The United States has approved exports of Nvidia’s H200 artificial intelligence chips to China, marking one of the sharpest reversals in Washington’s tech policy since restrictions tightened last year. The move comes as Beijing signals it will limit access to the processors, highlighting the fragile balance in U.S.–China technology relations.
BNN Bloomberg spoke with Ivana Delevska, founder and CIO at Spear Investment, about the potential impact on Nvidia, developments in China’s semiconductor ambitions and what the shift means for the broader AI investment cycle.
Key Takeaways
- The U.S. move to allow Nvidia’s H200 chip exports marks a major policy reversal after months of tighter tech controls.
- China is preparing to restrict access to the chips, complicating Nvidia’s ability to regain market share in the region.
- Market uncertainty remains high as tariffs, permit requirements and political tensions cloud the demand outlook.
- Nvidia maintains an edge in general-purpose AI computing through its CUDA software and frameworks, even as competitors advance.
- Analysts see little evidence of an AI bubble, noting strong innovation and bottlenecks in networking, components and power systems that continue to drive investment.
Read the full transcript below:
ROGER: U.S. President Donald Trump is allowing Nvidia to sell advanced chips to China, marking a shift in U.S. tech policy. High-end H200 chips are getting the green light, but China is set to limit access. Here to talk about that is founder and CIO of Spear Invest, Ivana Delevska. Ivana, thanks, as always, for joining us.
IVANA: Thanks for having me, Roger.
ROGER: OK, many angles to this. First up, Nvidia. How is this a win for Nvidia?
IVANA: Well, Roger, just for background, Nvidia was expected to sell approximately US$17 billion to China this year. That got brought down to zero. So in the current analyst estimates, there is nothing really included for China. So versus that, it’s a pretty significant incremental positive.Now, it does come with a lot of questions we don’t know the answer to, and those are whether China will be encouraging the sales of these chips or discouraging them. And that’s really what the market is a little uncertain about, in addition to the 25 per cent tariff. So it sounds like it may be a little bit too late for Nvidia to penetrate this market, but still, a lot of details to come.
ROGER: I know we’re already seeing China apparently considering limiting access — you’d have to get a permit and explain why you need it. How detrimental would that be?
IVANA: Well, Nvidia chips are very strong for general-purpose computing. So anytime you need Nvidia’s CUDA software, there is really no better substitute for it. From that perspective, I think it would be in China’s best interest to allow the sale of these chips.However, every time you have a situation where it’s not really a free and open market, it’s not as easy for the company to penetrate it. So I think it’s not necessarily detrimental — it’s just not going to be a very big positive surprise for the stock.I think if they were able to really penetrate the market in a more free-market type way, China could be a really big opportunity for Nvidia. But thus far, with this sort of friction and tariffs, I don’t know that we’ll be able to see a meaningful impact in the numbers. And this is why you see the stock not really responding all that positively to the news.
ROGER: Now, there are critics of Trump who are concerned this could help China win the AI war if they get these chips, because they’re considered ahead when it comes to AI. Is that the case, or is China still a long way off? They’re still a generation behind when it comes to chips — is that holding them back?
IVANA: Well, Roger, in our view, the way to win the war is with continued innovation. So it’s really all about staying ahead on the innovation front rather than restricting sales.I don’t necessarily think allowing the sales to China will accelerate China winning the war, because if they’re not provided, then China would develop them internally. So at this point in time, Nvidia being the dominant platform would be a lot more beneficial to the U.S. versus China having its own chips developed internally.Just to provide some background here, at this point, many companies know how to make this chip — so it’s not really a secret. It’s all about how quickly they can advance them and how quickly they can provide features that are beneficial to different use cases.We’re really beyond the stage where Nvidia was the only game in town. You see competitive chips from TPUs, from Google or AWS chips, and the same with Chinese competitors. There are many alternatives to GPUs today. However, Nvidia still wins in that general-purpose computing. So if you’re making these chips available broadly, the more likely you are able to establish dominance.If other companies or Chinese competitors develop this in-house, then they would obviously have that advantage.
ROGER: Does this say anything about the bigger picture for AI and concerns about a bubble in the tech sector, or is this just the next move in this situation?
IVANA: Well, I actually think all the developments we’ve seen over the past year have been very positive for the AI cycle, and I don’t really see a bubble anywhere. If you look at Nvidia’s valuation specifically, it’s trading at the low end of its valuation range, and some of that is because of the concerns we highlighted earlier.But if you look at the sector more broadly, you’re seeing a lot more innovation. You’re seeing companies come to market. You’re seeing a lot of benefits across the value chain. This is where we’re finding the most investment opportunities — companies in the networking space, whether it’s on the electrical component side, optics, power generation. This is where the bottlenecks are today, and that’s where most of the opportunities are.We disclose all of our holdings on our website, and you’ll find some of these companies there.
ROGER: There were concerns that if they didn’t allow the chips to go in, China would catch up and build its own internal system. Is China catching up when it comes to chips? Will they be able to catch up?
IVANA: Yes, absolutely. They’re definitely catching up when it comes to chips. They’re catching up on the hardware side. They’re doing very well on the networking side. So that’s another area where even if the chips themselves aren’t performing as well, once they’re connected as a system, they actually have pretty comparable performance.I think where Nvidia is really differentiated is in providing these frameworks that sit on top of the GPU. When you hear Jensen say even if we were to give away the chips, even if competitors were to give away the chips for free, our product would still be better — that comes from these frameworks that are free and come along with the Nvidia chip.So I think Chinese competitors will catch up on the hardware or bare-metal performance side, but they’re still going to be missing some of these frameworks, which are what allow competitors or new entrants to develop AI products quickly.
ROGER: OK, we’ll have to leave it there. Ivana, thanks, as always, for joining us.
---
This BNN Bloomberg summary and transcript of the Dec. 9, 2025 interview with Ivana Delevska are published with the assistance of AI. Original research, interview questions and added context was created by BNN Bloomberg journalists. An editor also reviewed this material before it was published to ensure its accuracy and adherence with BNN Bloomberg editorial policies and standards.
