While Nvidia's revenue surged 73% in the latest quarter and its earnings forecast for the next quarter was more optimistic than market expectations, investors are still worried: How long will the AI boom last?
This concern caused Nvidia's stock price to plummet 5.7% last Thursday, the biggest intraday drop in nearly three months.
What everyone is most concerned about now is: Will companies and governments continue to invest heavily in AI in the next few years? AI is shifting from "training models" to "practical application (reasoning)", can Nvidia maintain its technological leadership and market dominance?
In addition, whether some customers' large-scale purchase commitments can be fulfilled and the tight supply of memory chips have also made Nvidia's future growth a little more uncertain.
The market is full of doubts about whether "high growth can continue"
Last Thursday, Nvidia's stock price fell 5.7% in New York, reaching a low of $184.32, the biggest intraday drop since November 25 last year, and finally closed down 5.49% at $184.89.
Strangely, this decline occurred precisely after the company handed over a brilliant report card, with fourth-quarter revenue surging 73% year-on-year, and sales expectations for the next quarter were significantly higher than analysts' average forecasts.
But the market is "voting with their feet" - indicating that investors' confidence in Nvidia is shaking. In the past, relying on the explosive growth brought about by the AI boom, Nvidia once became the world's most valuable company.
But now, everyone is no longer just looking at the data in front of them, but more concerned: how long can this wave of AI investment frenzy last?
Analysts at British investment platform Hargreaves Lansdown pointed out that shareholders are most worried about two things: Will companies and governments continue to spend heavily on AI in the next few years? When AI shifts from "training large models" to "daily reasoning applications", can Nvidia still sit firmly in the leading position?
Nvidia executives are full of confidence
面对市场对AI热潮能否持续的担忧，英伟达CEO黄仁勋在周三的电话会上坚定回应：客户已经在用新买的算力赚钱了，所以他们会继续大手笔投资。
他说：“算力不是成本，而是直接带来增长和收入的工具。我非常有信心——客户的现金流正在变强。”
他还提到，行业已经迎来一个新拐点：“代理式AI”（agentic AI）开始真正落地，智能体能帮企业完成实际任务，这会进一步拉动对算力的需求。
不过，质疑声并未消失。因电影《大空头》走红的投资人迈克尔·伯里（Michael Burry）又泼了一盆冷水。他指出：英伟达目前的采购承诺高达952亿美元，而一年前才161亿美元；如果未来AI需求突然放缓，这么大的订单可能变成负担，带来库存或财务风险。
对此，英伟达CFO科莱特·克雷斯（Colette Kress）试图安抚市场。她承认，生产最先进的芯片（比如Blackwell系列）依然很难，但公司已经提前锁定了足够的零部件，能保障供应。
她还透露：最新的Blackwell芯片和下一代Rubin芯片的销售势头，比预期还要好；公司此前预测，这些产品到2026年底将带来5000亿美元收入；“我们已建立充足的库存和供应链承诺，足以覆盖到2027年的客户需求。”
|一边数钱，一边缺芯
尽管整体表现不错，但英伟达的最新财报也透露出一些隐忧。
先看亮点：公司预计第一财季营收约 780亿美元，高于分析师平均预期的 728亿美元（虽然略低于部分乐观预测的近800亿）；第四季度的实际营收、每股收益和毛利率也都超预期；数据中心业务（主要是AI芯片）大放异彩，收入达 623亿美元，超过预期的604亿。
但其他业务拖了后腿：游戏业务（曾是英伟达的“现金牛”）本季收入仅 37.3亿美元，低于预期的40.1亿；汽车业务收入为 6.04亿美元，也略低于华尔街预期的6.43亿。
为什么游戏业务疲软？关键卡在“存储芯片短缺”。
像英伟达这样的科技公司，高度依赖存储芯片——它们为从手机到AI服务器的各种设备提供临时数据存储。但目前全球存储芯片供应紧张，价格被推高，导致生产更多显卡变得更难、更贵。
CFO克雷斯坦言：她不确定这个问题今年能否缓解，游戏业务短期内可能难以恢复增长。不过，英伟达早已把重心转向AI。最近，Meta（Facebook母公司）宣布将在未来几年部署“数百万颗”英伟达AI芯片，进一步巩固双方合作。
有趣的是，英伟达的竞争对手AMD也宣布与Meta达成类似的大单，价值数百亿美元。这些大额长期订单，被芯片厂商用来证明：AI热潮不是泡沫，需求真实存在。
英伟达靠AI数据中心继续高飞，但传统业务受困于供应链问题；行业用大单“锁需求”来稳信心，可这种紧密绑定的模式，也让市场多了几分警惕。
竞太观点|市场到底在怕什么？
对长期投资者：
英伟达仍是AI算力核心基础设施提供者，生态壁垒极高；若相信AI渗透率将持续提升（尤其Agentic AI落地），回调或是布局机会。
Pay attention to two verification signals: whether enterprise-level AI spending will spread to non-Internet industries such as finance and manufacturing in H2 2026; Actual shipments and gross margin of Blackwell/Rubin chips.
For short-term traders:
Beware of "good news" fluctuations, 78 billion guidance has been price in;
If major customers such as Meta and Microsoft cut capital expenditures, it may trigger a chain reaction.
For the combination configurator:
It can be used as a satellite position with AMD, TSMC, Micron, etc. to diversify the risk of single dependence; Avoid purely conceptual AI stocks - only companies that can cash in on cash flow can withstand the bubble test.
