kind of shook everybody. There were so many day traders, Dan, back in the day trade, you literally, you throw, throw a line in and you're making money. And everybody was an expert. People quit their jobs and day trade.

‍

Daniel Newman: I had a meme somewhere, that a monkey was able to outperform your average day trader. Just blind monkey. You know. I don't, one thing about me is I love markets, but I don't really trade. I am a guy that buys companies, falls in love with them, holds their stock and rides them all the way up and then takes the same trip all the way back down when the market crashes. So, but over time it's still like one of those where you kind of look at the escalator, right, where it kind of does this and then it goes back and then it goes a little higher and it goes back and it goes a little higher and then goes back. But overall, directionally that's still a heck of a lot better than not being in the market. So again, not investment advice. We're not telling you what to buy or not buy. But I'm saying there are a heck of a lot of breadcrumbs with TSMC. People are going to realize this AI thing is the real deal. And yes, there will be some blow up, there will be some, there will be some companies that were completely nonsensical. But the ones we're talking about, there's no nonsense in terms of TSMC's moat. That thing is legit. Like they are, they are going to be okay even if it blows up.

‍

Patrick Moorhead: Totally. The only thing that derails them is China invading. So you can imagine, even if naturally the volumes come down. When you own a foundry, you have to sell what you make and you keep lowering costs because you have to cover your fixed costs. Your capex, you built this thing so.

‍

Daniel Newman: Sure do baby.

‍

Patrick Moorhead: Even if you sell it at 0% gross margin, right. You've got to do it to cover your fixed, fixed cost. Trust me, I was there when AMD had a head foundry. Heck, our fab wanted our 180 nanometer fab one fab of the year when I was literally the best foundry in the entire world. By the, by the sia. Boy, things change. Hey folks, I want to thank you for tuning in. This is a great show. Dan, thanks for coming in off the surfboard, out of the weight room, out of, you know, hanging out with your family to do this. You're a better man than I. I usually don't like to do these things when I'm on vacation, but I really, really appreciate you and I really appreciate the fans out there. If you're not subscribed, consider hitting that subscribe button. We have a hundred thousand subscribers now that I'm really proud of and it's a pretty awesome community joining.

‍

Daniel Newman: I heard it's a hundred thousand, three hundred now or something. I'm kidding.

‍

Patrick Moorhead: Oh yeah, yeah, possibly. That's cool. That's pretty cool. Thanks everybody. We'll see you next week.

‍