원문 보기
Here's a big topic that currently only affects two major companies. Deferred demand for mergers and acquisitions. Now that the Biden era of excessive antitrust regulation has been replaced by the Trump era, where this regulation is almost non- existent. Most companies don't believe this moment will last long, so they use it to make deals.
When the summer is over, I believe we will return and see blockbusters like the rumored Bristol Myers acquisition by AstraZeneca or Stripe for PayPal. I know the targets in these cases may not be interested in the merger, but the potential buyers, oh yes, they are totally ready. They seem very interested and I think they are willing to pay well.
How to play on this merger mania? Without choosing potential targets. This is a game for fools. Instead, buy shares of companies that facilitate these deals, and why not Goldman Sachs and Morgan Stanley? Both have excellent M&A departments. This M&A advisory business is a real goldmine . We are talking about huge profits per employee, and therefore earnings per share.
Finally, the technology sector has had such a boom that portfolio managers want exposure to other market sectors where there is still innovation. Think of them as non-tech technology stocks. For the club, this is a charitable trust club, we emphasize Eli Lilly and Johnson&Johnson. We talked about them today, I discussed them in detail with Jeff Marks in our morning meeting.
Both companies had great quarters and are worth buying during any downturn. BD has a lot of medical technology. We will hear them later in the program. Amgen just completed a great quarter, and it's time to acknowledge that Moderna has begun to realize its dream of specialized vaccines for specific diseases.
As I said, the company received FDA approval for its mRNA flu vaccine. Well, there's a reason it's up 83% in a year. Of course, there are no guarantees when you engage in thematic investing. You may mistakenly choose the wrong pharmaceutical promotion. Maybe sales and trading could cause failure at Goldman Sachs.
You buy a cruise line thinking consumers are living life to the fullest , and then someone in the group reports a weak quarter and drags everyone down. But the point is, I think this quarter's information is fresh enough that you could pick a stock in a travel company, a semiconductor equipment manufacturer, a cybersecurity company, something in the M&A world, or maybe the Department claims that the deal was structured to avoid regulatory oversight.
As for the stock, Nvidia is up about 18% year-to-date, but has fallen this week . Welcome back to our co-hosts, Tom White and Kevin Hanks. Back to Nvidia, Kevin, what are your thoughts? Well, the question on many's mind is: does Nvidia's valuation match its growth and long-term projections? People are trying to figure it out.
For the first time in a long time, an earnings report has led to a rise in stocks after its release. So for the last few quarters, it's been somewhat new. But I think what impressed investors the most was the projected 70% revenue growth through fiscal 2028 . They start to look at the longer term, and I think that helps.
However, even Jensen Huang called Nvidia's valuation one of the great mysteries a few months ago, and he's right. Given other estimates, it seems that the AI sector is being severely undervalued, as if such figures are impossible to sustain. However, they continue to do so . So we find ourselves in an interesting situation where we look at Nvidia, one of the largest companies in the world with such a huge market cap, and ask: is it undervalued?
This is a strange question, a strange situation we find ourselves in, Nicole and Tom. I'm glad you mentioned the 70% revenue growth he's forecasting. At the Goldman Sachs conference, he also denied claims that AI is slowing down and gave figures that this market could grow to $3-4 trillion by 2030. Tom?
Yes, Kevin mentioned that. The forward P/E ratio for this company is low compared to historical performance, right? I think it's below the 20x level for the next 12 months. If you look at their competitors, AMD, Intel, they are much higher. So, in that sense, it is inexpensive. I guess the law of large numbers applies here , right?
This is the largest market capitalization among stocks. This is, you know, the largest company on the market. That's part of the problem, and also, I think, this cyclical investment pattern— no, not a pattern, but what's happening now, investing in companies that buy GPUs and other technologies from them.
But they are trying to expand to a full stack of technologies, right? They just bought Hugging Face. They are entering the processor market with Vera Rubin. This will become a direct competitor to Intel and AMD and will start taking market share away from them . So this is a positive point. They were at the Goldman Sachs Communicopia conference and made a statement there in which they essentially stated that investments are not cyclical.
Overall, Grace Blackwell's deliveries increased by 27% month-on- month. So they brought some clarity to this growth rate. Now look at the latest financial results. What was the growth rate? 106% compared to last year, to 96.2 billion. Data center revenue grew 117% year-over-year . They demonstrate simply incredible results.
So one of the caveats was gross profits, right? They just made it 75%. They are expected to decline by a percentage point or two, stabilizing at around 71.5-72%. Hey, it's still pretty good for this particular niche. So, yes, it was a pretty strange trade, but if you expand on Nvidia's chart and look at what it has done over the past 5 years, it might clarify things, as we are up 17% this year but have seen a lot of consolidation over the past 12 months.
Before we continue, be sure to technologies driving the market. Nvidia is facing a new challenge in the field of AI. More and more companies are developing their own chips. Specialized output systems are becoming increasingly important. And instead of fighting this trend, Nvidia is choosing to embrace it.
Jensen Huang's message is simple. Nvidia already has the largest AI infrastructure in the world. Its platform is gaining share in both closed and open AI models. So, instead of trying to stop specialized chips, Nvidia wants to connect them to its ecosystem. This is exactly what we are seeing with D Matrix.
This startup has developed a specialized inference processor designed for ultra-low latency AI workloads. And now D Matrix is partnering with Nvidia to connect its technology to Nvidia's infrastructure using NVLink fusion. The reason why this is important is simple. Nvidia has already built a huge installed base.
Nvidia hardware is deployed in cloud providers, countries, and data centers around the world. For a smaller chip company, building an entirely new infrastructure ecosystem would take years. So, instead of competing with Nvidia's installed base, D Matrix can integrate into it. The company spent over 7 years developing its memory-based output computer technology.
And time is very important for this. Over the past year, AI agents and programming systems have dramatically increased the demand for ultra-fast inference. When developers interact with AI agents to code, latency matters. And the faster the system responds, the more useful the agent becomes. This is where specialized inference equipment can potentially add value.
D-Matrix's upcoming Raptor XPU is particularly interesting because it takes a different approach to memory. Instead of the traditional use of HBM, the company is developing a 3D stacked DRAM architecture that places memory much closer to the compute unit. The goal is to overcome one of the biggest problems in modern AI, the so-called " memory wall." AI systems increasingly require moving huge amounts of data between memory and the processor.
This creates issues with latency, power consumption, and cost. D-Matrix believes its architecture can reduce these bottlenecks. And the company expects the first integrated product to hit the market in about a year. The opportunities for customers are also wide. The company says it works with clients including hyperscalers, Frontier AI projects, cloud inference providers, and even high-frequency trading firms.
So, this isn't necessarily a replacement for Nvidia GPUs. This is adding specialized calculations alongside them. And this distinction is extremely important . It seems that Nvidia understands that the future of AI computing will not necessarily be controlled by one type of processor. Different workloads will require different architectures.
Some tasks will benefit from Nvidia GPUs, while others will benefit from specialized inference processors. Nvidia's strategy seems to be that if you can't beat every specialized chip, make your platform a place where those chips can connect. And that could make Nvidia's ecosystem even harder to replace.
Comments 0