Author: Aki Wu Talks Blockchain
At the end of October 2025, Nvidia's stock price reached a new all-time high, with its market capitalization surpassing the $5 trillion mark, making it the first company in the world to cross this threshold. Since the emergence of ChatGPT at the end of 2022, Nvidia's stock price has increased more than 12 times. The AI revolution has not only driven the S&P 500 index to new highs but has also sparked discussions about a technology valuation bubble. Today, Nvidia's market capitalization has even exceeded the total size of the entire cryptocurrency market. In terms of global GDP ranking, Nvidia's market capitalization is second only to the United States and China. It is remarkable that this superstar of the AI era also had a "honeymoon period" in the cryptocurrency field. This article will review Nvidia's tumultuous history with the cryptocurrency mining industry and why it chose to withdraw and shift its focus to its core AI business.
Crypto Bull Market Frenzy: Gaming Graphics Cards Transform into "Money Printing Machines"
Looking back at Nvidia's development history is also a legend of the continuous evolution of technological narratives. Founded in 1993, Nvidia started by inventing the GPU (Graphics Processing Unit) and rode the wave of the PC gaming boom in the late 1990s. Nvidia's GeForce series graphics cards were a huge success, and the company quickly rose to become a graphics card giant. However, when the gaming market gradually saturated and growth slowed, Nvidia also faced the dilemma of unsold inventory. Fortunately, opportunity always favors the prepared—a major turning point was the cryptocurrency boom.. In 2017, the prices of cryptocurrencies such as Bitcoin and Ethereum soared, sparking a "mining" craze. Because GPUs are ideally suited for parallel computing in mining, miners worldwide scrambled for graphics cards, turning them into money-printing machines with supply falling short of demand and prices skyrocketing. Nvidia became one of the biggest winners behind this crypto bull market, reaping huge profits from card sales. Starting in the second half of 2020, the crypto market rebounded after a two-year hiatus. Bitcoin prices surged from less than $15,000 in the middle of the year to a peak of over $60,000 in early 2021, while Ethereum rose from a few hundred dollars to over $2,000. This new wave of soaring cryptocurrency prices reignited the GPU mining frenzy. Miners went on a buying spree of the new generation of GeForce RTX 30 series graphics cards, leading to a shortage of high-end cards originally intended for gamers, and the market once again fell into a frenzy of "supply falling short of demand." When Nvidia's RTX 30 series graphics cards were released, they surprised gamers with their high performance and cost-effectiveness. However, with the surge in Ethereum mining profits, the actual selling price of these graphics cards was pushed to outrageous levels. The RTX 3060, with a suggested retail price of 2,499 RMB, was being sold for 5,499 RMB, and the flagship RTX 3090 was even being priced at nearly 20,000 RMB. However, the continued shortage of graphics cards has brought the conflict between gamers and miners to the forefront. Nvidia chose a "dual-track" approach, lowering the Ethereum hash rate for its GeForce cards (starting with the RTX 3060) aimed at gamers. However, this was later discovered to be a smokescreen. In reality, miners discovered that by plugging the RTX 3060 with a "dummy HDMI" adapter, they perceived other graphics cards as also acting as display adapters, thus bypassing the hash rate limitations in multi-GPU scenarios and achieving full-speed mining.

Andreas showed this demo on his Twitter
On the other hand, the Cryptocurrency Mining Processor (CMP) series was launched specifically for miners, attempting to achieve "diversion".
On the other hand, the Cryptocurrency Mining Processor (CMP) series was launched specifically for miners, attempting to achieve "diversion".

Andreas showed this demo on his Twitter ...p>On the other hand, the Crypto The official blog stated explicitly that day: "GeForce is born for gamers, CMP is born for professional mining." CMP would eliminate display output, use open baffles to improve airflow in densely packed mining racks, and lower peak voltage/frequency for stable energy efficiency. However, precisely because CMP lacked display output and had a short warranty, it was more difficult for miners to exit the market. GeForce, on the other hand, could be used for mining and refurbished for resale to struggling gamers, offering better residual value and liquidity. Therefore, the project ultimately generated a lot of hype but little substance, fading from public view. According to Nvidia's financial report, in the first fiscal quarter of 2021, graphics cards used for mining accounted for a quarter of the quarter's shipments, with cryptocurrency-specific chips (CMP series) generating $155 million in sales. Fueled by the crypto boom, Nvidia's revenue soared to $26.9 billion in 2021, a 61% increase from the previous year, and its market capitalization briefly surpassed $800 billion. However, this success was short-lived. On May 21, 2021, the Financial Stability and Development Committee of the State Council of China announced a crackdown on Bitcoin mining and trading. Subsequently, mining farms in Xinjiang, Qinghai, Sichuan, and other regions were shut down, bringing the mining industry to a rapid halt. Between that month and the following month, Bitcoin's hashrate and price both came under pressure, forcing miners to relocate or liquidate their equipment. By September 24, the People's Bank of China and several other departments issued a joint notice classifying all virtual currency-related transactions as illegal financial activities and proposing a nationwide "orderly withdrawal from the mining industry," further closing the policy loopholes. For those in the Huaqiangbei mining machine industry, the cycle of boom and bust is nothing new. Those who experienced the mining machine "crash" of early 2018 still vividly remember it; some withdrew from the market in despair, but a few persevered and weathered the storm, investing their unsold mining machines into their own mining farms, waiting for the next market boom. As it turned out, the bull market of 2020-2021 allowed those who held on to turn their fortunes around once again. In September 2022, a landmark event occurred in the crypto industry: the Ethereum blockchain completed its "merge" upgrade, transitioning from a Proof-of-Work (PoW) mechanism to a Proof-of-Stake (PoS) mechanism, eliminating the need for a large number of GPUs to participate in mining. This marks the end of the long-standing era of GPU mining. Lacking the specific needs of cryptocurrency miners, the global graphics card market cooled rapidly, directly impacting Nvidia's performance. In the third quarter of 2022, Nvidia's revenue declined by 17% year-on-year to $5.93 billion, with net profit at only $680 million, a 72% year-on-year decrease. Nvidia's stock price once fell to around $165 in 2022, nearly halved from its peak, and the former cryptocurrency boom instantly became a burden on its performance. Drawing a Line: Nvidia's Breakup with the Mining Industry Faced with the frenzy in the mining industry and complaints from gamers, as well as the problems brought about by cyclical profitability, Nvidia gradually realized that it had to find a balance in the cryptocurrency mining wave and, at the right time, "draw a line" with it. Amid concerns about a potential bubble arising from soaring cryptocurrency prices, the company also faced challenges regarding financial compliance. An investigation by the U.S. Securities and Exchange Commission (SEC) later found that Nvidia had failed to adequately disclose the contribution of cryptocurrency mining to its gaming graphics card revenue growth for two consecutive quarters in fiscal year 2018. This was deemed improper disclosure. In May 2022, Nvidia agreed to settle with the SEC and pay a $5.5 million fine. This incident forced Nvidia to re-evaluate its delicate relationship with the crypto industry; while the crypto mining boom brought substantial profits, its volatility and regulatory risks could potentially damage the company's reputation and performance. After Ethereum transitioned to PoS in 2022, GPU mining demand plummeted, and Nvidia's gaming graphics card business quickly returned to normal supply and demand. Jensen Huang has also repeatedly emphasized that the company's future growth will primarily come from areas such as artificial intelligence, data centers, and autonomous driving, rather than relying on speculative businesses like cryptocurrencies. It can be said that after experiencing the highs and lows of the "mining card craze," Nvidia decisively distanced itself from this highly volatile industry and invested more resources in the broader and more socially valuable AI computing landscape. Meanwhile, Nvidia's latest Inception program website for AI startups explicitly lists "unqualified organization types," including "companies related to cryptocurrencies." This clearly shows Nvidia's intention to distance itself from its past crypto partners. So, after fully embracing the AI industry, will Nvidia's chip business still have any overlap with the crypto industry? On the surface, since Ethereum ended its "mining era," the connection between GPUs and traditional crypto mining has significantly weakened. Major cryptocurrencies like Bitcoin already use dedicated ASIC miners, and GPUs are no longer the highly sought-after "golden goose" for crypto miners as they once were. However, the two fields are not entirely unconnected, and new points of convergence are emerging in different forms. Some companies that were once deeply involved in cryptocurrency mining are shifting their business focus to AI computing power services, becoming new customers of Nvidia. Moreover, traditional Bitcoin mining companies are also exploring using surplus electricity and site resources to undertake AI computing tasks. Some large mining companies have recently replaced some of their equipment with GPU hardware for training AI models, believing that AI training offers a more stable and reliable source of income compared to the volatile cryptocurrency mining industry. The person who made the most money in the AI gold rush—Nvidia, which sells "shovels" (November 2022) In [Month], OpenAI's ChatGPT burst onto the scene, causing a huge sensation globally for large-scale AI models. For NVIDIA, this was undoubtedly another once-in-a-century opportunity. The world suddenly realized that driving these computationally intensive AI monsters was inseparable from NVIDIA's GPU hardware support. Following ChatGPT's explosive popularity, major technology companies and startups flocked to the "large-scale model" track, leading to an explosive growth in the computing power required to train AI models. NVIDIA astutely grasped this essence: regardless of technological changes, computing power will always be the fundamental currency of the digital world. Currently, NVIDIA holds over 90% of the large-scale model training chip market share. GPUs such as the A100, H100, and the new generation Blackwell/H200 have become essential for AI... The industry standard for accelerated computing. Due to demand far exceeding supply, Nvidia possesses extraordinary pricing power and profit margins in high-end AI chips. Goldman Sachs predicts that from 2025 to 2027, the capital expenditures of just the five major cloud service providers—Amazon, Meta, Google, Microsoft, and Oracle—are expected to approach $1.4 trillion, nearly tripling compared to the previous three years. This massive investment has laid the foundation for Nvidia's sky-high market capitalization. However, the AI field once experienced a shockwave of "cost reduction and efficiency improvement"—the explosive popularity of the open-source large-scale model DeepSeek. The DeepSeek project claims to have trained a DeepSeek V3 model with performance comparable to GPT-4 at an extremely low cost of approximately $5.576 million, and subsequently released the R1 model with ultra-low inference costs. At the time, the industry was in an uproar, with many predicting Nvidia's demise, arguing that the emergence of such low-cost AI models meant that small and medium-sized enterprises could deploy large models with fewer GPUs, potentially impacting the demand for Nvidia's high-end GPUs. "Will the demand for AI computing power be replaced by an efficiency revolution?" This became a hot topic of discussion. Affected by this expectation, Nvidia's stock price plummeted, closing down about 17%, wiping out approximately $589 billion in market capitalization in a single day (considered one of the largest single-day market capitalization losses in US stock market history). However, just a few months later, it proved that this concern was short-sighted. DeepSeek did not reduce the demand for computing power; instead, it triggered a new surge in computing power demand. Its technical approach essentially achieved "computing power parity"—through algorithmic innovation and model distillation, it significantly lowered the hardware threshold for large models, making AI applications more affordable for more institutions and enterprises. On the surface, because of improved model efficiency, it seems that "less computing power is needed"; but in reality, the DeepSeek phenomenon greatly popularized AI applications, causing computing power demand to grow exponentially. A large number of enterprises rushed to connect to DeepSeek, triggering a wave of AI applications, with inference computing quickly becoming the new main force in computing power consumption. This precisely confirms the famous... The "Jeves Paradox"—increased technological efficiency ironically accelerates resource consumption. DeepSeek lowered the barrier to entry for AI, leading to a surge in applications, which in turn exacerbated the shortage of computing resources. As it turns out, every new AI model's emergence often translates into a surge of new GPU orders. The more AI innovation Nvidia generates, the stronger it becomes, a fact once again validated in the DeepSeek controversy. Nvidia's February 2025 financial report showed that its data center business significantly exceeded expectations. At a deeper level, DeepSeek's success is not a threat to Nvidia; rather, it demonstrates that "cost reduction and efficiency improvement" leads to larger-scale application expansion, thereby driving up total computing power demand. This time, DeepSeek has become the new fuel for Nvidia's computing empire. As AI pioneer Andrew Ng said, "AI is the new electricity." In the age of electricity, computing power providers like Nvidia undoubtedly play the role of power companies. Through massive data centers and GPU clusters, they continuously supply "energy" to various industries, driving intelligent transformation. This is also the core logic behind Nvidia's market capitalization soaring from $1 trillion to $5 trillion in two years—a qualitative leap in global demand for AI computing power, with tech giants from various countries vying to invest in computing power in an arms race-like manner. After reaching a market capitalization of $5 trillion, Nvidia's influence and scale have even surpassed the economic influence of many national governments. Nvidia is no longer just a "graphics card" manufacturer that makes game graphics smoother; it has transformed into the fuel of the AI era, becoming the undisputed leader in this gold rush. "The Shovel Seller." As Nvidia's size increased, the myth of its employees becoming incredibly wealthy spread throughout the industry, with many employees holding stock worth more than their annual salary. Nvidia itself has achieved one leap forward after another by constantly "telling" new technological narratives. Gaming graphics cards opened the first door, the mining boom provided a second wave of growth, and AI propelled Nvidia to its true peak.