large_cover-nvda 4 trillion.jpeg

medium_cover-nvda 4 trillion.jpeg

small_cover-nvda 4 trillion.jpeg

thumbnail_cover-nvda 4 trillion.jpeg

cover-nvda 4 trillion.jpeg



## Highlight:
NVIDIA benefits from the overall ecological Network Effect: CUDA, installation foundation, system-level integration, optimization, etc. Each advantage complements and strengthens each other, forming a strong technical barrier.
NVIDIA has achieved unexpected revenue for six consecutive quarters in the past. Surprisingly, despite raising expectations every quarter, it still finds ways to surprise investors with a further significant lead.
In the past, Internet Tech Giants considered growth more when deciding to use certain technologies to build business barriers. However, for AI giants, if they don't keep up, their business foundation may be lost. Today, their decision-making strategy is no longer based on calculating profits and returns, but on fear of absolute losses. Buying NVIDIA chips is essentially buying insurance, with no flexibility.
In the long run, tech giants will continue to seek high-performance GPU sources or internal solutions outside of NVIDIA to break free from their dependence on it. Most likely, these efforts will gradually weaken but cannot replace NVIDIA's dominant position in the AI field.


## Introduction

From time to time, a star company will emerge, attracting attention due to its skyrocketing stock price over the years, but also causing huge controversy due to intense fluctuations. This time it is NVIDIA, a global chip giant that powers AI. Its stock price has soared nearly 50 times in five years and is currently tied with Microsoft and Apple for the top three in global market value.

In a recent interview by a podcast named The Crossing, RockFlow founder Vakee expressed her opinion on whether Nvidia's stock price is in a bubble. She does not believe that Nvidia is in a bubble. A Forward P/E (Forward Price-To-Earnings Ratio) valuation of 40 times is a reasonable range for such a leading company.

The recent pullback of semiconductor stocks, including the overall pullback caused by excessive gains in the US stock market, is in line with normal market changes. In addition, the stronger expectation of interest rate cuts at that time strengthened risk appetite, and some funds flowed from top companies to more aggressive small-cap stocks, which is also a normal sector rotation. Furthermore, the strengthening of US export restrictions and geopolitical tensions a few weeks ago were also inducements for the decline of chip stocks represented by NVIDIA, but these are all normal market reactions.

As for the violent fluctuations in the US stock market in the past one or two weeks, it is mainly due to a combination of factors such as weak US labor market data, the lack of surprises from financial report season giants, the pullback of yen carry trading, and potential conflicts in the Middle East. However, setting aside these external factors, as far as the target of NVIDIA is concerned, we believe that with the long-term optimism towards AI (which is likely to be the biggest opportunity for change in our generation), the absolute leading stock in this industry still does not have a bubble.

NVIDIA's performance in the past year has been impressive: it has a market share of about 90% in the AI chip market, annual revenue exceeding $60 billion, and a Net Profit ratio exceeding 50%. In the past five years, its compound annual revenue growth rate has reached as high as 64%, far ahead of many S & P 500 companies.

Regarding the development history and investment value of this company, RockFlow's research team made a detailed analysis in an article at the beginning of last year - https://mp.weixin.qq.com/s/6HD8UjKIH6c-BJx3bgERiw. In this article, we hope to answer why the market has not yet paid enough attention to NVIDIA's true Competitive Edge, and why we believe that NVIDIA is not only a great company, but also an investment target with high potential returns.


## What does NVIDIA Competitive Edge rely on besides CUDA?


Investors who are bullish on Nvidia believe that they are betting on its future. After considering Nvidia's higher expected growth, it is not more expensive than other tech giants.

However, there is a problem: the farther the market predicts returns, the more uncertain their predictions become. Microsoft and Apple are mature companies that rely on existing stable customer bases to make money. Nvidia, on the other hand, serves a newer but more promising market, so investors are much more divided about Nvidia's prospects than Microsoft and Apple.

For a long time, NVIDIA has been considered a top producer of gaming graphics cards. With the rise of cryptocurrency mining, GPUs, which serve as the core of graphics cards, have become increasingly popular. NVIDIA's GPUs are highly optimized for "parallel processing" - decomposing computationally difficult problems and assigning them to thousands of processor cores on the GPU at the same time, thus solving problems faster and more efficiently than traditional computing methods.

In addition to designing the most advanced GPUs on the market, NVIDIA has also created a programming model and parallel computing platform - Computing Unified Device Architecture (CUDA), which has become an industry standard and makes it easier for developers to use the features of NVIDIA GPUs.

![nvda CUDA system](https://web-cdn.rockflow.tech/broker-cms/nvda_CUDA_system_a882ef25c8.JPEG)

But what NVIDIA relies on is not just the CUDA platform that everyone talks about today. The RockFlow research team believes that NVIDIA benefits from the overall ecological Network Effect: CUDA, installation foundation, system-level integration, optimization, etc. It also actively chooses advanced solutions in bandwidth and network to improve computing performance. Each advantage complements and strengthens each other, forming a strong technical barrier.

First, let's look at the installation base. CUDA's decades-long leading advantage means that it benefits from a strong Network Effect - its huge installation base motivates frameworks and developers to target it, thereby attracting more users to adopt it. NVIDIA has a large number of users in fields such as gaming, professional visualization, and data center. The huge user base provides NVIDIA with a continuous source of revenue and product feedback, and this scale effect also enables NVIDIA to continue investing in research and development and maintain technological leadership.

Secondly, it is the system-level integration capability. NVIDIA not only provides GPU hardware, but also provides supporting software stacks. It can be said that from drivers to power builders, and then to optimization libraries, it has already formed a complete ecosystem. This vertical integration enables NVIDIA to optimize at the system level, providing better performance and User Experience.

As for the optimization level, NVIDIA has conducted deep optimization in both hardware and software. In terms of hardware, it continuously improves the GPU architecture to enhance performance and energy efficiency; in terms of software, it fully unleashes the potential of hardware through driver and library optimization. System-level optimization, such as multi-GPU collaborative work and GPU direct memory access technologies, further improves overall performance.

In order to solve bandwidth and network problems, NVIDIA has made many attempts. It has launched a series of technologies to improve the efficiency of data transmission between GPUs, CPUs, and GPUs, the most important of which is NVLink. NVLink directly connects multiple GPUs with high bandwidth, significantly improving AI computing efficiency. This has enabled NVIDIA to maintain a strong position in the fields of autonomous driving and AI computing. In addition, betting on InfiniBand, acquiring Mellanox, and approaching the Ethernet platform through NVIDIA Spectrum-X are also NVIDIA's active layout in the network field.

The RockFlow research team believes that for today's NVIDIA, the huge installation base provides the driving force and data for its continuous optimization of GPU design, while system-level integration and optimization enhance user stickiness and expand the installation base. Coupled with the continuous iteration of many advanced solutions for bandwidth and network issues, the virtuous cycle between them keeps NVIDIA in a leading position in the GPU and accelerated computing fields.

Therefore, even though competitors AMD and Intel have launched similar GPU products with price advantages, NVIDIA still holds an absolute dominant position in the AI chip market.

It is estimated that NVIDIA has firmly occupied more than 90% of the data center GPU market share in the past 7 years. In 2023, its share will reach 98%. The operation of all large data centers and large model training rely on GPUs developed by NVIDIA.

![data center GPU market](https://web-cdn.rockflow.tech/broker-cms/data_center_GPU_market_235e960e18.jpeg)

In the long run, it may be difficult for NVIDIA to fully maintain its current market share, but with the continuous growth of the data center GPU and other AI chip markets, its Competitive Edge will ensure that it receives the vast majority of orders. NVIDIA is expected to always occupy an important position in this new industrial revolution and maintain benign and healthy long-term growth.


## NVIDIA's performance miracle: expectations are repeatedly raised, but still bring surprises


More than two months ago, NVIDIA announced its Quarter 1 performance, and its stock price soared 10% the next day. Since then, its market value has skyrocketed, once replacing Microsoft as the world's most valuable company.

Last quarter, NVIDIA's specific revenue situation is as follows:

![nvda Q1 FY25 income](https://web-cdn.rockflow.tech/broker-cms/nvda_Q1_FY_25_income_7c761ec116.jpeg)

Revenue grew 18% quarter-over-quarter to $26 billion (exceeding expectations by $1.50 billion). Data center revenue grew 23% quarter-over-quarter to $22.60 billion, making it the largest and fastest-growing segment.

How exaggerated is its growth rate? It can be seen from the following chart:

![nvda revenue breakdown](https://web-cdn.rockflow.tech/broker-cms/nvda_revenue_breakdown_1f7a3fafe7.jpeg)

In fact, NVIDIA has achieved higher-than-expected revenue for six consecutive quarters in the past. Surprisingly, despite raising expectations every quarter, it can still find ways to further surprise investors with a significant lead. At the same time, the profit margin of NVIDIA's chip business is also significantly improving.

At the subsequent financial report meeting, Huang Renxun stated:

The next industrial revolution has already begun. Major companies and countries are collaborating with NVIDIA to turn traditional data centers worth trillions of dollars into accelerated computing and establish new types of data centers - AI factories to produce a new commodity - AI.

From the perspective of the two core stages of AI system operation: training - AI learns from large amounts of data, develops intelligence and pattern recognition. Currently, NVIDIA's powerful GPU dominates this stage; inference - AI applies its knowledge to real-world tasks and decisions. Despite facing more intense competition, NVIDIA is making significant progress.

Inference workloads contributed about 40% of NVIDIA's data center revenue in the past year. The current market consensus is that with more and more generative AI applications emerging, inference is expected to become a huge market, bringing considerable investment returns to NVIDIA customers.

Currently, NVIDIA mainly divides customers into three categories.
Cloud as a Service Provider (CSP) contributes nearly half of Data center revenue. All giants (Amazon, Microsoft, Google, etc.) are customers of NVIDIA.
B-end enterprises have driven strong continuous growth. Taking Tesla as an example, it expanded its training AI cluster to 35,000 H100 GPUs and used it for FSD V12.
C-end companies are also a key vertical class. Taking Meta as an example, its Llama 3 is trained on 24,000 H100 GPU clusters and is expected to use 240,000 GPUs and ten times the computing power of Llama 3 to train the next generation MultiModal Machine Learning Llama 4.

NVIDIA management also pointed out some important directions during the conference call.

Regarding Data Center, they believe that "as generative AI enters more and more 2C internet applications, we expect to see continued growth opportunities, as the scale of inference will expand with the complexity of the model and the increase in the number of users and queries per user, thereby driving more demand for AI computing."

In addition, NVIDIA is very optimistic about the concept of "sovereign AI": "Sovereign AI refers to a country's ability to produce AI using its own infrastructure, data, labor, and commercial networks. The importance of AI has attracted the attention of every country, and major powers will pay more attention to the control of AI technology. We believe that sovereign AI will bring billions of dollars in revenue this year."

Regarding the impact of US export restrictions, the new generation H200, and Blackwell architecture, Huang Renxun has also made corresponding disclosures. The consecutive rise in NVIDIA's stock price in the following weeks fully demonstrates the market's recognition of its views and optimism about the long-term development of the AI wave.


## The biggest challenge is still self-developed by giants


As mentioned earlier, NVIDIA's Network Effect built around the CUDA ecosystem is an excellent example of a commercial company shaping a complete ecosystem. NVIDIA's huge success is not only due to the excellent performance of its chips, but also because it has firmly established itself at the center of the generative AI wave with the entire hardware and software network layout.

However, at the same time, its strategy of bundling basic software with chips has also attracted strong criticism from customers and competitors, and even regulatory agencies are repeatedly launching Anti-Trust investigations due to its high market share.

Currently, NVIDIA is facing competition from AMD and other chip manufacturers (including Qualcomm, Intel, etc.). These companies are essentially chip designers, and they almost all use the same outsourcing company to manufacture chips - TSMC.

AMD and Intel have both launched their own Data Center GPUs, aiming to regain market share from Nvidia's H100/H200. Intel has launched the Gaudi3 AI acceleration chip, while AMD has launched the MI300 series. In 2024, it may be the moment when Nvidia hands over a small part of the market share to AMD for the first time, and Intel is also expected to regain a small part of the market share.

Just look at one statistic to understand why competitors are so determined: According to the latest quarterly financial reports, Microsoft, Google, and Meta's total capital expenditures for the second quarter of 2024 exceeded $40 billion - most of which was spent on AI.

![big tech capital spending](https://web-cdn.rockflow.tech/broker-cms/big_tech_capital_spending_b285970193.jpeg)

With so many giants choosing to invest heavily, is it possible that their decision-makers will one day be forced by internal and external pressures to calculate the ROI of AI costs? They find it not cost-effective, so they start to slow down AI investment?

Our answer is almost impossible. In the past, when Internet Tech Giants decided to use certain technologies and enhance certain capabilities to build business barriers, they considered more about growth - whether it could accelerate business development. But this time, generative AI has a big change. For these giants, if they don't keep up, their companies and business foundations may be lost. Therefore, the arms race of generative AI this time is essentially more like buying "medicine" logic for the procurement of NVIDIA chips. When a technology is the difference between life and death for a company, it will not think too much about the details.

This is not an exaggeration. For example, the search business may have been completely changed by AI. Whether it's Google or Baidu, their decision-making strategy today is no longer based on calculating profits and returns, but on fearing absolute losses. Buying NVIDIA chips is essentially buying insurance, and there is no flexibility at this point.

It is precisely because high-performance chips are so important that while giants spend huge capital expenditures, they are also determined to develop their own or find alternative solutions. Faced with NVIDIA's comprehensive CUDA ecosystem layout, other companies (in fact, almost all competitors) are trying to jointly develop open solutions to break NVIDIA's monopoly on AI software and hardware ecology.

Intel, Google, ARM, Qualcomm, Samsung, and other technology companies are collaborating to develop a new software suite that allows developers' code to run on any machine equipped with any chip. OpenAI is also working hard, releasing an open-source language that allows researchers without CUDA experience to write GPU code, and the open-source PyTorch Foundation incubated by Meta is also making similar attempts.

These companies are also striving to replace NVIDIA's proprietary hardware, including developing new solutions to connect multiple AI chips within and across servers. Companies such as Intel, Microsoft, Meta, AMD, and Broadcom hope to establish new industry standards for this crucial connectivity technology for modern data centers. The conflict between proprietary and open solutions is a bit like Apple and Google Android in the smartphone market, and as we have seen, both closed and open visions are enough to create winners.

In the long run, tech giants will continue to seek high-performance GPU sources or internal solutions outside of NVIDIA to break free from their dependence on NVIDIA. Most likely, these efforts will gradually weaken but cannot replace NVIDIA's dominant position in the AI field.


## Conclusion


In the past few years, NVIDIA's chips have pushed the company's profitability to new heights. The RockFlow research team has long been optimistic about NVIDIA. Previously, RockFlow founder Vakee was asked on the podcast The Crossing, "Are there any signals that NVIDIA may be attacked by competitors?" Her opinion was that there are no such signals.

NVIDIA's advantage today is not just a single product or the hardware advantage of a simple GPU. NVIDIA has built its core competitiveness in a complex ecosystem, so it is particularly difficult to break through. Its advantages include not only the CUDA platform, but also the current user base, the entire basic installation volume, the overall system integration capability, and the ability to continuously optimize. These advantages reinforce each other, so the overall competitive barrier will always be prominent.

Faced with potential problems in the ecosystem, NVIDIA can spare no effort to solve them, whether through acquisitions or investments. This continuous investment is constantly increasing its own barriers in hardware, system integration capabilities, or internet bandwidth issues. This virtuous cycle will make its advantageous position more stable.

Therefore, we believe it can still replicate its previous growth momentum and release greater value in the long run.


Why the market has not yet paid enough attention to NVIDIA's true Competitive Edge？and the reason NVIDIA is an investment target with high potential returns.

large_0 封面.jpeg

medium_0 封面.jpeg

small_0 封面.jpeg

thumbnail_0 封面.jpeg

cover-nvidia flash crash.jpeg




## Highlight:

1）DeepSeek claims to train models at one-tenth the cost of GPT-4, causing concerns about the collapse of computing power demand. However, "underwater demands" such as MultiModal Machine Learning models and real-time inference are gradually emerging, coupled with the acceleration of Model Iteration, which has instead pushed up the long-term demand for computing power in the market.

2）In addition to the explosive data given in previous financial reports, NVIDIA's true Competitive Edge is hidden in the ecological barrier built by 280 million lines of CUDA code. Its software stack advantage has formed an industry standard, and the migration cost is extremely high. Even if the hardware performance of competitors is close, the software ecosystem gap is still difficult to bridge.

3）Current data proves that although DeepSeek's breakthrough is significant, no tech giant has cut capital expenditures in computing and data centers. The upcoming Q4 2024 financial report may once again boost investor sentiment. It is expected that Nvidia's revenue in 2025 will remain extremely optimistic, and new highlights will be revealed at the GTC conference on March 17th (focusing on physical AI projects such as GB300, Rubin, and robots).


## Introduction

On January 27, 2025, the US stock market witnessed a historic scene: NVIDIA plummeted 17% in a single day, with a market value evaporation of $590 billion, setting a record for the largest single-day market value loss in the history of the US stock market.

The epicenter of this earthquake came from a Chinese AI company, DeepSeek. It claimed to have trained a model with equivalent performance at one-tenth of the cost of GPT-4, instantly tearing apart the consensus on the market's demand for AI computing power.

But what's even more dramatic is that in less than a month, NVIDIA's stock price has basically recovered all lost ground. The bulls who remain optimistic and choose bottom fishing have bought low again.

RockFlow's research team believes that behind this NVIDIA roller coaster ride, the market's overreaction to short-term sentiment is reflected, and the subsequent rebound verifies the irreplaceability of its core competitiveness and industry position. In August last year, we analyzed the true Competitive Edge of NVIDIA that the market has not yet paid enough attention to, and why we believe that NVIDIA is not only a great company, but also an investment target with high potential returns.

This article will deeply review why the 590 billion dollar flash crash was misjudged, as well as the long-term reasons and short-term catalysts for our optimism about NVIDIA.


## The essence of market panic: misjudging the technical path and demand structure


The release of DeepSeek R1 model is like a "bomb" thrown at the computing power market: it can be trained with only 2000 H800 GPUs and costs 6 million dollars, compared with GPT-4 which costs as much as 100 million dollars, and Llama 3 which uses 16000 H100. This order of magnitude of cost difference makes the market suddenly fall into the panic of "computing power demand collapse theory".

Some investors mistakenly believe that improving model efficiency will directly reduce reliance on NVIDIA GPUs, leading to short-term pressure on its stock price.

![1.png](https://web-cdn.rockflow.tech/broker-cms/1_67874f8df1.png)

However, there is a misjudgment point here - the illusion of cost accounting. In fact, 6 million only covers the cost of GPU leasing and electricity, while the real training cost also includes many implicit expenses such as data cleaning, algorithm engineer salaries, and experimental losses. Research firm Semianalysis pointed out that if the full cost is included, DeepSeek's actual expenditure may reach the level of 30 million dollars.

More importantly, market panics have overlooked the "iceberg structure" of AI computing power demand: the current Model Training demand is only visible and may only account for 10%, while the underwater part includes exponential growth in computing power consumption of MultiModal Machine Learning models (video/3D generation), rigid requirements for low-latency hardware for real-time inference needs (such as Tesla's autonomous driving), and repetitive training needs brought about by the acceleration of Model Iteration speed.

The real disruptive aspect of DeepSeek this time is that its open-source strategy (free and open R1 model) directly challenges OpenAI's subscription system ($20-200/month), causing turmoil in the AIaaS valuation system. In addition, it may also lead to the diversion of the long-tail market, especially for small and medium-sized developers who may turn to cost-effective solutions such as Trainium 2, but NVIDIA's core customers (large-scale data centers) are difficult to migrate due to CUDA ecosystem binding.

Overall, there are two core logics for misjudging short-term market panic:

On one hand, although model optimization (such as sparse computing and algorithm improvement) reduces training costs, the acceleration of large-scale Model Iteration (such as MultiModal Machine Learning and real-time learning) actually increases the total long-term computing power demand.

Secondly, underestimating the resilience of NVIDIA's demand structure: core customers are concentrated in cloud vendors (AWS, Azure) and leading AI companies (OpenAI, Meta), and its capital expenditure plan is centered on "infrastructure expansion" rather than simply pursuing single-task costs.


## NVIDIA's rebound driving force: fundamentals and ecological barriers


When the market is boiling for DeepSeek, don't forget the reality revealed by NVIDIA's Q3 2024 financial report: Hopper architecture chips (H200) are taking over data centers at the fastest speed in history. The core highlights of the last quarter's financial report include but are not limited to:

Data center revenue: $30.80 billion (+ 112% YoY), with Cloud as a Service Provider (CSP) contributing over 50%
Blackwell demand: The first batch of samples sparked a buying frenzy, Huang Renxun said "demand exceeds the most optimistic expectations"
Profit margin game: Although Blackwell's initial gross profit margin dropped to 75% in mass production, it will recover to the middle of 75% as production capacity climbs

![2.jpeg](https://web-cdn.rockflow.tech/broker-cms/2_05343675b6.jpeg)

Hopper's outstanding sales in the third quarter helped the Data Center Division's revenue grow 112% year-over-year to $30.80 billion. CFO Colette Kress said on the earnings call:

The sales of H200 have grown continuously to billions of dollars, making it the fastest-growing product in the company's history. The inference performance of H200 has increased by 2 times, and the TCO has increased by 50%. Cloud as a Service provider accounts for about half of the sales of Data Center, and the revenue has increased by more than 2 times year-on-year.

Now, AWS, CoreWeave, and Microsoft Azure all offer cloud instances based on NVIDIA H200, and Google Cloud and OCI (Oracle Cloud Infrastructure) are about to be launched. In addition to the significant growth of large CSPs, as North America, EMEA, and the Asia-Pacific region increase the construction of NVIDIA cloud instances and sovereign clouds, NVIDIA GPU regional cloud revenue has doubled year-on-year.

In addition to CSP, NVIDIA also receives more than double the revenue from consumer internet companies that purchase Hopper chips to support the training of next-generation AI models, multimodal and proxy AI, deep learning recommendation engines, generative AI, and content creation.

Kress stated during the financial report conference call that the company has delivered the first batch of Blackwell samples to customers in the third quarter. Blackwell is its latest architecture series, and due to its powerful performance, there is a high demand for it. Two or three years ago, it took weeks or even months to train large AI models on hardware, but Blackwell can greatly shorten this time. In the rapidly developing AI industry, the faster developers bring innovative products to market, the greater the chance of success.

Management forecasts total revenue of $37.50 billion for Q4. If this expectation is met, Q4 revenue will increase by 69.7% YoY, and full-year revenue in 2024 will increase by 111% compared to 2023, reaching $128.66 billion.

In addition to the explosive data given in previous financial reports, NVIDIA's true Competitive Edge is hidden in the ecological barrier built by 280 million lines of CUDA code. Its software stack advantages: CUDA and AI libraries (such as TensorRT) have formed industry standards, and the migration cost is extremely high. Even if competitors (such as AMD) have similar hardware performance, the software ecosystem gap is still difficult to bridge.

In addition, NVIDIA's system-level solutions (such as DGX SuperPOD, OVX servers, and other full-stack solutions) have been deeply bound to customer infrastructure, replacing the need to reconstruct the entire technology stack. This ecological control was further strengthened in the Blackwell era: its NVLink 5.0 Technology Implementation ultra-high-speed chip interconnection bandwidth is several times that of AMD MI300X. When the hardware performance gap exceeds a certain critical point, the cost-effectiveness loses its discussion significance.

NVIDIA was originally a hardware company that produced GPUs. However, it is developing into a company that provides end-to-end AI solutions. It provides software tools for customers to build chatbots, AI virtual assistants, and virtual agents. This is not just a chip provider, but a mature AI giant.

![3.png](https://web-cdn.rockflow.tech/broker-cms/3_bf5b479361.png)

It also emphasizes the total cost of ownership (TCO) of its entire AI infrastructure solution, rather than just focusing on manufacturing chips, which makes it harder for competitors who only sell cost-effective chips to compete. NVIDIA includes the entire hardware and software ecosystem, support, operating expenses, and the ability to quickly deploy AI solutions in its TCO calculation. Management hopes to tell customers, "Our AI chips may have higher upfront costs, but in the long run, the entire AI solution can save costs."

Therefore, although DeepSeek and the trend of reducing AI costs pose a certain threat to Nvidia, Nvidia does not need to worry too much.


## The "arms dealer" logic of AI competition remains unchanged


The current theme of tech giants is that business growth is still limited by the ability to establish data centers and provide computing power. None of them have cut capital expenditures in computing and data centers. Moreover, their growth in Cloud Services remains strong.

In the past 12 months, the top three CSPs have invested $186 billion to expand their computing power. In the new year, Meta expects capital expenditures of $600- 65 billion, Microsoft expects to invest about $80 billion to expand its data center, Amazon sets a tone of at least $100 billion, and Alphabet expects capital expenditures to reach $75 billion.

Meanwhile, the recently launched "Stargate" project claims to cost $500 billion and aims to promote the development of US AI. Oracle alone has identified 100 data centers for future development. In the field of physical AI, Tesla completed the assembly of 50,000 H100 clusters at the Texas Gigafactory in 2024, which will be used for autonomous driving. Musk said that the computing power of the Optimus humanoid robot needs to be increased by 10 times.

Morgan Stanley research director Vishwanath Tirupattur believes that although DeepSeek's breakthrough is significant, it will not lead to a collapse in capital expenditures of giants with significant influence in AI and related fields.

He mentioned that the sharp decline in computing costs in the 1990s provided a useful reference for this. At that time, the investment boom was mainly driven by two factors: the speed at which companies replaced depreciation capital, and the sustained and significant decline in computing capital prices relative to output prices. If the efficiency improvements brought by DeepSeek reflect similar phenomena, then the cost of AI capital may decrease and support the expenditure prospects of companies.

Tech giants continue to increase spending, undoubtedly providing strong support for computing companies such as Nvidia. Amazon CEO Andy Jassy pointed out in a conference call that most companies rely on Nvidia's chips for AI computing, and in the foreseeable future, Amazon will continue to maintain a cooperative relationship with Nvidia.

Not to mention, Huang has repeatedly stated last year that countries around the world are planning to build and operate their own AI infrastructure domestically, which will drive demand for NVIDIA products.

He emphasized in an interview with Bloomberg that countries such as India, Japan, France, and Canada are discussing the importance of investing in "sovereign AI capabilities". "Their natural resources - data - should be refined and produced for their countries. The recognition of sovereign AI capabilities is global."

Of course, it needs to be acknowledged that as US tech giants firmly invest heavily in the AI field, NVIDIA is also facing increasing uncertainties: for example, the two AI ASIC giants that have benefited the most recently are Broadcom and Mywell Technology.

With their technological leadership in chip-to-chip communication and high-speed data transmission, Broadcom and Mywell Technology have become the core forces in the AI ASIC market. Tech giants such as Microsoft, Amazon, Google, and Meta are collaborating with Broadcom or Mywell Technology to develop their own AI ASIC chips for massive inference AI computing deployment.

In addition, according to Reuters, OpenAI is advancing plans to reduce its reliance on NVIDIA by developing its first generation of internal artificial intelligence chips, opening up a new chapter in its chip supply. Sources said it will complete the design of its first internal chip in the coming months and plans to send it to TSMC for manufacturing. The latest news shows that OpenAI is expected to achieve its mass production goal by 2026.

However, these factors do not hinder NVIDIA's AI competition "arms dealer" logic, and the selling caused by NVIDIA believers shouting DeepSeek is a bottom fishing opportunity. They still believe in three supporting factors.

Currently, the market's confidence in Hopper and Blackwell chips is constantly increasing.
Although investors are under pressure for large training clusters, there are signs that large clusters are still under construction.
The inference market is expected to drive NVIDIA's growth for many years, and NVIDIA's position in the inference field is solid.

In the short term, the upcoming Q4 2024 financial report may once again boost investor sentiment. It is expected that NVIDIA will reconfirm Blackwell's execution; Data Center 2025 revenue will increase by more than 60% year-on-year; and NVIDIA will build momentum for the GTC conference on March 17th (focusing on physical AI projects such as GB300, Rubin, and robots).


## Conclusion


The RockFlow research team believes that the DeepSeek impact event will eventually be recorded in the history of technology, not because it changed the rules of the game, but because it verified the unshakable nature of the rules: in the infinite war of AI, computing power is not an option, but a must; it is not a cost item, but an asset item.

As the "infrastructure provider" of the global AI arms race, NVIDIA's irreplaceability will continue to strengthen during the industry expansion cycle. Despite the long-term disturbance caused by ASIC and self-developed chips, NVIDIA will still be the biggest winner in the AI computing power expansion cycle in the next 3-5 years.

When the market cheered for the "efficiency revolution", NVIDIA proved in a month that true Competitive Edge is never in the gross profit margin numbers of financial statements, but in the "import torch.cuda" command typed by every developer, in the roaring DGX SuperPOD of every Data center, and in the eternal desire of human beings to expand the boundaries of intelligence.

It deeply review why the 590 billion dollar flash crash was misjudged, as well as the long-term reasons and short-term catalysts for our optimism about NVIDIA.

NVIDIA 590 billion flash crash and rebound: How DeepSeek caused AI computing power misjudgment? 

2025

cover-nuclear energy.jpeg

An in-depth analysis of nuclear energy, and take stock of the investment value of old giants and emerging companies in the nuclear energy concept stocks.

The trillion-dollar nuclear energy track, which 10x stocks will be born?

Reasons for NVIDIA's Breakthrough 4 trillion




## Highlight:
The high performance expectations of the financial report and the lack of "willingness to protect the market" by the management team have suppressed the short-term confidence of the Pinduoduo bulls. However, Duan Yongping chose an advanced strategy of bottom fishing, "starting to put selling points".
Buffett and Duan Yongping are both proficient in using option strategies. The classic case of the former is his investment in Coca-Cola, while the latter has repeatedly practiced similar strategies in the technology sector, and both bottom fishing Pinduoduo and Alibaba have achieved considerable profits.
Options can express more diverse investment perspectives. Reasonable use of option tools not only reduces risk levels, but also provides a source of high annualized returns.
RockFlow's "Dip Sniper" product, based on the Sell Put strategy, helps users obtain Fixed Income in market fluctuations while preparing for low-priced positions.



## Introduction

On Monday, Pinduoduo fell nearly 30% overnight, the largest drop since its IPO, as lower-than-expected performance and warnings from company executives dented market confidence.

Goldman Sachs believes that the main reasons for this big drop are due to three aspects.

Before the financial report was released, the market had high expectations. Due to strong expected performance, Pinduoduo's stock price has risen by about 20% since the end of July, while the KWEB (China Overseas Internet ETF) index has fallen by 4% during the same period.

Pinduoduo's online marketing service growth has shown signs of normalization for the first time, with a year-on-year growth rate slowing to 29%, which is lower than expected. However, to be fair, Pinduoduo's performance is still good and significantly better than Alibaba and Kuaishou e-commerce.

The management's comments scared the market. Pinduoduo's management pointed out in the conference call that due to intensified competition and possible decline in long-term profitability to promote high-quality development, it is expected that future revenue growth will slow down and sacrifice short-term profits. As the overall investment is still in the stage, there will be no buybacks or dividends in the next few years.

With high expectations falling short and management showing no "willingness to protect the market", the short-term confidence of Pinduoduo's bulls was suppressed, and the stock price almost fell below the center of the earth.

Under such an astonishing decline, is it suitable for bottom fishing now? How to bottom fishing?

 Duan Yongping released a statement that night, stating that although he did not fully understand Pinduoduo's business model, he was bottom fishing and "starting to put selling points".



## Buffett and Duan Yongping love the steady bottom fishing strategy


In fact, this is not Duan Yongping's first time bottom fishing Pinduoduo.

At the end of 2021, Pinduoduo's quarterly report collapsed, and the stock price was nearly halved within a month. Duan Yongping posted a statement saying, "Although I still don't understand their business model, I feel that their support for agricultural products is still very meaningful. I am ready to invest in Pinduoduo again." The specific operation is exactly the same as this time-selling put. At that time, he also attached 7,000 settlement orders for Pinduoduo put options, with a potential transaction amount of about 35 million US dollars.

More than a month later, Pinduoduo's stock price has risen by more than 20% compared to Duan Yongping's bottom fishing. He posted again, "Pinduoduo is a typical example of earning but not earning much."

This bottom fishing, Duan Yongping is obviously continuing to implement the previous investment logic. He is expressing more of an attitude: not fully understanding, but optimistic.

The so-called "selling point put" actually refers to selling put options, that is, the counterparty pays a certain option fee to Duan Yongping, so as to have the right to sell Pinduoduo stocks in the US stock market to Duan Yongping at the agreed price on the agreed date. Simply put, this means that the target is moderately optimistic about the rise.

Duan Yongping's consistent investment principle is Value Investment. Value Investment favors companies and bottom fishing at low prices, which is his strategy. This time, Pinduoduo is no exception.
 
Before bottom fishing Pinduoduo, Duan Yongping also made a move on Tencent and Alibaba. Duan Yongping's first investment in Tencent was in 2018, with the logic behind it being "a long slope, thick snow". This is also Buffett's famous saying, "Life is like a snowball, the most important thing is to discover wet snow and a long mountain slope".

As the head of the Value Investment faction, Buffett's most famous case for the application of options is undoubtedly Coca-Cola.

On April 2, 1993, Marlboro cigarettes announced a 20% price cut. After the announcement, stock prices, including Coca-Cola, fell one after another, but Buffett believed it was a good stock based on his research and understanding of Coca-Cola.

He thought he could buy 3 million shares of Coca-Cola at about $35 per share and was confident that the price would not fall below that level. So he decided to sell the put option on Coca-Cola.

In April 1993, Buffett sold a put option on 3 million shares of Coca-Cola stock for $1.50 each. The expiration date of the option was December 17, 1993. At Berkshire Hathaway's annual shareholder meeting, Buffett stated that he was prepared to increase 2 million shares in a similar manner.

Buffett's only risk in selling the put option this time is that if Coca-Cola's stock price falls below $33.5 in December, he must buy Coca-Cola stock at $35 per share, but he is not worried because he originally wanted to buy Coca-Cola stock at $35.

As a result, by the end of 1993, with Coca-Cola's stock still above $35, the put buyer abandoned the contract, and Buffett made a net profit of $7.50 million.

Buffett's use of option strategies is quite proficient. He understands that selling put options can steadily harvest THETA, and based on his research and understanding of Coca-Cola, even call-overs after expiration are very cost-effective. It can be seen that Duan Yongping obviously recognizes and has practiced similar investment strategies in the technology stock field many times.


## Why is it more cost-effective to be an option seller than to buy stocks directly?


Call and Put are respectively bullish and put options in American options. Buying and selling represent different market views.

Selling put options means two possibilities:

If the expiration stock price is higher than the exercise price, the buyer gives up the call-over and the seller earns a net profit from the premium. If it is a small increase and the increase cannot cover the premium, then the ROI rate outperforms buying the stock. If there is a large increase, only the premium can be obtained, and higher returns cannot be obtained, which is equivalent to missing out.

If the expiration stock price is lower than the exercise price, the buyer will call over, and the seller's specific profit and loss depends on the difference between the premium and the price difference. If there is still a surplus after deducting the decline in the premium, the seller will still make a profit; if the decline swallows up the premium, the seller will lose money, but the loss is still relative to the market price of the stock, and the actual purchase of the corresponding stock is still the same.

Therefore, selling put will only result in a big loss when there is a big drop. As long as the stock price does not fall below the range covered by the premium, the seller will never lose money. 

Buffett and Duan Yongping, who are well-versed in investment, have repeatedly verified a truth with their decades of investment experience: options have become a very useful tool in Value Investment due to their many advantages.
 
First of all, Value Investment is a long-term investment based on the fundamentals of a company. Based on a long-term optimistic view of its future, profits are made through long-term holding and short-term volatility selling. Unless held for a long time and not sold, risks are inevitably associated with volatility. With the assistance of option tools, risks can be effectively reduced.
 
Secondly, options can express more diverse views. Duan Yongping firmly buys Apple in the US stock market, and has repeatedly used the strategy of selling put to express his bullish views on Alibaba and Pinduoduo, rather than directly buying a large amount of stocks.
 
The reason why Duan Yongping did not directly buy Pinduoduo's stock this time may be because he believed that there was not much room for the stock price to rise, but it has basically bottomed out. Therefore, he sold the put, earned the premium when it rose, and built a position at a low price when it fell. Among them, the application of option tools not only reduces the risk level, but also provides a source of high annualized returns.


## A more concise and convenient innovative option financial product - RockFlow [Dip Sniper](https://mobile.rockflow.ai/en/docs/options/better-buy)

"Dip Sniper" is an innovative options wealth management product launched by RockFlow. It is based on the Sell Put strategy to help users obtain Fixed Income in market fluctuations and prepare for low-priced positions.

Users can choose a suitable exercise price and expiration date for the "Dip Sniper" product, and receive a premium immediately after the subscription is completed. At the close of the expiration date, if the stock price falls below the exercise price, investors will buy stocks at the exercise price; if the Closing Price does not reach the exercise price, there is no need to buy stocks. Since 100% of the corresponding stock's cash is frozen in the user's account during the subscription, the "Dip Sniper" product will not have any position squaring risk.

Taking Pinduoduo as an example before the opening on Tuesday, when the stock price was 100 dollars, if you buy the "Dip Sniper" product that expires this Friday and has an exercise price of 90 dollars, you can immediately receive a premium of 30% annualized return. This money is Fixed Income, and regardless of how the stock price changes, the premium has already been received.

By the close of this Friday, if the Pinduoduo Closing Price does not fall below 90 dollars, the user's funds will be unfrozen and the transaction will be completed. If it falls below 90 dollars, the user will buy Pinduoduo stock at a price of 90 dollars and complete a low-priced position. The actual purchase price is lower than 90 dollars because they received an annualized premium of 30% at the beginning.

In contrast, ordinary Sell Put requires users to have a certain understanding of option trading, including exercise price, margin requirements, option contracts, etc. RockFlow's "Dip Sniper" product simplifies the trading process, simplifying complex option trading into simple and easy-to-understand financial products. You only need to choose the exercise price and expiration date (usually this Friday or next Friday) to efficiently complete option trading, express your investment ideas, and greatly reduce the participation threshold.

In summary, there are two advantages to Dip Sniper: firstly, it allows users to lock in opportunities to build positions at low prices during market fluctuations, and at the same time, they can obtain a Fixed Income regardless of how the market changes; secondly, it is an ideal choice to achieve stable returns during the financial reporting season and prepare for low-price positions.



Pinduoduo fell nearly 30%, and Duan Yongping chose an advanced strategy for bottom fishing, "starting to sell put".

Reasons for NVIDIA's Breakthrough 4 trillion

Related Content

NVIDIA 590 billion flash crash and rebound: How DeepSeek caused AI computing power misjudgment?

The trillion-dollar nuclear energy track, which 10x stocks will be born?