Nvidia delivered another record quarter as revenue and profit topped expectations, reinforcing confidence that spending on artificial intelligence infrastructure remains strong. The chipmaker also issued a robust outlook for the next quarter, while investors continued to watch the broader impact of AI adoption and competition across the semiconductor sector.
BNN Bloomberg spoke with Ivana Delevska, founder and chief investment officer at Spear Invest, about Nvidia’s latest earnings, the rise of agentic AI, growing demand for AI computing power and why she believes opportunities remain across the broader AI supply chain.
Key Takeaways
- Nvidia forecast quarterly revenue above US$90 billion, with expectations that demand tied to AI infrastructure will continue to grow.
- Growing adoption of agentic AI could drive higher demand for CPUs and computing power beyond traditional AI model training.
- Nvidia maintained its leadership position in AI chips through strong efficiency and performance advantages over competitors.
- Delevska said China is unlikely to become a major growth driver for Nvidia as domestic alternatives continue to develop.
- SpaceX’s long-term growth potential could expand significantly if lower launch costs make new space-based applications commercially viable.
Read the full transcript below:
ROGER: Well, Nvidia reporting yet another record quarter, with sales and profits meeting expectations, seemingly suggesting that there’s no slowdown in sight for the AI trade. Nvidia, of course, provides chips to leading AI model developers, including OpenAI and Meta, and it’s the world’s most valuable company. Let’s get more now from Ivana Delevska, founder and CIO at Spear Investment. Ivana, thank you very much for joining us today.
IVANA: Thanks for having me, Roger.
ROGER: Another one for Nvidia, but the market today is saying, “We expected it.”
IVANA: Yeah, well, Roger, interestingly, the numbers came out very strong, especially the guide for next quarter, over US$90 billion. So while Nvidia investors were positioned for a strong report, the implications across the value chain are actually pretty positive. So we’re seeing a lot of the component suppliers, like their partner for CPUs, Arm, or their networking suppliers, those stocks are going up much more than Nvidia, which is seeing a bit of a muted reaction to the numbers.
ROGER: And that $90 billion, is that realistic?
IVANA: Yeah, absolutely. And we believe that it could be conservative. What’s happening in the second half of the year, they’re launching the Vera Rubin platform, but even more importantly, as agentic AI takes off, and we’re in the early innings of that, they’re going to be selling a lot more CPUs. So they guided on the call for US$20 billion incremental CPUs, and we believe that this could be conservative.
ROGER: And what gives you the confidence with, I mean, agentic? I think everybody is now starting to understand AI, and now agentic is the next phase. What makes you think it’s going to take off as much as it will?
IVANA: So, Roger, if you look at what’s been happening, the first wave of AI was really mostly model-training related, and really you would be prompting these models and getting a one-off response. What we see coming next is these models that are reasoning and iterating, and that’s really what Anthropic’s really strong revenue is based on, right? So they launched their new model, Opus, and they’ve seen just skyrocketing revenue to over US$40 billion on an annualized basis in a matter of a few months. So we think that these are early signs that agentic adoption is coming, but we’re going to see more of this with OpenAI rollouts, with enterprise adoption. So yeah, we think we’re in the early innings, but Anthropic’s revenue is really the first sign that we saw.
ROGER: Okay, and with Vera Rubin, what makes this the next level? What takes it to the next level?
IVANA: Well, with each generation, the key is tokens per watt, so they’re really using the power unit that you have to generate the most tokens possible. And basically, as we go from generation to generation, the tokens are becoming much more accessible, broadening applications to a much wider range. So Nvidia, we believe, still remains the leader in terms of cost per watt. We’re seeing competitors like Google coming up with the TPU, the AMD chip — they also have strong value propositions — but nothing really compares to Nvidia for general-purpose computing.
ROGER: Okay, and I have to ask — I’m going to say I’m asking for other people — but tokens, what exactly is that? What’s that concept?
IVANA: So that’s basically AI usage. That’s how AI usage is being measured. So every time you go and prompt ChatGPT for something, you are generating tokens. So it’s interesting because some of the tech companies now are measuring people’s productivity, like employee productivity, in terms of how many tokens have you been using, right? So that means that you have been productive and using AI and generating insights. So that’s a key metric as we go forward to monitor the usage of AI.
ROGER: And I think one of the big things still, $0 from China. Does it need to get into China, or can it even, do you think?
IVANA: Well, we think the boat for Nvidia to sell revenues to China has sailed. They still may be able to capture a small part of that market, but really the opportunity was to capture that market early. Now the Chinese companies are developing their own internal solutions. They’re nowhere near as good as Nvidia, but they’re compensating for it on the software side. So we think that while companies still may like it, the incentives are not quite there for Nvidia to sell to China. We expect that it will be a small market for them, but nowhere near as big as it could have been if they were able to sell these chips early to China.
ROGER: Okay, and we have to talk about this one. SpaceX, the IPO — we’re finally seeing some numbers. What do you think of those numbers?
IVANA: Well, Roger, the key thing with SpaceX is launch, right? So as we go forward into the next several years, space is opening up to be a huge potential for several different applications, data centres in space being one of them, communications being the earlier one, and the one that we’re seeing the revenue numbers mostly reflect. But the key will be launch, because only SpaceX has been able to prove launching rockets at a reasonable cost, where commercial applications can become viable. So we think with their new rocket, the Starship, they’re going to be able to significantly drop costs from the thousands — like US$3,000 per kilogram — to less than US$300 per kilogram, and that will broaden out significantly the type of applications that you’re going to be able to see. Hence the timing of this IPO is really coming ahead of that cost decline.
ROGER: I mean, we’re now hearing that $5 trillion for a valuation. Is that realistic?
IVANA: Five trillion is probably not realistic. I think something closer to $2 trillion at the IPO, but over the long run, I think they can easily trend towards that number. So the first use case that we see today, as I mentioned, is communications with Starlink. I think that’s going to be just one of many use cases. So as we develop data centres in space, that’s going to be a three- to five-year timeline just to prove the concept. I think that’s when you’re going to be seeing the valuation trend towards the $5 trillion. I don’t know that it would be coming out at that valuation.
ROGER: Okay, we have to wrap it up there. Ivana, but always a pleasure to talk to you. Thanks for joining us.
---
This BNN Bloomberg summary and transcript of the May 21, 2026 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.
