$NVIDIA (NVDA.US)$ The content of the GTC keynote address is summarized in an easy-to-understand manner, so it is recommended. - YouTube
Hearing this makes it clear that NVIDIA's Global Strategy is significantly ahead of competitors in terms of breadth, foresight, and feasibility.

$NVIDIA (NVDA.US)$ Generative AI is still in a transitional phase as a technology, and while its growth is rapid, it holds the potential for various innovations to occur worldwide. For instance, the currently dominant Transformer architecture in generative AI may be replaced by something else as innovations happen in the future, which could result in the risk that investments in ASICs specifically designed for certain calculations would become wasted if they could not adapt. Additionally, creating many types of ASICs for each specific calculation is not realistic in terms of cost and production efficiency. It seems unlikely that hyperscalers will concentrate the majority of their investments in specific ASICs at such a stage, and the superiority of NVIDIA, which is a GPGPU, is expected to continue for the time being.
The following is a description of the risks of ASICs and the superiority of GPUs as asked in Chat GPT.
————
The development and mass production of ASICs is a highly advanced project, featuring the following characteristics.
Development period and cost
• Period:
In general, it is said that it takes about 2 to 3 years from conceptual design to the start of mass production. Advanced processes (for example, 7nm or 5nm) are used in...

$NVIDIA (NVDA.US)$ It's right that Buffett advocated that I don't invest in something I don't understand. I think Buffett already doesn't understand AI. If it is an industrial revolution at the level of changing from horses to cars, there is a high possibility that it will be a serious loss of opportunity if you believe Buffett's words today. I also feel that Apple, which likes Buffett, seems to have changed its position quite a bit after AI penetrated the world. There is a high probability that Buffett won't be in this world around that time. In that sense, he may be the best perfect investor in the world in his life, but there is a growing possibility that a completely different world will unfold in the future, and in that sense, I think stories about geniuses and investors at the forefront of innovation rather than successive investors are overwhelmingly useful.

$NVIDIA (NVDA.US)$
As the shortage of high-quality and high-entropy data from the Internet progresses, scaling AI is facing revenue decline while the industry is shifting towards more abundant synthetic data.
Nvidia is strategically positioned by the Omniverse platform and the Cosmos world model that have the potential to infinitely generate high-entropy synthetic data essential for Reinforcement Learning.
RL-driven AI training enabled by Nvidia's simulation environments may dominate next-generation applications in Robotics, autonomous Autos, finance, Health Care, and could elevate Nvidia's valuation to 10 trillion dollars.
However, significant risks remain. Nvidia is facing intense competition from tech giants (Google, Amazon, Microsoft) and specialized startups, and there is a possibility that RL itself may stagnate or be replaced by new paradigms.

$NVIDIA (NVDA.US)$ A study was conducted on AI investments in major countries by Deep Research. In all countries, future AI investments are expected to expand significantly. As the USA is no longer the world's police and countries strengthen their national defense, AI investment will become a key factor, and without high-performance GPUs, it will be difficult to compete with other nations. The demand and importance of high-performance GPUs will continue to increase in the future.
USA
Federal government spending on AI-related initiatives was approximately 3.3 billion dollars in the fiscal year 2022.
• Private sector investment is expected to reach approximately 47.4 billion dollars and double in the next five years.
• Focus on defense, advanced research, and development of Datacenters.

China
• Through the advisory fund, approximately 940 billion dollars in funding has been投入.
• Significant investments in Smart City, military, and AI Semiconductors fields.
• An ongoing investment of several hundred billion dollars is expected with public and private collaboration.
European Index and UK.
• The EU is promoting a public-private investment plan of approximately 4 billion euros (about 4.3 billion dollars).
• The UK is approaching nearly 47 billion dollars in international investment projects.
• The aim is to realize high-performance computing infrastructure, data sharing, and a digital government.
Japan :
The government is strengthening the domestic AI Computer infrastructure with a scale of approximately 0.74 billion dollars....

$NVIDIA (NVDA.US)$ Future global AI infrastructure investments, including companies and nations, are publicly disclosed, and are expected to increase by 240 billion dollars in 2024 and 420 billion dollars or more in 2025, and even after that, global AI capital investment, including hyperscaler companies, is expected to increase, and even with low estimates, it is expected that it will be 500 billion dollars or more in 2026 and 600 billion or more in 2027. There is a very high possibility that NVIDIA, which currently occupies an exclusive position, can enjoy a lot with these trends. Since the overall market will expand even if the monopoly position falls slightly, it is expected that 2025 will have sales of 130 billion dollars, 2026 will be 180 billion dollars or more, and 2027 will grow drastically in the future. Moreover, the current stock price valuation of NVIDIA is by no means high. Opinions have suddenly emerged that the AI boom has ended this year, but it seems to be an overly pessimistic assumption that has arisen due to short-term market turmoil. They cite examples of past IT bubbles, etc., but they ignore growth forecasts based on AI investment plans etc. of each company and country that have already been announced...

$NVIDIA (NVDA.US)$ This is purely a personal opinion. At some point, I believe LLM will be personalized. The current LLM is focused on creating a single public model at a tremendous cost. Using the method of fine-tuning to retrain the public LLM to be usable with internal information requires the same enormous cost of initial training. Even major companies find it unrealistic at this stage to cover such vast retraining costs. Currently, there are mechanisms like RAG to meet the need for public LLMs to refer to corporate data for responses, but this is actually just a search function and is a transitional technology. However, with the evolution of software like Deepseek and hardware like GPUs, I can imagine a future where personalized LLMs will be widespread at the level of individual companies and even individuals, anticipating that the computational power required will be at a level incomparable to now. This direction is likely not something that will reverse, and while the timing is uncertain, it is expected to happen someday. In that case, personalized L...

$NVIDIA (NVDA.US)$ Institutions initiate a short sell using algorithms, causing fearful S investors to sell excessively, while behind the scenes, institutions are gradually accumulating Buy orders below. In the medium to long term, institutions cannot outperform the fundamentals, but in the short term, there is a relatively high probability of making a profit. Institutions gradually extract funds from general investors who have less information and are easily swayed by emotions. However, there is always a risk of incurring significant losses if taken too far. The situation becomes even more complicated when options are involved, not just Short Sell.

$NVIDIA (NVDA.US)$ The current quality of generative AI has not yet reached a truly useful level. At this stage, it is still too early for Commodityization. No one would pay to use such a thing. It is essential to fundamentally improve the quality of generative AI to create products or services that can drastically change the world, and for that, not only small-scale efficiency improvements but also a significant increase in parameters through brute-force scaling is necessary. The situation where there is still not enough GPU has not changed at all. The opinion that the growth of AI demand will slow down in the future is based on the premise that generative AI will stop innovating at the current level, which I believe is not the case.

$NVIDIA (NVDA.US)$ When looking at options trading today, it seems that a typical Gan Ltd mask squeeze occurred at 130. The next peak of calls is at 135, and even bigger at 140. From the perspective of options trading, institutions who have sold a large amount of calls at 140 as a risk hedge may need to buy a large amount of NVIDIA's physical stocks to offset the risk, which could lead to another similar Gan Ltd mask squeeze around 140 before the next quarterly earnings report this month. In that case, the stock price is likely to quickly reach 145.