AI computing power investment is undergoing a major shift! Market is betting big on the strong rise of ASIC.

As of Friday's closing of the US stock market, NVIDIA Corporation (NVDA.US) fell by nearly 3%, facing the significant risk of losing its milestone $4 trillion market value for the first time in nearly two months. The "AI chip leader" NVIDIA Corporation fell by nearly 5% during Friday's trading session, and based on the weekly indicators, NVIDIA Corporation's stock price has been falling for four consecutive weeks. For a stock with a market value exceeding $4 trillion, a nearly 5% drop is considered a huge decline, highlighting the market's panic triggered by the disappointing US non-farm payroll data, as well as the increasing pressure from the surge in the AI ASIC market size brought about by Broadcom Inc. Additionally, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR stock ADR price has shown a strong upward trend recently, ranking second only to Broadcom Inc. in the performance of chip stocks in the US market. In the chip industry chain, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR is considered to be the "eternal god" (YYDS), as the demand for AI GPUs and AI ASICs cannot do without Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR. With decades of experience in chip manufacturing technology and being at the forefront of chip manufacturing technology innovation and improvement, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR has dominated the majority of global chip orders for a long time, especially for chip manufacturing orders at the most advanced processes of 5nm and below. Looking back, although the extremely weak non-farm payroll data has lowered expectations for a "soft landing" in the US economy, the market's expectations of a rate cut by the Federal Reserve have significantly increased, with the possibility of a 50 basis point rate cut in September. This is a positive factor for mega-cap growth stocks like NVIDIA Corporation. Mark Malek, Chief Investment Officer of Siebert Financial, stated that a 50 basis point move "will bring tailwinds to the stock market." "This will undoubtedly boost mega-cap growth stocks and give investors the green light to take on more risks," he said. Therefore, the strength and future prospects of Broadcom Inc. bring a core logic of market panic, as the emergence of AI ASIC products from Broadcom Inc., jointly created by Alphabet Inc. Class C, Meta, and Microsoft Corporation, among other tech giants, have led to a rapid rise in revenue. This development has caused top Wall Street investment institutions and individual investors who have been bullish on NVIDIA Corporation's stock price and performance trends to become worried. Due to the strong growth potential of AI ASIC products and the possibility of them soon taking market share away from AI GPUs, their long-term growth expectations for NVIDIA Corporation's performance have started to adjust. AI ASIC and NVIDIA Corporation's AI GPU are two completely different technological paths in the field of AI chips. Currently, they are to a large extent competitors in the market. Especially for AI giants in cloud computing and AI leadership like OpenAI, AI ASICs have a very clear cost-effectiveness and efficiency advantage in the AI training/inference field. The $10 billion "super order" brought in by OpenAI has boosted Broadcom Inc.'s AI ASIC demand, and the explosive growth in Broadcom Inc.'s performance has led Wall Street analysts to adjust down their previously high expectations for NVIDIA Corporation's performance from 2026 to 2030. Broadcom Inc.'s strong rise has forced them to moderately adjust their previously high growth expectations, which is the core logic behind NVIDIA Corporation's stock price plummet on Friday. The status of NVIDIA Corporation in the "$4 trillion club" is under pressure, with a significant retraction in stock price in recent weeks. According to sources close to the matter, Broadcom Inc. is helping OpenAI design and produce a custom artificial intelligence accelerator chip (i.e. AI ASIC chip). The two companies plan to start delivering the first batch of AI chips in this series as early as next year, which will directly place NVIDIA Corporation, the absolute leader in AI computing infrastructure, under long-term competition pressure from Broadcom Inc. Broadcom Inc.'s semiconductor revenue related to AI infrastructure in the third quarter was approximately $5.2 billion, with a year-on-year growth rate of 63%, higher than the Wall Street average expectation of $5.11 billion. Broadcom Inc.'s management expects this category's revenue to reach approximately $6.2 billion in the fourth quarter, an increase of nearly 70% year-on-year, higher than analysts' previous expectation of approximately $5.82 billion. Before the earnings release, the market had very high expectations for Broadcom Inc.'s performance and future outlook data, which significantly boosted investors' bullish sentiment towards Broadcom Inc. and the entire AI computing industry chain. After receiving over $10 billion in AI infrastructure orders from an unnamed new super-large-scale customer (later revealed to be OpenAI), Broadcom Inc. CEO Hock Tan stated during the earnings call that the company expects stronger growth in AI-related revenue for the 2026 fiscal year than previously expected. In a previous earnings call, Tan had indicated that the prospects for AI-related revenue in 2026 would follow a growth trajectory similar to this year an expected growth rate of approximately 50% to 60%. In the Chinese stock market, Horizon is indisputably the hottest stock, betting on the ASIC route. It has surged by as much as 95% so far this year. Goldman Sachs Group, Inc., a Wall Street giant, recently raised its target price for Horizon, the "top AI chip leader in China" and the leader in "domestic chip substitution," for the second time in just a week. In its latest report released on September 1, Goldman Sachs Group, Inc. raised Horizon's 12-month target price from RMB 1835 to 2104, an increase of 14.7%, and maintained its "buy" rating. The new target price implies a 41% increase from the closing price on August 29. Horizon has recently shown a resonance between stock price and performance, highlighting the intense investment and "domestic chip substitution" trends in the Chinese stock market. In terms of performance, in the first half of 2025, Horizon's revenue reached 2.881 billion yuan, a year-on-year increase of 4347.82%, with a net profit of 1.038 billion yuan, compared to a loss of 530 million yuan in the same period last year. The logic of the bullish market for the AI computing industry chain remains strong, but funding is beginning to focus entirely on ASICs. It is undeniable that Broadcom Inc.'s strong performance and future prospects have reinforced the long-term bullish narrative for the AI computing sector. However, the core of the global net long position in capital is shifting from NVIDIA Corporation's AI GPU chain to the AI ASIC chain. With its absolute leadership in chip-to-chip communication and high-speed data transfer, Broadcom Inc. has emerged as the most important player in the field of custom AI ASIC chips for AI infrastructure in recent years. For example, Broadcom Inc. is a core player in the development of Alphabet Inc. Class C's data center server AI chipTPU AI acceleration chip. Aside from chip design, Broadcom Inc. also provides Alphabet Inc. Class C with critical chip-to-chip communication intellectual property and is responsible for the manufacturing, testing, and packaging of the new chips, thus providing essential support for Alphabet Inc. Class C's expansion into new AI data center territories. As US tech giants continue to invest heavily in artificial intelligence, the biggest beneficiaries include not only NVIDIA Corporation but also other AI ASIC giants such as Broadcom Inc., Marvell Technology, Inc., and Taiwan Semiconductor Manufacturing Co., Ltd. from Taiwan. Companies like Microsoft Corporation, Amazon.com, Inc., Alphabet Inc. Class C, and Meta, as well as the AI leader OpenAI, are all working with Broadcom Inc. or other ASIC giants to constantly update and iterate AI ASIC chips, mainly for large-scale deployment in AI inference endpoints. As a result, the AI ASIC's market share expansion is expected to be significantly stronger than that of AI GPUs, moving towards a more balanced market share rather than the current dominance of NVIDIA Corporation's AI GPUoccupying up to 90% of the AI chip market. The chip is expected to use an advanced 3nm process technology in collaboration with Broadcom Inc. to start mass production in the second half of 2025. It is estimated that Ironwood will bring in approximately $10 billion in revenue for Broadcom Inc. within the next 6-7 months. The cost-effectiveness advantage of AI ASICs is a major advantage for giants like Alphabet Inc. Class C, Microsoft Corporation, and OpenAI over NVIDIA Corporation. Meta's internal AI ASICMTIA v2 has shown in third-party GEMM/inference test models that it consumes less power compared to NVIDIA Corporation's H100 but has similar throughput (about 5.5-5.7 times relative to T4). With H100 priced at over $30k and an expected cost of $2-3k for MTIA v2, the cost-effectiveness of MTIA far surpasses that of NVIDIA Corporation. Therefore, although AI ASICs cannot completely replace NVIDIA Corporation, their market share is expected to expand significantly, rather than the current situation where NVIDIA Corporation's AI GPU dominates. Especially in the actual AI data center computing infrastructure configuration, a combination of AI ASICs and NVIDIA Corporation's AI GPUs (GPU for training/exploration and ASIC for scalable inference/partial training) significantly improves energy efficiency and reduces TCO. In standardized mainstream inference and partial training scenarios (especially continuous long-tail training/fine-tuning), custom AI ASICs demonstrate significantly better unit throughput cost/energy consumption compared to GPU solutions. However, in rapid exploration, cutting-edge large model training, and multi-modal experimentation, NVIDIA Corporation's AI GPUs remain dominant. Therefore, in current AI engineering practices, tech giants are increasingly inclined to adopt a "ASIC for normal workload and GPU for exploration/peak model development" hybrid architecture to minimize TCO. Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR exemplifies what it means to be "classic and timeless"! Since 2023, the booming demand for AI GPUs, and the recent explosive demand for AI ASICs, cannot do without Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR. Currently, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR is the world's largest contract chip manufacturer, and as the frenzy of AI continues unabated and continues to sweep the globe, its customers like NVIDIA Corporation, AMD, Broadcom Inc., and other chip giants are benefiting from the rapidly growing demand for AI chips, the core infrastructure of AI. These chip giants have seen a surge in chip manufacturing orders from Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR, which, in turn, has led to Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's performance consistently exceeding strong growth expectations since last year. This is a key logic support for the continued rise of Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's stock price in the Taiwan Stock Exchange and the US ADR market this year. Leveraging decades of experience in chip manufacturing technology and being at the forefront of chip manufacturing technology innovation and improvement, having pioneered the FinFET era and ushering in the 2nm GAA era, as well as leading the world in advanced manufacturing processes and packaging technology, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR has long dominated the vast majority of global chip foundry orders, especially for advanced processes of 5nm and below. Furthermore, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR is currently taking nearly all of the high-end chip packaging orders for 5nm and below processes with its leading 2.5D and 3D chiplet advanced packaging technology, and its advanced packaging capacity is inadequately meeting demand. The supply of NVIDIA Corporation's Blackwell AI GPUs since the end of last year remains constrained by Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's CoWoS packaging capacity at the 2.5D level. Currently, Apple Inc., AMD, NVIDIA Corporation, and Broadcom Inc., among other chip giants, are switching to Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's 3D advanced packaging capacity, which may further drive Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's advanced packaging capacity to be insufficient to meet demand. The latest performance of Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR shows that the booming demand for AI computing power is driving a 61% increase in net profit for Q2, and Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR expects sales revenue in 2025 to increase by around 30% in USD terms, higher than the previous expectation of "close to the middle of the 20% range." This growth is mainly driven by the continuous surge in AI chip orders using advanced 3nm and 5nm process technologies. As the demand for AI computing power remains incredibly strong, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR is actively expanding its backend capacity to increase the actual output of CoWoS advanced packaging, primarily for NVIDIA Corporation's AI GPU capacity, demonstrating the company's high confidence in the incredibly strong demand for AI chips to continue until 2026.