727 FinanceFor Nvidia, the key going forward is to sustain customers' willingness to purchase and to ensure that the revenue and profit growth driven by its new-generation products meets market expectations.
Written by: Li Hailun
Source: Tencent Technology
After several years of an AI computing boom, global chip giant Nvidia is undergoing a sharp valuation correction.
As of the close on October 16, 2026, Nvidia's stock price had cumulatively corrected about 12.02% from the all-time high set on October 6. The company's total market capitalization fell from about $5.76 trillion to $5.16 trillion, a loss of about $600 billion (about RMB 4 trillion) over nine trading sessions. The entire semiconductor sector also came under pressure in tandem, with the Philadelphia Semiconductor Index suffering its longest streak of eight consecutive trading days of declines since 2008.
Selling pressure subsequently spread to multiple parts of the AI industry chain. According to statistics from U.S. financial media CNBC, in just four trading days from Monday to Thursday of this week (October 12 to October 15), the combined market capitalization of the world's top 20 AI and semiconductor companies shrank by about $1 trillion; as deleveraging pressure persisted on Friday, volatility across the sector widened further. From hardware supply chain companies such as Samsung Electronics and SK Hynix to emerging computing rental providers that rely on debt financing to build data centers, all were affected to varying degrees.
This market turmoil triggered by the correction in tech stocks such as Nvidia has also brought a question back to the surface: over the past few years, the rapid expansion of AI infrastructure investment has driven up valuations across Nvidia and the entire computing industry chain. But as investment scale continues to expand, can the commercialization revenue of AI companies support such massive capital expenditures?
As the market begins to pay closer attention to the actual returns on AI investment, the growth expectations that drove Nvidia's stock price higher over the past few years are also undergoing new scrutiny.
Nvidia's current stock correction has been affected by multiple factors, including changes in downstream AI companies' commercialization expectations, rising macroeconomic financing costs, and deleveraging in funding conditions, reflecting that capital markets' assessment of the AI industry is changing.
**The market has begun to re-examine whether the revenue growth of AI companies can continue to support the massive capital expenditures of the upstream computing industry chain.** As one of the important drivers of Nvidia's computing demand, OpenAI's revenue performance has become a window through which the market observes progress in AI commercialization.
According to disclosures by the Financial Times, as of the end of September 2026, OpenAI's annualized revenue run rate was close to $50 billion, about $18 billion short of the previously rumored $68 billion to $70 billion.
This discrepancy has sparked discussion about the statistical basis of AI companies' revenue and the actual pace of commercialization. Especially against the backdrop of continued expansion of AI infrastructure investment, if downstream companies' revenue growth falls short of expectations, the market will worry about whether they can maintain the current scale of computing procurement and capital expenditures in the future, which in turn affects judgments about growth in demand for Nvidia GPUs.
However, the annualized revenue, cloud service transaction value, and channel revenue-sharing metrics disclosed by different AI companies differ in basis and cannot be directly compared in a simple way. Bloomberg subsequently reported that, driven by growth in enterprise business, OpenAI's annualized revenue by year-end is still expected to reach $70 billion.
Nevertheless, the related discussion reflects that the market's focus is changing: investors are beginning to place greater emphasis on downstream AI companies' actual revenue and investment returns, and on whether such revenue can support the long-term growth of the upstream computing industry chain. This shift in expectations has also become one of the factors behind the valuation pressure facing Nvidia and related computing stocks.
**Rising macroeconomic financing costs constitute the second source of pressure.** U.S. inflation data coming in above expectations has cooled rate-cut expectations. Data center construction is highly dependent on heavy capital investment in chips, servers, and power infrastructure, and high interest rates directly push up funding costs.
Take computing rental provider CoreWeave as an example. Its business expansion requires continuous investment in data centers, chips, and supporting infrastructure, and its accumulated debt and lease liabilities also put considerable pressure on its cash flow. As the market raises its requirements for returns on AI infrastructure investment, the financing capacity, debt-servicing ability, and cash flow returns of such highly indebted companies are being scrutinized more strictly by investors.
**In addition, rumors about memory chip supply and demand have added further variables.** Market discussion of technological breakthroughs by Chinese memory manufacturers has raised concerns about the future global memory supply landscape. Although high-bandwidth memory (HBM) and ordinary DRAM differ significantly in technical specifications and application scenarios, and medium- to long-term supply-demand forecasts are highly uncertain, this still disturbed short-term market sentiment.
Overall, the perspective from which capital markets assess the AI industry is changing. In the past, investors were more willing to pay high premiums for the rapid growth of computing demand; now, revenue growth rates, financing costs, and return on investment are being scrutinized more strictly.
In addition to changes in expectations for the fundamentals of the AI industry, concentrated reduction of risk exposure by institutional funds has further amplified the decline in tech stocks this round.
According to a client report released by Goldman Sachs Prime Brokerage on October 16, global long-short funds had net sold AI concept stocks for six consecutive trading days prior to that. For funds that use leverage and hold diversified positions, when market volatility intensifies and position risk rises, risk control mechanisms are often triggered, forcing cuts to both long and short positions.
This involves the de-grossing operations commonly used by hedge funds.
Over the past few years, as AI concept stocks continued to rise, some institutions amplified investment returns through leveraged trading and used long-short portfolios to hedge risk. When related stocks fall in a concentrated manner and market volatility intensifies, funds may, out of risk management needs, simultaneously reduce long and short positions to lower overall portfolio risk.
This means that even if some investors remain optimistic about Nvidia's long-term growth prospects, they may still sell shares because they need to reduce leverage or adjust their portfolios. When multiple institutions take similar actions within a short period, concentrated selling may further depress the stock price, forming a short-term negative loop.
In addition to the concentrated withdrawal of institutional funds, Nvidia's own stock trading activity has also drawn market attention.
Public filing records show that some Nvidia executives and directors sold company shares during the preceding period. For example, Nvidia director Mark Stevens was among the investors with the largest scale of insider stock sales in the AI sector in the third quarter of 2026. SEC filing data show that Stevens made multiple large stock sales in September alone.
However, insider stock sales mostly involve pre-established 10b5-1 trading plans, personal asset allocation, or tax arrangements, and share reductions alone cannot be used to conclude that management is bearish on the company's prospects.
In contrast to the risk-averse withdrawal of short-term funds, some investors have begun to reassess the valuation attractiveness of Nvidia at current levels.
Along with the stock price decline, Nvidia's previously elevated valuation premium has been somewhat digested. However, forward valuation metrics are highly dependent on the basis of future earnings forecasts, and if future performance growth slows, valuation multiples may still have room for adjustment.
The current market is not simply one-directionally bullish or bearish: some funds worry that AI capital expenditures are too large and the payback period too long; other investors believe that as long as earnings growth can continue, the stock correction is an opportunity to reposition. The disagreement between the two sides lies in whether actual AI demand in the coming years can live up to today's market expectations.
The stock correction has not hindered Nvidia's pace in advancing its next-generation computing platform. As AI applications accelerate their extension from model training to large-scale inference, the market's focus has shifted from the performance of a single GPU to an overall computing framework composed of chips, CPUs, memory, networking, and software working in concert.
As the core of this strategy, Nvidia is advancing the Vera Rubin computing platform, integrating the Vera CPU, Rubin GPU, and high-speed interconnect technology to improve the computing efficiency and energy efficiency of large AI systems and meet the requirements of hyperscale clusters for computing power, memory bandwidth, and system coordination.
This reflects the evolution of Nvidia's competitive moat: its advantages have long extended beyond the hardware performance of a single GPU to the CUDA software ecosystem, network interconnection, system integration, and supporting tools for large-scale deployment. For companies building large data centers, the energy efficiency and deployment efficiency of the entire system are more critical than the metrics of any single chip.
Technological advantages and product iteration provide the foundation for Nvidia to sustain growth, but whether the new platform can generate actual performance contribution still depends on mass production and delivery progress, the scale of customer deployment, and ultimately return on investment. Even if total computing demand continues to grow, if downstream customers find that additional capital investment is unlikely to bring matching business revenue, the pace of future procurement and deployment may face adjustment.
For Nvidia, the test in the next stage is to prove that stronger computing performance can translate into measurable economic returns for customers. As AI investment enters deeper waters, training and inference costs, energy consumption, and actual business monetization capability will jointly determine downstream customers' future procurement decisions.Facing market volatility, NVIDIA CEO Jensen Huang expressed confidence in the long-term prospects of AI while attending an event in Berlin, Germany. He said: "The reason is very simple: AI is transformative, and it will have a very positive impact on every industry and every country. There is no doubt that this transformation will continue, and we are only at the starting point."
These remarks continue his consistent judgment: AI is not a short-lived hardware upgrade, but a technological transformation that is reshaping production methods across industries. From data centers to enterprise software, and from scientific research to industrial automation, there is still broad room for the expansion of AI applications.
For NVIDIA, the key going forward is to maintain customers' willingness to purchase and to ensure that the revenue and profit growth brought by its new-generation products meets market expectations. Whether NVIDIA's growth can be sustained will ultimately depend on whether downstream customers can convert computing power investment into actual returns and continue to pay for a new round of expansion.