According to recent reports, global AI chip leader NVIDIA
Tech giants like Microsoft, Amazon, Google, and Meta, the parent company of Facebook, have been aggressively investing to meet the widespread industry's massive appetite for AI computing resources. This intense competition for AI infrastructure has further escalated the “AI arms race,” with substantial capital being funneled into acquiring AI GPU and AI ASIC compute clusters. Moreover, these tech leaders, alongside AI pioneers like OpenAI, are developing groundbreaking AI technologies that aim to match or exceed human intelligence.
Underpinned by the robust stock price surges and strong performance of major tech firms such as NVIDIA, Meta, Google, Oracle, TSMC, and Broadcom, an unprecedented tide of AI investment is sweeping through the U.S. and global stock markets, propelling both the S&P 500 and MSCI World Index to reach historical highs since April. As earnings season approaches, capitalizing on TSMC's and ASML’s unexpected stellar results further solidifies the narrative of a “long-term bull market” in AI, reassuring bulls of tech stocks.
This indicates that the ongoing “super bull” market in the global AI supply chain, led by NVIDIA, TSMC, Broadcom, and Micron Technology, is far from over. The AI supply chain remains the most favored sector for global capital in the upcoming period. This year, NVIDIA’s stock price has repeatedly hit new records, boasting a year-to-date increase of 40%, with a current market cap of approximately $4.45 trillion. Analysts on Wall Street believe NVIDIA will be the primary beneficiary of the soon-to-be-unleashed trillion-dollar wave of AI spending, with expectations for its market cap to potentially reach $5 trillion.
TSMC’s U.S. ADR has also reached new heights, with a year-to-date increase surpassing 50%, outpacing NVIDIA. In the chip supply chain, TSMC has become a “forever god” (YYDS), epitomizing timeless excellence—both the in-demand AI GPU and AI ASIC heavily rely on TSMC’s advanced manufacturing capabilities. TSMC’s decades-long expertise in semiconductor production positions it at the forefront of technological advancements, allowing it to dominate most global chip manufacturing orders, especially for the most advanced nodes of 5nm and below.
NVIDIA and TSMC’s commitment to domestic chips aims to significantly strengthen the U.S. chip supply chain. In a recent blog post, NVIDIA remarked that this move “greatly enhances the U.S. chip supply chain” and aims to bring back AI technology stacks—comprising both hardware and software—to ensure the U.S. maintains its leadership in the global AI era.
This action aligns perfectly with former President Donald Trump's policy to bolster American leadership in technology and manufacturing, resonating with his enduring vision of “bringing chip manufacturing back to the U.S.” Their collaboration to ramp up chip capacity domestically could also be a significant response to Trump’s administration for waiving tariffs on chip products—without expanding chip production in the U.S., chips and consumer electronics may not remain exempt from tariffs.
“Bringing chip manufacturing back to the U.S.” has been a focus of Trump's manufacturing ambitions since he took office in 2016, viewing the resurgence of high-end manufacturing as a significant political achievement. According to data from the Semiconductor Industry Association (SIA), the U.S. share of global semiconductor manufacturing capacity has plunged from a high of 37% in 1990 to merely 12% after 2020. Hence, both Biden and Trump prioritized this initiative as a core task during their terms.
NVIDIA has stated that TSMC's factory in Arizona will start producing advanced process chips including 2nm, 3nm, and 4nm nodes, as well as the cutting-edge A16 chip—which represents the pinnacle of semiconductor technology. These advanced chips will be critical for AI, communications, and high-performance computing applications.
In March, TSMC announced its expansion plans in the U.S., aiming to establish “three major chip manufacturing factories, two advanced packaging plants, and one advanced chip R&D center,” with an investment increase to $165 billion. This strategy also signifies the introduction of advanced packaging capabilities such as CoWoS, SoIC, and InFO in the U.S., forming a robust “closed loop” for domestic AI chips.
As the world’s leading advanced-process chip manufacturer, TSMC reported record overall performance last Thursday while unexpectedly raising its forecast, significantly enhancing investor optimism regarding expansion expectations tied to AI infrastructure spending. TSMC also raised its lower limit of capital expenditure targets for this year, shedding light on the booming demand for AI chips.
Recent earnings reports reveal TSMC’s net profit for Q3 reached an all-time high, exceeding expectations by 39%. Significantly, TSMC raised its revenue growth forecast for 2025 to a midpoint of 30%. Regarding the critical CoWoS advanced packaging capacity vital for NVIDIA's AI GPUs and Broadcom’s AI ASICs, TSMC has emphasized a focus on ramping up capacity significantly by 2026, indicating that advanced packaging capabilities linked to AI remain exceedingly tight.
The unstoppable tide of AI is pushing the leading players’ stock prices to new heights. Recently, partnerships among leaders in AI application like OpenAI and major U.S. chip giants—such as NVIDIA, AMD, and Broadcom—have led to several large-scale AI infrastructure agreements worth billions, accelerating the establishment of colossal AI data centers worldwide, including the “Interstellar Gateway” AI project.
Despite some experienced analysts, recalling the bursting of the “Internet bubble,” labeling the scale of AI processing orders by OpenAI and these chip giants as potentially exacerbating AI bubble inflation through “circular financing,” analysts at Goldman Sachs maintain that while valuations in the tech sector have risen, they have not yet reached historically bubble levels. Goldman highlights that the current tech stock growth is fundamentally driven, rather than rooted in irrational speculation associated with unprofitable companies like during the Internet bubble era. Notably, AI leaders such as NVIDIA and Google boast exceptionally robust balance sheets, marking a crucial difference from previous bubble characteristics.
Recently, the soaring prices of global DRAM and NAND high-performance storage products, along with OpenAI's over $1 trillion in AI infrastructure deals and TSMC's staggering results, which led to a raised revenue growth forecast for 2025 to a midpoint of 30%, have collectively reinforced the bullish narrative surrounding AI infrastructure sectors, including AI GPUs, ASICs, HBM, data center SSDs, liquid cooling systems, and core power equipment.
The AI computing demand driven by generative AI applications and intelligence agents is described as a “vast ocean of stars,” anticipated to drive exponential growth in the AI computing infrastructure market. Huang Renxun considers “AI inference systems” to be the largest source of future revenue for NVIDIA.
Top Wall Street firms such as Morgan Stanley, Citigroup, Loop Capital, and Wedbush regard the global AI infrastructure investment surge driven by AI computing hardware as just emerging, likely reaching $2 trillion to $3 trillion amidst an unprecedented “AI computing demand storm.”
Leading institutions like Cantor Fitzgerald, HSBC, and Morgan Stanley firmly believe that NVIDIA will remain the principal beneficiary of this trillion-dollar AI spending wave, reason why Cantor and HSBC recently raised their price targets for NVIDIA significantly.
HSBC raised its rating for NVIDIA's stock from “hold” to “buy,” strongly countering the prevailing market theories concerning an “AI bubble.” Even more remarkable, HSBC significantly increased its target price for NVIDIA from $200 to $320, the highest on Wall Street, suggesting an astounding 80% potential upside for NVIDIA's stock price. If achieved, NVIDIA's market cap will soar to an astonishing approximate $8 trillion, compared to its current market cap of around $4.40 trillion, which has consistently held the top position globally. With NVIDIA's stock price reaching record highs this year, it has climbed 40% thus far in 2025.
Analysts from Cantor Fitzgerald emphasize that the rapid and widespread adoption of generative AI applications substantiates that this AI wave is not a bubble in any sense. “Over the past 12 months, global large recommendation systems have embraced generative AI. Searches have transitioned to generative AI. Social media has shifted to generative AI.” “User-generated innovative content, AI-driven advertising recommendation engines—everything has moved from traditional machine learning to generative AI. Based solely on the large-scale transition from traditional computing to generative AI, NVIDIA forecasts up to $2 trillion in capital expenditure. We are absolutely not in a bubble; rather, the market is beginning to recognize how ‘high-quality’ AI can lead to significantly positive investment returns,” Cantor Fitzgerald stated in a bullish report projecting NVIDIA’s stock price to reach $300.