Is it time for NVIDIA to buy the dip?
On November 12, NVIDIA announced that Jensen Huang resigned from his positions as President, CEO, and Director, effective immediately. After the U.S. stock market opened that day, NVIDIA's stock plummeted over 14%, hitting a low of $148.8, nearly a 40% drop from its October peak, with the market value evaporating by over $600 billion at one point.
Image source: Visual China
The founder's departure is the most direct trigger for this round of sell-off— the market is not just responding to a single risk but is re-evaluating the entire company. Meanwhile, when a leading AI computing power company sees its valuation drop to historical lows despite strong demand, discussions about whether to buy the dip are also heating up. If we shift our focus away from this personnel change, we may find that the real questions NVIDIA needs to address are whether its valuation, competitive moat, and demand have undergone irreversible changes.
The sell-off triggered by Jensen Huang's resignation has pushed NVIDIA's valuation to a rare position: approximately 12 times forward PE, the lowest level for the company's stock price in fifteen years. During the same period, the expected price-to-earnings ratio for the S&P 500 index is about 18 times. In other words, after a governance crisis, NVIDIA's valuation has become cheaper than the overall U.S. stock market.
This valuation level corresponds to a still strong business. For the fiscal year 2026 (ending January 2026), NVIDIA's revenue is projected to be $215.9 billion, a year-on-year increase of 65%, with data center revenue reaching $62.3 billion in a single quarter and a gross margin maintained at around 75%. In the second quarter of fiscal year 2027, revenue is expected to be $96.2 billion, with data center revenue at $89 billion, a year-on-year increase of 117%, an operating profit margin of 66.2%, and free cash flow reaching $21.3 billion.
Although several investment banks have lowered their target prices for NVIDIA following the Jensen Huang incident, most still maintain a buy or overweight rating.
Barclays lowered its target price from $275 to $210, maintaining an overweight rating, clearly stating, "We are reducing the valuation multiple by about 25% to reflect the reputation and transition risks in the coming months," with the current stock price corresponding to a potential upside of about 20.9%. Goldman Sachs maintains a buy rating, lowering its target price from $300 to $220; Citigroup lowered its target from $285 to $225, also maintaining a buy rating; Royal Bank of Canada lowered its target from $330 to $230. Truist is one of the few institutions that directly downgraded its rating from "buy" to "hold," cutting the target price to $174, which is just around the current stock price.
Clearly, the consensus among institutions is: the business is not bad, but the multiple given to the business needs to be reset. The 12 times forward PE has already largely completed this reset. The main support behind this is that the long-term narrative on the demand side has not been disproven.
At the 2026 GTC conference, Jensen Huang predicted that by 2027, the demand for AI chips from the Blackwell and Rubin systems would reach at least $1 trillion, directly doubling the previous forecast of $500 billion for 2026. Anthropic's latest disclosed annual recurring revenue (ARR) is still growing, and the total forecast for AI capital expenditures has not been revised down due to short-term fluctuations; only the slope of spending has shifted backward.
However, a cheap valuation does not mean that risks have disappeared. The risks NVIDIA currently faces may be more fundamental than "who will be the CEO."
First, the lock-in effect of CUDA is being eroded from the inference side.
CUDA is considered a competitive moat mainly because the cost of migrating models from NVIDIA GPUs to other chip architectures is high. However, Anthropic disclosed a significant advancement during its IPO roadshow: its model Claude Mythos Preview completed the migration of the core GPU kernel autonomously in just three days, with minimal human intervention, allowing Claude to run critical inference tasks on AMD's flagship GPU.
Silicon Valley investor Marcelo P. Lima's comment has been widely quoted: "NVIDIA's CUDA moat no longer exists." This comment may have some exaggeration, but the direction of the trend is clear: as large models begin to possess the ability to rewrite system software, the migration costs previously built at the expense of human engineering labor are being rapidly flattened by automation—CUDA will not collapse overnight, but the process of it shifting from "everyone must use it" to "it must be used on the training side, but not necessarily on the inference side" is already happening.
Secondly, the supply bottleneck NVIDIA faces is not in wafers, but in HBM and packaging.
Industry data shows that TSMC can cut over 400 million GPU dies in a year, but NVIDIA's actual annual delivery volume is only about 10 million. The real constraint lies in HBM: the global annual HBM capacity is about 3.75EB, with NVIDIA able to secure around 70% of it, and each GPU requires an average of 260GB, which can only support about 10 million GPUs in total. At the same time, packaging capacity is also tight, and adjustments in HBM4 supply have already led to potential delays in Rubin's mass production schedule. This means that even if demand is unlimited, NVIDIA's shipment volume in the medium term is constrained by a physical limit, rather than its design capabilities or customer orders.
Moreover, most importantly, what Jensen Huang left for NVIDIA is not only the organizational assets of CUDA and architecture accumulation but also three highly personalized elements: strategic judgment on the technology roadmap, supply chain negotiation capabilities with TSMC and SK Hynix, and direct communication channels with customers and the government. Interim CEO Colette Kress is a familiar face in the capital markets, having long served as CFO, but questions about who will make decisions on product roadmaps and maintain supply chain relationships remain unanswered until a formal successor is in place. This means that while NVIDIA's stock price is unlikely to collapse after "the departure of Jensen Huang," the premium the market was willing to pay for Jensen Huang will not automatically transfer to his successor.
Therefore, if we look at opportunities and risks together, NVIDIA's current position can perhaps be summarized in one sentence: the business is still there, but the "faith premium" given to the business is being systematically stripped away.
Indeed, the 12 times forward PE has absorbed a considerable degree of pessimistic expectations. If the long-term direction of AI capital expenditures remains unchanged, the demand for Blackwell and Rubin materializes, and the successor can stabilize supply chain and customer relationships, then the current price offers a favorable entry window. Barclays, Goldman Sachs, and Citigroup maintain buy ratings and provide potential upsides of over 20%, reflecting this judgment.
However, "cheap" alone still does not constitute a sufficient reason to buy. The loosening of the CUDA lock-in effect is a slow structural change that occurs over years, but its direction is difficult to reverse. The physical bottlenecks on the supply side in HBM and packaging limit the revenue ceiling NVIDIA can realize during demand surges. The uncertainty surrounding the successor means that the stripping away of the "founder premium" may not yet be over.
Bernstein bluntly stated in its latest research report: "If you see two ants in the kitchen, there are definitely more than two ants in the kitchen." It pointed out the potential chain of developments: from personal scandals to governance vacuums and business news, to government investigations, each link could trigger a new round of sell-offs. Each chain reaction brings NVIDIA's valuation closer from "Jensen Huang premium" to "ordinary semiconductor company premium."
Undoubtedly, what the market fears most is not bad news itself, but not knowing when or in what form bad news will appear. Until everything settles, "how much is NVIDIA worth without Jensen Huang" remains an open question.