Loading summary
Host
Nvidia is set to start shipping chips back to China again after months of this thing being in limbo. And really, for the last few years, Nvidia has been trying to figure out all the regulatory restrictions that the United States is putting on it. So we're going to get into the background of where it's been in the past, why they're able to start shipping this again, what changed, and also how many billions of dollars this is going to add to Nvidia's overall revenue. This is really crazy numbers. And the reason why they're back on track is also quite a crazy story. So we're going to be diving into all of that on the podcast today. But before we did, I wanted to mention if you want to try all of the top AI models without having to get subscriptions to every single platform that exists out there, I recommend checking out AI Box AI. This is my own platform that I'm building. We're currently in beta. And you can try the top 40 AI models, including a model from Nvidia called Numitron 70B that they put out. There's a ton of other models from all the top companies like Anthropic, OpenAI, Meta, Google that you'd expect as well, a bunch of interesting ones for image generation from Black forest labs, ideogram, we have text to speech from OpenAI and 11 labs and tons of other models. So if you want to try out all of that for 20amonth, you don't need subscriptions to every platform. Go check out AI Box AI and lots more exciting stuff launching from my startup in the future. All right, let's get into what's going on with Nvidia. So this is going to be Monday that Nvidia made this big announcement, and essentially they're able to start with, they're able to restart the sale of their H20AI chips to China. So this is something that they were formally selling, essentially, and then the Trump administration put a bunch of restrictions on. They eventually change those restrictions and they're able to start selling again. So the company essentially is expecting to receive all of the licenses from the government soon that they need to ship these out. Nvidia is also creating a new RTX Pro chip that's designed specifically for the Chinese market. So they're saying, you know, this is fully compliant with all the regulations. Basically the regulations came out, and a lot of this also came out, I think, under Biden with the CHIPS act and other things. But essentially the government said we don't want China to, you Know, compete with us, with our adversary and be, you know, number one in AI. We want to give American companies a leg up. So they essentially just restricted what Nvidia was allowed to ship to China. They weren't allowed to ship the most powerful chips. Now did this stop China? Like, no. We see companies like Deep Seek and other big companies coming out of China with pretty powerful AI. And a lot of this is due to the fact that companies would just register, would just purchase chips in Singapore, which doesn't have any restrictions. They ship them to China and just kind of make the profit. I think one of the big things that shocked a lot of people was the fact that 30% of Nvidia's revenue came from Singapore. It's like, oh wow, Singapore accounts for like almost as much as any other country in the entire world. But really Singapore is just kind of a fill in for China. So there's a bunch of restrictions, there's a bunch of loopholes. Nvidia denies any wrongdoing in this situation. But regardless, it is a, is a tricky situation. They're trying to, they're trying to work with the US Government and get around this. So these new chips are supposed to be, supposed to be, you know, fully compliant and able to ship to China. So these are the H20 chips. Of course, there's a huge kind of fight between the US and China. Most of Nvidia's advanced AI processors are not allowed to be shipped to China. The H20 is the most powerful chip that they can legally sell to China under the current export controls. But even this got blocked for a while. So I think we saw a bunch of big companies in China, including ByteDance, Alibaba, Tencent, all of them have been like super aggressively Stockpiling the, the H20 for the first three months of this year because they were anticipating much stricter export controls. And basically this is because this chip in particular has really good memory bandwidth compared to Chinese alternatives which you could get in China. And also Nvidia has a really widely adopted software ecosystem that makes it much easier to deploy. So if you don't have this, it's quite tricky. And being able to use Nvidia chips speeds up. You know, you're able to not just have the hardware, but also all of Nvidia's software, which is really impressive. So all of the back and forth began in April. The Trump administration restricted even the H20s, right? So H20s were selling like hotcakes to every single Chinese, you know, big tech company. They were stockpiling them April he shut it down. And some people say that this potentially cost Nvidia 15 to $16 billion in revenue just since April. So you can imagine how much they want to get these things back on the market and selling again, judging by how much, you know, Chinese firms are trying to get, we're trying to get them in the, the first quarter of this year. These are very important chips that they still really want. But the thing that I think is interesting is the original move that they just did in April was trying to target chips that were exceeding specific performance thresholds. So that was anything that had total memory bandwidth of 1400 gigabytes per second or an input output bandwidth of 1100 gigabytes per second. So they essentially, you know, it's like, can't be too fast, can't be too powerful. But these restrictions were really short lived. Now a lot of people will point to one specific meeting between Jensen Huang and Trump. Jensen essentially went to a, it was like a million dollars per head at this dinner at Mar? A Lago where Trump was there and Jensen Huang went there. And shortly after the, you know, the administration put a pause on the ban. So some people are kind of pointing to that and pointing to, I don't know, all sorts of people are like, you know, this is a complicated balance between government and industry. I mean, at the end of the day, Jensen Huang has, some people are, I don't know, whatever. Jensen Wang, obviously, in my opinion there's no, you know, it's no surprise would have went to that dinner with this issue, tried to bring it up to Trump and probably had a conversation about it. So whether it makes sense or not was then up to them to, to decide. I think it's kind of interesting. But one other thing that people are pointing out is out of that dinner, not just the chip ban, but kind of getting paused. There was also another outcome that was shortly and shortly announced afterwards because within a week of, you know, this kind of report being published about Nvidia's doing that, Nvidia announced plans to build an AI server in the US which was worth as much as $500 billion over the next four years. And they're also going to be doing this with help from TSMC, Taiwan Semiconductor Company. So two things happened. One, the chip export got paused and two, Nvidia promised a 500 billion dollar chip manufacturing project or I guess an AI server build out in the U.S. so this is pretty interesting. This is pretty interesting for a lot of people. You know, I would assume there was some sort of negotiation there was nothing formal that was kind of published. But you could assume that Nvidia is like, hey, if we want to sell these chips, you it's hurting a company based in America and American jobs. And Trump's like, all right, well, maybe you spend $500 billion then. Well, we'll figure something out. So however the conversation went down and some people are like, no, nothing happened. It was just a coincidence. I personally don't really believe it. I believe there's probably some sort of deal that was struck in the process. But regardless, this is kind of where the chips are on the ground. So some people are criticizing this, especially because we see China's AI capabilities are growing much stronger. We see Deep Seq, who came out, you know, really kicking, took the world by storm with their model when it first came out. And that was built like they essentially Deep Seq was completely trained using Nvidia's H800 chips, which are more powerful than the H20s. But the US banned the sales of those back in October of 2023. Somehow the Chinese suppliers managed to figure out a workaround to continue to get them. So I'm not personally know, shocked by this. And to be honest, like, a lot of this I feel like is sort of theater. You know, I mentioned what's going on with Singapore, but like, Nvidia wants to be able to have a more direct, more open line of. Of sale, probably directly into China that can maybe cut out the middleman and save some money or make more money or sell more. But, like, at the end of the day, China's going to get their hands on advanced chips no matter what. Even if they had to build them themselves, I don't think you could just permanently cripple China's AI ambitions personally, even with export controls. But in the meantime, because it does take some time to stand these companies up, it does take some time. Nvidia does have superior technology. Anyone else? So this definitely is the fastest way to. To, you know, for Nvidia, get the chips there and for the Chinese companies to get these kind of powerful chips. So what's interesting is Jensen Huang has been meeting with officials Both in Washington, D.C. and in Beijing this month, and he's emphasized, quote, the benefits that AI will bring to business and society worldwide. This is a big thing he's talking about. I think this is definitely very interesting as we have US Policymakers attempting to make this sort of balancing act with national security and really powerful commercial interests. Given everything we've seen in 2025, I don't think that this trend is going to slow down at all. I bet we'll see more export restrictions on Nvidia on certain chips, while also perhaps as we get better and better chips, they'll be able to ship, like faster, faster ones that are kind of the older, older model. And maybe that will give American companies a slight edge, even if it's just that it takes, you know, months longer to get chips to China. We're going to be months ahead and I'm sure there's going to be a whole battle that continues throughout the, throughout the rest of the decade, to be honest, in this, in this regard. So in any case, if you learned anything new, if this was an interesting episode to you, what's going on with Nvidia, make sure to subscribe over on YouTube and drop a review over on Apple or Spotify. It helps out the channel a ton. So thank you so much for tuning in to the AI Chat podcast. I hope you have a fantastic rest of your day. And if you're looking for a platform to test out all the models, make sure to check out AI box AI. I'll leave a link to that in the description so you can go check it out. $20 a month for 40 of the top AI models. Thanks so much for tuning in and I'll catch you next time.
The Mark Cuban Podcast: Detailed Summary of "Nvidia's Sales Recovery in China"
Release Date: July 21, 2025
In the episode titled "Nvidia's Sales Recovery in China," hosted by Mark Cuban, the discussion delves deep into the recent developments surrounding Nvidia's strategic maneuvers to resume chip shipments to China amidst a complex web of regulatory challenges and geopolitical tensions. The episode provides a comprehensive analysis of the factors influencing Nvidia's decisions, the implications for both the company and the broader tech landscape, and the ongoing tug-of-war between the United States and China in the realm of artificial intelligence (AI).
Mark Cuban begins by outlining the historical context of Nvidia's attempts to ship advanced AI chips to China. Over the past few years, Nvidia has faced increasing regulatory restrictions imposed by the U.S. government, aimed at limiting China's access to cutting-edge technology that could bolster its AI capabilities.
Quote: "Nvidia is set to start shipping chips back to China again after months of this thing being in limbo." ([00:00])
These restrictions were part of broader efforts to prevent China from surpassing the U.S. in AI advancements, ensuring that American companies maintain a competitive edge.
In a significant development, Nvidia announced its ability to restart the sale of its H20AI chips to China. This marks a pivotal shift from the stringent export controls that previously hindered such transactions.
Quote: "These new chips are supposed to be fully compliant and able to ship to China." ([00:01])
The H20AI chip represents the pinnacle of Nvidia's offerings that can legally be exported to China under current U.S. regulations. The decision to resume shipments comes after Nvidia secures the necessary government licenses, signaling a potential recovery in its sales pipeline.
The temporary halt in chip sales, initiated in April during the Trump administration, is estimated to have cost Nvidia between $15 to $16 billion in revenue. This financial impact underscores the significance of the Chinese market to Nvidia's overall business strategy.
Furthermore, Chinese tech giants such as ByteDance, Alibaba, and Tencent had been aggressively stockpiling the H20 chips before the restrictions took effect. These companies favored the H20 due to its superior memory bandwidth and Nvidia's robust software ecosystem, which facilitates easier deployment and integration.
Quote: "The H20 is the most powerful chip that they can legally sell to China under the current export controls." ([00:03])
Despite the restrictions, Chinese companies managed to circumvent some limitations by sourcing chips through Singapore, which accounted for approximately 30% of Nvidia's revenue. This workaround highlights the challenges in enforcing export controls and the resilience of global supply chains.
A pivotal moment in Nvidia's journey back to the Chinese market was the interaction between Nvidia's CEO, Jensen Huang, and former President Donald Trump. Huang's attendance at a high-profile dinner at Mar-a-Lago, where discussions about Nvidia's future may have occurred, coincided closely with the decision to pause the chip ban.
Quote: "Jensen Huang went to that dinner with this issue, tried to bring it up to Trump and probably had a conversation about it." ([00:16])
Shortly after this meeting, Nvidia unveiled plans to invest $500 billion in building AI servers in the United States, in collaboration with Taiwan Semiconductor Manufacturing Company (TSMC). This monumental project likely served as leverage in negotiations, resulting in the easing of export restrictions.
Quote: "Nvidia announced plans to build an AI server in the US which was worth as much as $500 billion over the next four years." ([00:14])
Cuban speculates that this deal was a strategic move by Nvidia to secure favorable terms for chip exports to China, balancing national security interests with commercial imperatives.
Singapore has emerged as a critical hub in Nvidia's sales strategy to China. By channeling chip sales through Singapore, Nvidia effectively bypassed some of the stringent U.S. export controls, allowing continued access to the Chinese market despite regulatory hurdles.
Quote: "Singapore is just kind of a fill in for China." ([00:05])
This approach, while profitable, raised questions about the transparency and compliance of Nvidia's operations. However, Nvidia maintains that it adheres to all applicable laws and regulations, denying any wrongdoing in its international dealings.
Despite the regulatory challenges faced by Nvidia, China's AI sector continues to flourish. Companies like Deep Seek have launched advanced AI models built using the more powerful H800 chips, which were initially banned for export in October 2023. The ability of Chinese suppliers to find workarounds underscores the limitations of export controls in completely stifling technological advancements.
Quote: "Deep Seek... was built like they essentially Deep Seq was completely trained using Nvidia's H800 chips, which are more powerful than the H20s." ([00:10])
Cuban expresses skepticism about the efficacy of long-term export restrictions, suggesting that China's determination to advance its AI capabilities may eventually overcome regulatory barriers.
Looking ahead, Cuban anticipates that the battle over AI dominance between the U.S. and China will intensify. He expects continued export restrictions on Nvidia's higher-tier chips, coupled with efforts by Nvidia to adapt by introducing compliant products and expanding domestic manufacturing initiatives.
Quote: "I personally don't really believe it... but regardless, this is kind of where the chips are on the ground." ([00:19])
The strategic investments in AI infrastructure within the United States, such as the proposed $500 billion project, are seen as efforts to secure technological leadership and mitigate reliance on foreign technologies.
Cuban concludes that while immediate gains may be realized through resumed chip sales to China, the long-term landscape will be shaped by ongoing negotiations, technological innovations, and shifting geopolitical alliances.
Conclusion
The episode provides an in-depth exploration of Nvidia's navigation through complex regulatory environments to re-establish its presence in the Chinese market. Through strategic negotiations, investment in domestic AI infrastructure, and adaptive sales strategies, Nvidia aims to balance commercial success with compliance to U.S. export controls. However, the persistent growth of China's AI capabilities and the inherent challenges in enforcing export restrictions suggest that the tech rivalry between the U.S. and China will remain a defining factor in the global AI landscape for years to come.
For listeners interested in staying informed about the evolving dynamics of business and technology, subscribing to "The Mark Cuban Podcast" offers valuable insights and analyses on pivotal industry developments.