원문 보기
Let's dive deeper into Nvidia. Nvidia is in the spotlight . With me now is Ray Wang, CEO of Constellation Research and author of "Everybody Wants to Rule the World." Looking at Nvidia in more detail, we take into account that the P/E ratio has decreased. Many guests say: " Look, it's better and cheaper now than it was in January." What do you think about Nvidia's recent results?
What impresses you the most ? Nicole, Nvidia is running at full capacity, but not getting all the benefits from it. This is mostly because the market is unsure whether there is still capacity left, and what will actually happen after Vera Rubin, Rubin Ultra, and other upcoming chips. But the reality is that the demand is there, and I think many are just being cautious.
They are waiting to see what the hyperscalers and the Magic Seven will do next week. Will they invest more? Are they going to retreat? Do they have the profits to support such growth? But this was the largest transfer of wealth from hyperscalers and software to hardware and chips that we have ever seen.
And you have a target price . You recommend buying for 280. Is that true? What will be the catalyst for us to grow from 209 to 280? I think as the fall season approaches, we will see the harsh reality that only the tech sector is growing . I know there is a rotation of capital out of tech right now, but when we start thinking about who is showing double-digit growth and who is poised to create the next wave of innovation, it will be tech again, and that is what drives the price target.
Another point is that we have not yet reached sovereign AI. You talked about agent-based AI. You talked about physical AI. All of these areas have not yet been fully assessed. I mean, these are huge markets that are coming up, especially in terms of sovereignty, as each country tries to figure out whether they can provide AI to their citizens and whether they can do it on their own terms.
This is going to be one of the biggest trends , and, you know, this expansion is going to continue. So we're only in the third inning of a nine-inning battle for AI, and we do n't even know who the winners will be. We know who these players are, but we're not sure who the ultimate winner will be , and that's a long way off .
Investors are punishing those who spend too much on AI. For example, this week Alphabet Google talked about increasing capital spending to 205 billion at the top end. And as for hyperscalers, now we see numbers from 1.2 to 1.7 trillion. Do you think companies should rely less on capital expenditures in a different business model?
That's a different matter, right? When we thought about the internet, we thought about clouds, right? It was about decentralization, openness, many players, prices were falling, and brick-and-mortar stores were losing out. Here everything is the opposite. It's centralized, it's more expensive, and you need— essentially, these are closed systems going to market.
So you need capital to play. You need scale to play . And there are only a few companies that will be able to compete in this market in the long term to claim to have created their own universal AI. They will provide opportunities for inference. They will be able to provide work at the periphery thanks to decentralized physical AI.
There are few players. And that's why hyperscalers are betting big on it. I expected Google to leave the capital expenditure forecast at 190, but something else happened: they said, "No, inflation and other things ." We will continue on this path. We had a great quarter and we're going to spend . This confused investors.
I'm thinking about what to expect from Meta, Microsoft, Amazon next week, but you said that what you saw at Google, where you expected stagnation, you think Alphabet's increase in capital spending is a good signal for durability. What does this mean? Why this is important is the fact that Alphabet continues to believe in investing in the years ahead.
So when at the beginning of the year we saw predictions of 700 billion for AI infrastructure from hyperscalers and others, and then we see new predictions: 800, 900 and eventually 1 trillion—it is this momentum that is driving the market. We are seeing continued acceleration in growth, not leveling off.
So, once these numbers drop, you will see a shift from investments in AI hardware, memory, and data storage back to software and services. Companies that can deliver results and solutions will take care of the rest, and this is where the likes of Palantir come back into the picture. That's when some stocks that have fallen sharply, such as ServiceNow, Salesforce, or Workday, will get a chance, because by then they will have implemented agent systems that will use this infrastructure.
How do you compare these companies to hyperscalers? We're waiting for news, and while you may not be giving investment advice, how do you assess the growth prospects of names like Salesforce compared to Amazon or Meta next week? Look, ServiceNow is a great example of what these metrics are all about.
I mean, ServiceNow has a "rule of 56." That's just a crazy 24.5% increase, as I remember from the charts. This, you know, means working at full capacity. They just acquired a company that will help them enter the Indian market and the financial services sector. They create a platform for agent orchestration.
Salesforce does the same thing. They are working at full capacity. They have excellent performance, and of course, they are laying the foundation so that they are not considered obsolete technology in the future. So I believe that if you have data and distribution, you have a "moat" around the business.
And this is not something that can be easily replaced by software that is considered free, or where the cost of the software is close to zero. And of course, we're all waiting for Nvidia's reports after that. This is one of the first companies to report. We heard a lot of interesting things from AMD today.
Does this mean anything to you? After all, we know that they are competitors. AMD is often called Nvidia's " little sister or brother" because the latter is a true giant and leader, but is AMD making real progress given what we heard today at the Advancing AI conference? There have been many comments from Lisa Su that the next phase of AI will encompass advanced models, AI agents, and physical AI.
Analysts talked about the company's leadership in the server segment and the expansion of its customer base in the field of AI. What does this mean for AMD and Nvidia? When I saw OpenAI , Anthropic, and Meta confirm that they were using Helios, that they were using chips that competed with Nvidia, and that they were counting on AMD to provide the compute layer for inference—it reassured me.
This shows that despite Nvidia's leadership in GPUs, AMD is also providing power. Moreover, if you compare the actions of AMD and Nvidia, they are reducing the cost of a single chip, performance is increasing, power consumption is decreasing, and people are really starting to see these benefits. Isn't that right?
This will be one of those times where we encounter the Jevons paradox: when costs fall, we increase consumption, but the demand for AI now seems endless, even though it is n't. It's just rapid, exponential growth. Yes, some of these things do demonstrate that there is demand, and Lisa Su noted that they are ready and able to provide the computing infrastructure needed to significantly scale this capability. Thank you, Ray Wong. Thank you.
Comments 0