--- title: "NVIDIA (NVDA) vs AMD, for long-term investment I choose the latter." type: "Topics" locale: "zh-CN" url: "https://longbridge.com/zh-CN/topics/39718378.md" description: "In the AI chip race, NVIDIA (NVDA) is undoubtedly the most dazzling star at the moment, with a gross margin exceeding 70% and a net profit margin as high as 55%. The moat built by the CUDA ecosystem seems impregnable, almost becoming synonymous with "AI chips." However, if I had to choose between NVIDIA and AMD (Advanced Micro Devices), I might choose AMD. This is not to deny NVIDIA's excellence, but rather to see through a core logic: NVIDIA's long-term investment thesis is built on the assumption of "unchanged dominance," and this assumption..." datetime: "2026-04-03T11:41:59.000Z" locales: - [en](https://longbridge.com/en/topics/39718378.md) - [zh-CN](https://longbridge.com/zh-CN/topics/39718378.md) - [zh-HK](https://longbridge.com/zh-HK/topics/39718378.md) author: "[Sam港美股日记](https://longbridge.com/zh-CN/profiles/13275072.md)" --- > 支持的语言: [English](https://longbridge.com/en/topics/39718378.md) | [繁體中文](https://longbridge.com/zh-HK/topics/39718378.md) # NVIDIA (NVDA) vs AMD, for long-term investment I choose the latter. In the AI chip arena, NVIDIA (NVDA) is undoubtedly the most dazzling star at present, with a gross margin exceeding 70% and a net profit margin as high as 55%. The moat built by its CUDA ecosystem appears unbreakable, almost becoming synonymous with "AI chips." However, if I had to choose between NVIDIA and AMD (Advanced Micro Devices), I might choose AMD. This is not to deny NVIDIA's excellence, but to see through a core logic: **NVIDIA's long-term investment thesis is based on the assumption that its "dominance will remain unchanged long-term," and this assumption could be wrong in the fast-iterating tech industry**. # First, Look at the Gap: NVIDIA and AMD, Not Even in the Same League? Looking at fundamental data alone, the gap between NVIDIA and AMD is indeed staggering, even making one question if they belong to the same industry. In terms of scale, NVIDIA's market cap is approximately over $4 trillion, while AMD is less than $400 billion, a difference of more than 10 times. Over the past 12 months, NVIDIA's revenue was $215.9 billion, more than 6 times that of AMD ($34.6 billion). In terms of operating cash flow, NVIDIA's $102.7 billion is 15 times that of AMD ($7.7 billion). Looking at profitability, the gap is even more astonishing: ● Gross Margin: NVIDIA 71.07% vs AMD 52.49% ● Net Profit Margin: NVIDIA 55.60% vs AMD 12.51% ● Return on Equity: NVIDIA 101.49% vs AMD 7.08% Looking at these figures alone, NVIDIA's crushing advantage seems unshakeable. But investment is never just about current strength; it's about future potential. # NVIDIA's Confidence: The CUDA Ecosystem, Really Unshakeable? The core reason NVIDIA has long held its dominant position in AI chips is not how strong its chip design is, but that it has built an ecosystem that is "hard to exit"—CUDA. Simply put, CUDA is a parallel computing platform and programming model introduced by NVIDIA. Developers need CUDA training, and companies' AI workflows are deeply dependent on it. Once a company chooses CUDA, switching to another platform is not only technically cumbersome but also faces huge operational risks, like "checking into the Hotel California—you can check out any time you like, but you can never leave." This ecosystem barrier gives NVIDIA an extremely high customer lock-in rate and the confidence to charge high prices, thereby maintaining a net profit margin of over 50%. This is also the core reason the market is willing to give NVIDIA a $4.4 trillion market cap—everyone assumes this ecosystem advantage will persist long-term, and NVIDIA will continue to capture most of the profits in the AI industry. But the problem is: **No ecosystem barrier is permanent.** NVIDIA's advantage is different from TSMC's chip manufacturing or Coca-Cola's secret formula—the latter two rely on proprietary technology or trade secrets, creating an unbreakable barrier. CUDA's advantage is more of a "first-mover advantage" rather than an "unreplicable barrier." AMD is not sitting idly by; it is quietly researching and narrowing the gap, and there is no insurmountable structural obstacle between the two. To use a vivid analogy: NVIDIA is like the hottest architectural firm at the moment, designing the AI computing solutions everyone wants; AMD is like the competitor across the street. Although not as famous now, it has been studying the former's designs. Once it finds a breakthrough, it can catch up quickly. # AMD's Confidence: Under Lisa Su's Helm, It Has Created a Comeback Miracle Before Many are bearish on AMD, thinking it can never catch up to NVIDIA, but they overlook a key figure—AMD CEO Lisa Su, and the comeback miracle AMD has created before. Lisa Su is a legendary woman in the semiconductor industry. She is the first female recipient of the IEEE Robert N. Noyce Award, the "Nobel Prize" of semiconductors, recognized for her outstanding leadership in semiconductor products and business strategy, contributing significantly to the industry's development. When Lisa Su became AMD's CEO in 2014, the company was on the brink of bankruptcy, with its stock price below $2. Under her leadership, AMD's stock price soared to around $110, returning to the mainstream market, and flagship products like Ryzen and EPYC became strong competitors to Intel. Ten years ago, the gap between AMD and Intel was as huge as the current gap with NVIDIA: Intel was leading comprehensively in scale, profitability, and ecosystem, and AMD was widely considered to have "no chance of winning." But with precise strategic planning and strong execution, Lisa Su not only led AMD to catch up with Intel but even achieved an overtake—Intel's own bureaucratic bloat and short-sightedness also reduced some obstacles for AMD's comeback path. This is not to say AMD will replicate the miracle of overtaking Intel and replace NVIDIA's dominance, but Lisa Su's past track record is enough to prove: **Don't underestimate AMD's ability to catch up, and don't overestimate the long-term stability of an industry leader.** More importantly, this competition is not a zero-sum game—the AI chip market is like a rapidly expanding "toy pool," big enough to accommodate more than one winner. NVIDIA's current lead does not mean AMD has no room for survival and growth; AMD's rise does not mean NVIDIA will decline completely. # Core Logic: Choosing AMD, Not Because It's Stronger, But Because It's More "Valuable" As Chris Miller mentioned in "Chip War," leadership in the semiconductor industry rarely lasts as long as people expect; technological changes often dim the luster of former champions. For AMD, it doesn't need to surpass NVIDIA; as long as it can capture enough market share, it can achieve a significant increase in market value. And now, it happens to have such an opportunity: First, diversified customer demand. As AI computing demand surges, hyperscale data center operators are increasingly worried about the risk of over-reliance on NVIDIA and are starting to seek supplier diversification. Meanwhile, many customers don't need a "perfect" GPU; a "good enough," more economical alternative is sufficient to meet their needs—this is precisely AMD's positioning. Second, the continuous expansion of market size. According to an IDC report, the global AI server market size reached $125.1 billion in 2024, will increase to $158.7 billion in 2025, and is expected to reach $222.7 billion by 2028, indicating huge market growth potential. Even if AMD only captures 25% of that share, it can achieve leapfrog growth in revenue. We can do a simple calculation: Assuming the AI accelerator market grows to $300 billion in the next five years, and AMD captures a 25% share, its revenue would reach $75 billion. If the net profit margin is conservatively estimated at 25% (only half of NVIDIA's current level), net profit would be approximately $18.5 billion. At a 25x P/E ratio, the valuation would be about $460 billion—compared to the current $360 billion market cap, that's nearly a 30% upside. If the market size grows to $400 billion, or if AMD's net profit margin improves to 35%, its valuation could exceed $600 billion, offering even greater upside potential. When the entire market acknowledges NVIDIA's dominance, its stock price has already reflected this consensus. The potential to outperform the market average in the future has significantly diminished. As for AMD, despite the huge current gap, its underestimated potential, Lisa Su's leadership, and the opportunities brought by AI market expansion all make it a better choice for long-term investment. ### 相关股票 - [AMD (AMD.US)](https://longbridge.com/zh-CN/quote/AMD.US.md) - [NVIDIA (NVDA.US)](https://longbridge.com/zh-CN/quote/NVDA.US.md) - [Taiwan Semiconductor (TSM.US)](https://longbridge.com/zh-CN/quote/TSM.US.md) - [Intel (INTC.US)](https://longbridge.com/zh-CN/quote/INTC.US.md) - [GraniteShares 2x Long AMD Daily ETF (AMDL.US)](https://longbridge.com/zh-CN/quote/AMDL.US.md) - [GraniteShares 2x Long NVDA Daily ETF (NVDL.US)](https://longbridge.com/zh-CN/quote/NVDL.US.md) - [XL2CSOPNVDA (07788.HK)](https://longbridge.com/zh-CN/quote/07788.HK.md) - [XI2CSOPNVDA (07388.HK)](https://longbridge.com/zh-CN/quote/07388.HK.md) - [YieldMax NVDA Option Income Strategy ETF (NVDY.US)](https://longbridge.com/zh-CN/quote/NVDY.US.md) - [Direxion Daily NVDA Bear 1X ETF (NVDD.US)](https://longbridge.com/zh-CN/quote/NVDD.US.md) - [T-Rex 2X Long NVIDIA Daily Target ETF (NVDX.US)](https://longbridge.com/zh-CN/quote/NVDX.US.md) - [T-Rex 2X Inverse NVIDIA Daily Target ETF (NVDQ.US)](https://longbridge.com/zh-CN/quote/NVDQ.US.md) - [Alphabet - C (GOOG.US)](https://longbridge.com/zh-CN/quote/GOOG.US.md) - [Alphabet (GOOGL.US)](https://longbridge.com/zh-CN/quote/GOOGL.US.md) - [INTEL-T (04335.HK)](https://longbridge.com/zh-CN/quote/04335.HK.md)