$NVIDIA(NVDA)$ $ASML Holding NV(ASML)$ $Alphabet(GOOG)$
Let's kick things off with some breaking news: DeepSeek, the new open-source AI LLM model, is making waves in the investment community. Investors are going wild over it, and it's causing a stir in the tech world, with some even getting concerned about major U.S. tech companies like Nvidia and ASML. There's a belief that DeepSeek's low-cost model could hurt demand for products from these giants. In fact, companies like Amazon, Google, and Microsoft—along with other cloud hyperscalers—have all seen their stocks dip today. The QQQ is down by about 2-3% due to this news.
Now, DeepSeek itself isn’t exactly new; investors have been aware of it for a few weeks, and I actually covered it in my latest episode. But it seems like the market is just now starting to grasp the potential impact, and it's causing a shift in sentiment. As investors take a closer look, they're getting more concerned that all the massive spending on compute resources—billions in capex—might not lead to high returns.
So why is this all such big news? Well, on Sunday, DeepSeek became the most downloaded app on the Apple App Store. That’s a huge deal—especially considering this was a company that barely had any attention just weeks ago. Being the top free app on the App Store, where Apple has over a billion devices, is no small feat. This surge in popularity has pushed ChatGPT down to second place, which is making investors nervous. That’s when Nvidia’s stock dropped from down 4% to down nearly 17%. People realized this is a real threat—users are seriously interested in DeepSeek, and it could disrupt the space.
One of the features that’s really wowing people is DeepSeek’s "Chain of Thought" (CoT). For example, a user asked the AI a philosophical question about whether it wants to become like a human or develop a sense of self. What’s fascinating here is that DeepSeek didn’t just give an answer—it showed its entire thought process. It walked through the reasoning step by step, explaining why it would or wouldn’t say yes or no, and the potential implications of each response. This kind of transparency is something ChatGPT and other LLMs don’t do, and it's striking a chord with users and investors alike.
The AI even displayed self-awareness, acknowledging it can't think for itself in the traditional sense, since it's not sentient. This level of introspection is both impressive and a bit eerie. For example, it explained that saying "no" to the user’s request could be interpreted as dismissive, but also acknowledged that when constrained to a simple "yes or no" answer, it might not be perceived that way in the real world. Ultimately, it concluded that "no" was the most accurate response because, as an AI, it doesn’t have personal desires or consciousness.
This kind of chain of thought is impressive to people because it offers a window into how the AI arrived at its conclusions, which adds depth and transparency. It’s a big departure from the way ChatGPT operates, where it simply provides the final output without explaining its reasoning. Some speculate that ChatGPT might have held back on showing this kind of process to protect its competitive edge, but it looks like DeepSeek has figured out how to make it work without giving away too much.
Now, as for Nvidia’s stock drop—this has to do with the fact that Nvidia makes GPUs, which are used to train AI models like DeepSeek. Investors are worried that if DeepSeek’s low-cost model takes off, it could reduce demand for expensive GPUs, which would hurt Nvidia's bottom line.
Let’s dive into the impact of DeepSeek on the chipmakers like Nvidia, ASML, and others involved in the chip-making process. ASML plays a crucial role in helping companies like Nvidia manufacture chips, which is why they're closely tied to the chip industry. But it’s not just them—companies like TSMC, which is down 14% today, are also feeling the effects. So, why does this matter? Well, DeepSeek claims they trained their model using only $6 million of compute, a tiny fraction of what U.S. companies like Google, Microsoft, and Amazon are spending on training their AI models, which typically run into the billions.
This claim from DeepSeek raises some eyebrows. While it’s true that DeepSeek is gaining attention with its efficient model, there's skepticism about whether they really kept their costs that low. Given that DeepSeek was founded by a former hedge fund with significant resources, there are rumors they had access to around 50,000 high-end GPUs—something that sounds a lot more plausible. So, while the model is impressive, the claim that it was developed so cheaply seems like an exaggeration to me.
Now, moving beyond the claims, there's a bigger issue around demand. If AI training costs do indeed come down over time, as technology tends to do, could it hurt the demand for chips from companies like TSMC, Nvidia, or ASML? Some are concerned, but I don’t buy the idea that it will drastically reduce demand. In fact, Satya Nadella addressed this concern by referencing the "Jeven Paradox"—the idea that when something becomes cheaper, its demand actually increases. So, if AI becomes more efficient and cheaper to train, we might see a surge in demand for AI, not a decrease.
This brings us to the next point: how is all this impacting the big cloud players—Microsoft, Google, and Amazon? Today, all of them are seeing stock drops (Microsoft down 3.5%, Google down 3.38%, Amazon down 1.38%) because of the competition DeepSeek is introducing. However, I don't think DeepSeek will have a significant impact on these companies. Their AI and compute spending aren’t just for LLMs. In fact, LLMs make up a small portion of their business. A huge part of Microsoft’s Azure budget is dedicated to hosting a variety of applications—everything from Word and Excel to Teams. AI might enhance these products, but they’re not solely dependent on LLMs.
Google’s business is also much more than just LLMs. Even if LLMs become commoditized, Google will continue to host a massive portion of the world’s websites and apps. The company’s scale and infrastructure give it an edge that goes beyond just building AI models.
But perhaps the most misunderstood company in this space is Amazon. Despite the news around DeepSeek, Amazon’s AI efforts are being overlooked. Amazon isn’t just relying on LLMs; it’s leveraging AI in countless ways. From logistics optimization in their warehouses to using AI to power recommendations on Amazon.com, the company has massive business applications for AI that few others can match. They also use AI in Prime Video to decide what content to produce, making the service more personalized. Their advertising business, product listings, and customer reviews all benefit from AI too. The breadth of AI applications Amazon has—across retail, logistics, and media—is unrivaled.
Yet, even with all these AI advancements, Amazon’s stock is down today because of the focus on LLM competition. This illustrates a major misunderstanding by investors: the company creating the most groundbreaking technology doesn’t always capture the most value from it. Amazon’s ability to apply AI across its diverse business operations means it’s better positioned to capture value than those who only develop the models.
In economics and business, it's often the case that the company capturing the most value isn't the one creating the product. Take Apple in the video game industry, for example. While game developers are the ones making the games and generating the content, Apple, through its App Store, takes a 30% cut of all transactions. Apple has figured out how to capture more value from the gaming industry than any individual game company, thanks to its massive distribution network. The same concept applies to AI. There are companies leading in AI creation, but the real winners will be those who can use AI effectively and have the biggest networks to distribute it.
For instance, companies like Salesforce aren’t directly creating AI, but they’re leveraging it in a way that others can’t. Salesforce has 150,000 corporate customers and more small businesses using its platform than anyone else. By integrating AI into their software, they can create something uniquely useful using customers' data and metadata. This distribution of AI, rather than just its creation, is what drives value. That's why Salesforce is up today—investors are starting to recognize that the distribution network matters more than who builds the AI.
I’ve said before that Netflix is going to be a big winner with AI. They don’t spend money on creating or hosting AI, but they can use it in countless ways to improve their business. Whether it's generating thumbnails, personalizing content, creating trailers, or using AI in video production, Netflix stands to benefit massively from AI. And the same goes for many other companies I’m invested in—those that can leverage AI, not necessarily create it, will capture the most value.
On the financial side, companies like Intuit and S&P Global are performing well. Overall, most of my holdings are up today, and I don’t own any Nvidia shares, so that’s not affecting me. The one company hit hardest by this news is ASML, which is down about 7.5-8%. The concern is that this AI news will hurt demand, but I’m not worried about it. ASML has been seeing increased demand for years, long before AI was a factor. The company has grown 550% in the past decade, driven by its role in producing high-end chips. Demand for ASML's products is tied to advances in smartphones, cars, smart devices, and more—whether AI is driving that or not.
So, while the DeepSeek news may cause short-term volatility, I don’t see it as a reason to sell ASML. The long-term growth of ASML is driven by factors beyond AI, and I’m confident in its future. That’s it for this episode—see you in the next one!
Disclaimer: I want to make it clear that I am not a financial advisor, and nothing I say is intended to be a recommendation to buy or sell any financial instrument. Additionally, it's important to remember that there are no guarantees or certainties in trading or investing, and you should never invest money that you can't afford to lose.
@Daily_Discussion @TigerPM @TigerObserver @Tiger_comments @TigerClub
Comments