Breaking Down Nvidia's Q2 Results and Forecast

Show Transcript
Bloomberg 23:11 · 10.5k Views

The Futurum Group CEO Daniel Newman breaks down Nvidia second-quarter earnings and forecast on “Bloomberg The Close.” Nvidia said third-quarter revenue will be about $32.5 billion, the company said. Though analysts had predicted $31.9 billion on average, estimates ranged as high as $37.9 billion.

Disclaimer: This content is for informational and educational purposes only and does not constitute a recommendation or endorsement of any specific investment or investment strategy. Read more

Transcript

  • 00:00 I mean, look, Dan, if you need a chip
  • 00:02 that can sort of power these data servers that can sort of be at the focal point
  • 00:06 of what you need to crunch these numbers to be at the center of AI, you have to go to NVIDIA.
  • 00:12 That's going to change.
  • 00:13 At some point.
  • 00:14 They're going to have competitors and they're going to have meaningful competitors.
  • 00:16 How soon do you think we're going to start to see that competition?
  • 00:21 NVIDIA is off to a fast start.
  • 00:22 And of course, more than a decade of research and investment that went into building this platform.
  • 00:27 It wasn't just hardware.
  • 00:28 You're not buying a GPU,
  • 00:29 you know, you're buying a system.
  • 00:31 They're putting it all together, the software, the libraries, the frameworks, and they've made AI accessible, enabling and developers to write programs to this hardware that's really powered this whole movement.
  • 00:42 But let's be honest,
  • 00:44 Romain,
  • 00:45 AMD is going to enter the fray in a big way.
  • 00:47 We're seeing what Lisa Sue is doing.
  • 00:48 And of course,
  • 00:49 we're seeing every hyperscaler, whether that's AWS with Inferentia and Tranium, whether that's Microsoft Maya,
  • 00:55 Google's training, its newest Gemini models on its TPU.
  • 00:58 They all want to enter the fray.
  • 00:59 They understand the importance of building this vertical integration.
  • 01:03 But I do think it's going to come slow.
  • 01:05 So,
  • 01:05 you know, this was a clean quarter.
  • 01:07 NVIDIA did what it said it would do, but the market doesn't want 100%.
  • 01:10 They want 100 and 5200% acceleration of growth.
  • 01:13 They want to know AI is going to land in Blackwell, is going to be right on time
  • 01:16 because that's what has made this a $3 trillion company.
  • 01:19 It's interesting when we look at, I guess, what NVIDIA says and how they characterize it in their presentations, but also with some of the folks we've spoken to in this program,
  • 01:27 they're basically said that right now there is no demand issue.
  • 01:29 The demand is there and it is ample, is right now still a supply issue.
  • 01:33 And as long as NVIDIA doesn't necessarily have an AMD or TSMC or some other company
  • 01:39 capable of providing comparable chips, this is still their world, isn't it?
  • 01:44 Yeah, this is
  • 01:45 this is not a demand issue.
  • 01:47 Of course, we are looking at that longer tail.
  • 01:48 We've seen the comments from Goldman, the comments from Sequoia
  • 01:51 on whether or not AI is being deployed at the, you know, at the enterprise level, financial services companies, hotels, you know, our restaurants and retail using this stuff.
  • 02:00 But these CapEx investments, you heard from Satya, you heard from Sundar, Mark Zuckerberg,
  • 02:05 they're not going to slow down anytime soon, but they are going to hedge.
  • 02:08 So back to your last question, Romaine, they are going to build their own accelerators.
  • 02:11 They're partnering with companies like Broadcom
  • 02:13 to build these XP, which we actually see growing faster than these GPU systems according to our intelligence.
  • 02:18 And over time, as we see,
  • 02:20 you know, inference become the killer workload rather than so much focus on training, we're going to see different architectures of chips start to play a really meaningful role in this AI,
  • 02:30 you know, this AI build out.
  • 02:31 So
  • 02:31 I see it both ways.
  • 02:32 I think for the next three or four quarters, Nvidia's in the driver's seat.
  • 02:35 And if they do things right, they're going to have so much market share that you think about what Intel had with its CPU business in the data center.
  • 02:41 They're going to be in a great position, and it's going to be their market to lose.
  • 02:44 OK.
  • 02:45 So I just want to go back for a moment to what investors are reacting to right now, which is the outlook for this current quarter.
  • 02:52 NVIDIA says it sees revenue this quarter, third quarter, 32 1/2 billion dollars ±2%.
  • 02:58 So
  • 02:59 the consensus estimate was for $31.9 billion, so just under $32 billion, which would already be a 76% jump
  • 03:06 versus last year.
  • 03:07 The highest estimate, though,
  • 03:09 was just under $38 billion.
  • 03:12 So,
  • 03:13 Daniel, when we look at a company like NVIDIA and the kinds of stock moves it's had of 160% this year,
  • 03:20 up
  • 03:20 35% since its last earnings report,
  • 03:23 you know,
  • 03:25 investors want to see it delivering
  • 03:26 in in every way on all cylinders.
  • 03:29 Is it a case where anything other than the high assessment would just automatically get some kind of a pull down?
  • 03:35 And
  • 03:36 in order for the stock to to,
  • 03:38 you know, attract
  • 03:39 investors once again,
  • 03:40 the commentary on the conference call really has to
  • 03:43 outperform.
  • 03:45 Yeah,
  • 03:46 we've seen so many quarters in a row where they've come and beaten substantially above even what that sort of
  • 03:51 guide was.
  • 03:52 And and over the past several quarters we're starting to see,
  • 03:54 you know, the deceleration quarter on quarter and year on year and that's because they've had those massive 3 and 400% growth.
  • 04:01 Now what we're really up against is, you know, we are up against whether or not this is sustainable.
  • 04:06 Do investors want to get out after making thousand plus percent returns?
  • 04:10 Are they looking at this and saying maybe we'll let it cool, we'll let it breathe?
  • 04:14 And I said this coming into the quarter, Scarlet, there was no result other than perfect.
  • 04:18 That would probably create a big lift.
  • 04:21 But there was a lot of downside, not just for NVIDIA, the whole market if they didn't come up big.
  • 04:25 And I think futures are showing that as well.
  • 04:27 So NVIDIA is really the only game in town right now when it comes to these AI chips to power data centers.
  • 04:34 We know that NVIDIA has pricing power and therefore it has margin power.
  • 04:37 Do you therefore look at the gross margin number as an accurate gauge of demand for NVIDIA chips or is it not all that sensitive because the pricing is is not dynamic?
  • 04:47 Yeah, I think, I think the margin power is indicative of the influence that NVIDIA has.
  • 04:52 Of course,
  • 04:53 you know, it has its large partners, it's building systems, not general, not necessarily selling just a chip.
  • 04:58 It's systems and the system.
  • 05:00 Margins are very high because the demand is, is, is substantial.
  • 05:03 You know, we're hearing everyone from open AI to, you know, to other smaller enterprises and tier twos having a problem getting access to the chips and systems that they need.
  • 05:11 And therefore companies like Meta X, Tesla,
  • 05:14 Microsoft, Amazon, taking all of that demand and it gives them great pricing power.
  • 05:17 I don't think until AMD
  • 05:20 really starts to see acceleration in its business and then to some extent, when these hyperscalers start to really see growth in their homegrown platforms,
  • 05:28 is NVIDIA going to be under substantial pressure to to play in the pricing battle?
  • 05:33 But I do think over probably 4 to 8 quarters, we're going to see some pressure on margin, and that is going to be something to watch.
  • 05:38 Yeah,
  • 05:39 absolutely.
  • 05:39 And we should point out our team here that's been crunching all these numbers in there
  • 05:43 now pointing out this is the smallest top line beat that we've seen for NVIDIA in six quarters here.
  • 05:49 Nevertheless, still, again, we always have to point out triple, almost triple digit growth here,
  • 05:54 which would normally be something to celebrate.
  • 05:56 Dan, you mentioned the hyperscalers.
  • 05:58 I am about the end use use cases for AI because ultimately that will probably determine the longer term growth for NVIDIA and some of its peers here,
  • 06:07 even past the hyperscalers.
  • 06:09 When do you start to see a more meaningful use case for AI and a meaningful use case for AI that's actually going to be profitable for the companies doing it?
  • 06:18 We're starting to see them Roman, we're seeing it with Walmart.
  • 06:21 They announced some pretty substantial efficiency gains
  • 06:23 in their systems just in this most recent quarter.
  • 06:26 We heard Andy Jassy talking about some of their code development and the efficiencies talking about,
  • 06:31 you know,
  • 06:31 years of developer time that was able to be reduced because it was
  • 06:37 because of what they've been doing with with generative AI tools for code development.
  • 06:41 But that is exactly where my head is at to remain.
  • 06:43 I mean, building this into Salesforce instances, you know, they had their entrance today.
  • 06:47 Marc Benioff's going to talk about this.
  • 06:49 But companies
  • 06:50 and their CRM systems, can they get better understanding of when customers are going to buy and getting them the right message in the right channel at the right time and using generative tools to do this.
  • 06:59 We all want to see this digestion of all this CapEx turn into really meaningful consumption on the OpEx side.
  • 07:06 And that's happening more slowly.
  • 07:07 I, I've been saying to the market that I think this is a four to even 12 quarter lag depending on industries.
  • 07:13 We are getting value out of AI.
  • 07:14 So that's also kind of a misconception.
  • 07:16 We've been using AI algorithms are 4 decades old.
  • 07:18 We've been optimizing Bloomberg's been using it in its own assessments for a long time.
  • 07:22 But generative tools need to bring more value and we need to understand monetization better in these next quarters for more of these software and and other business.