英伟达的财报已不再仅仅关乎公司自身。这家市值高达4万亿美元的芯片巨头的季度财报已成为人工智能热潮的试金石，进而成为整个股市的风向标。作为在标普500指数（按市值加权）中占比达8%的成份股，且在支撑生成式人工智能发展的芯片领域占据无可匹敌的地位，英伟达的业绩在华尔街眼中早已超越“单一公司成绩单”的范畴，更像是宏观经济指标。其财报发布甚至演变成一种文化现象，催生出“观摩派对”。
投资者正屏息等待英伟达周三收盘后发布的最新季度财报。从英伟达期权交易情况来看，市场预期该股将上下波动6%——这意味着其市值可能发生2600亿美元的变动。
自五月份上次季度财报发布以来，英伟达股价已飙升35%。鉴于近期市场对人工智能相关股票存在危险金融泡沫的担忧加剧，本季度最受瞩目的财报盛事的紧张氛围进一步升级。此外，英伟达在华业务的不确定性依然存在。
华尔街分析师预计英伟达第二季度营收将同比增长53%至460亿美元，达到公司业绩指引区间的上限，每股收益为1.01美元。作为英伟达核心业务的数据中心销售额预计将接近400亿美元。鉴于英伟达股价近几个月已累积可观涨幅，若周三公布的财报未达预期，抑或因中国限制政策而发布谨慎业绩指引，其股价恐将暴跌。
英伟达陷中美博弈漩涡
尽管英伟达仍是生成式人工智能浪潮的最大受益者之一，但随着中美争夺科技主导权，该公司关键业务已沦为地缘政治博弈的筹码。今年4月，美国政府开始要求该公司出口H20芯片前必须申请相关许可证。H20芯片是英伟达为遵守2022年末生效、2023年进一步收紧的美国出口管制规定，专门研发的顶级人工智能芯片的简化版本。更为严苛的出口许可证要求迫使该公司在第一季度计提45亿美元的资产减值损失，涉及未售库存和采购承诺。
此后英伟达在华业务愈发复杂。在英伟达首席执行官黄仁勋前往海湖庄园与特朗普会面之后，白宫宣布将批准该公司销售H20芯片。英伟达虽已提交出口许可证申请，但受美国政府强硬立场及中国买家采购时犹豫不决的影响，申请进程遭遇严重延误。本月初，英伟达与超微半导体（AMD）与特朗普政府达成协议，以对华芯片销售收入的15%换取出口许可证。
然而在H20芯片恢复出货之际，中国政府开始劝阻国内企业，推动其放弃采购该芯片，原因是担忧英伟达要求客户提交供美国政府审查的信息可能包含敏感内容。据报道，中国政府还宣称发现证据，表明英伟达芯片可能存在后门，允许美国情报机构提取芯片使用数据。
上周黄仁勋在中国台北宣布，英伟达已开始逐步停产H20芯片，并着手研发性能更强的后续产品。他表示公司正致力于推出“面向人工智能数据中心的新产品”，该产品将按美国要求进行调整，降低部分性能。他表示正在寻求特朗普政府批准该芯片的对华销售许可。
黄仁勋表示：“毋庸置疑，这取决于美国政府。我们正与他们沟通，但目前尚无定论。”
受上述诸多不确定性影响，分析师预测英伟达在本次财报中不会提及中国市场的营收情况。
“我推测他们既不会对中国市场的营收进行统计，也不会作出相关预测，原因是存在太多不确定性。”Cambrian-AI Research创始人兼首席分析师卡尔·弗罗伊德（Karl Freund）指出。
J. Gold Associates创始人兼首席分析师杰克·高德（Jack Gold）向《财富》杂志指出，英伟达如今需同时取悦两大群体：股东和特朗普政府。“他们正陷入进退两难的境地，”他坦言，“当前局势异常诡谲——美国政府实际上已将手伸进了这些企业的钱袋子。”
人工智能泡沫难题
除地缘政治因素外，英伟达还面临另一重挑战：人们愈发担忧人工智能热潮开始步入泡沫阶段。这将直接冲击英伟达的业务核心及其极高估值——该公司当前市盈率超40倍——依赖于对其高性能图形处理单元（GPU）持续增长的需求。英伟达的增长高度依赖少数几家云计算巨头，包括Meta、亚马逊（Amazon）、谷歌（Google）和微软（Microsoft），以及OpenAI等资金雄厚的人工智能初创企业。倘若这些公司放缓支出步伐，英伟达可能突然失去最大买家。
“我确实认为所有人都担心人工智能存在泡沫，”弗罗伊德表示，不过他补充道，这种担忧已持续三年之久。他强调，自己并不认为泡沫会在当下破裂。“我认为未来还有二到五年的增长期。”他说道。
高德对此表示认同，称英伟达“至少在接下来的几个季度内，甚至几年内都能获得可观利润”，但他警告称，在某个特定时刻，倘若市场崩盘，用于购买芯片的资金便会消失殆尽。
“这着实令我忧心不已，”他坦言。“这次，我笃定英伟达的财报数据依然亮眼——（英伟达）以高得离谱的价格出售所能制造的所有产品，这倒也无妨，只要它能一直这般蒙混过关。”但他补充道，从更广泛的市场角度来看，大规模的人工智能数据中心建设“不可能永远持续下去”。
弗罗伊德指出，这正是黄仁勋实际上正着力引导投资者，将关注焦点从以数据中心为中心的视角转向英伟达业务其他板块的原因，包括其汽车和机器人业务：“这便是他当下的策略，即随着人工智能从数据中心走向现实世界，如何让投资者转向以更全面的视角看待人工智能。”（*）
译者：中慧言-王芳
英伟达的财报已不再仅仅关乎公司自身。这家市值高达4万亿美元的芯片巨头的季度财报已成为人工智能热潮的试金石，进而成为整个股市的风向标。作为在标普500指数（按市值加权）中占比达8%的成份股，且在支撑生成式人工智能发展的芯片领域占据无可匹敌的地位，英伟达的业绩在华尔街眼中早已超越“单一公司成绩单”的范畴，更像是宏观经济指标。其财报发布甚至演变成一种文化现象，催生出“观摩派对”。
投资者正屏息等待英伟达周三收盘后发布的最新季度财报。从英伟达期权交易情况来看，市场预期该股将上下波动6%——这意味着其市值可能发生2600亿美元的变动。
自五月份上次季度财报发布以来，英伟达股价已飙升35%。鉴于近期市场对人工智能相关股票存在危险金融泡沫的担忧加剧，本季度最受瞩目的财报盛事的紧张氛围进一步升级。此外，英伟达在华业务的不确定性依然存在。
华尔街分析师预计英伟达第二季度营收将同比增长53%至460亿美元，达到公司业绩指引区间的上限，每股收益为1.01美元。作为英伟达核心业务的数据中心销售额预计将接近400亿美元。鉴于英伟达股价近几个月已累积可观涨幅，若周三公布的财报未达预期，抑或因中国限制政策而发布谨慎业绩指引，其股价恐将暴跌。
英伟达陷中美博弈漩涡
尽管英伟达仍是生成式人工智能浪潮的最大受益者之一，但随着中美争夺科技主导权，该公司关键业务已沦为地缘政治博弈的筹码。今年4月，美国政府开始要求该公司出口H20芯片前必须申请相关许可证。H20芯片是英伟达为遵守2022年末生效、2023年进一步收紧的美国出口管制规定，专门研发的顶级人工智能芯片的简化版本。更为严苛的出口许可证要求迫使该公司在第一季度计提45亿美元的资产减值损失，涉及未售库存和采购承诺。
此后英伟达在华业务愈发复杂。在英伟达首席执行官黄仁勋前往海湖庄园与特朗普会面之后，白宫宣布将批准该公司销售H20芯片。英伟达虽已提交出口许可证申请，但受美国政府强硬立场及中国买家采购时犹豫不决的影响，申请进程遭遇严重延误。本月初，英伟达与超微半导体（AMD）与特朗普政府达成协议，以对华芯片销售收入的15%换取出口许可证。
然而在H20芯片恢复出货之际，中国政府开始劝阻国内企业，推动其放弃采购该芯片，原因是担忧英伟达要求客户提交供美国政府审查的信息可能包含敏感内容。据报道，中国政府还宣称发现证据，表明英伟达芯片可能存在后门，允许美国情报机构提取芯片使用数据。
上周黄仁勋在中国台北宣布，英伟达已开始逐步停产H20芯片，并着手研发性能更强的后续产品。他表示公司正致力于推出“面向人工智能数据中心的新产品”，该产品将按美国要求进行调整，降低部分性能。他表示正在寻求特朗普政府批准该芯片的对华销售许可。
黄仁勋表示：“毋庸置疑，这取决于美国政府。我们正与他们沟通，但目前尚无定论。”
受上述诸多不确定性影响，分析师预测英伟达在本次财报中不会提及中国市场的营收情况。
“我推测他们既不会对中国市场的营收进行统计，也不会作出相关预测，原因是存在太多不确定性。”Cambrian-AI Research创始人兼首席分析师卡尔·弗罗伊德（Karl Freund）指出。
J. Gold Associates创始人兼首席分析师杰克·高德（Jack Gold）向《财富》杂志指出，英伟达如今需同时取悦两大群体：股东和特朗普政府。“他们正陷入进退两难的境地，”他坦言，“当前局势异常诡谲——美国政府实际上已将手伸进了这些企业的钱袋子。”
人工智能泡沫难题
除地缘政治因素外，英伟达还面临另一重挑战：人们愈发担忧人工智能热潮开始步入泡沫阶段。这将直接冲击英伟达的业务核心及其极高估值——该公司当前市盈率超40倍——依赖于对其高性能图形处理单元（GPU）持续增长的需求。英伟达的增长高度依赖少数几家云计算巨头，包括Meta、亚马逊（Amazon）、谷歌（Google）和微软（Microsoft），以及OpenAI等资金雄厚的人工智能初创企业。倘若这些公司放缓支出步伐，英伟达可能突然失去最大买家。
“我确实认为所有人都担心人工智能存在泡沫，”弗罗伊德表示，不过他补充道，这种担忧已持续三年之久。他强调，自己并不认为泡沫会在当下破裂。“我认为未来还有二到五年的增长期。”他说道。
高德对此表示认同，称英伟达“至少在接下来的几个季度内，甚至几年内都能获得可观利润”，但他警告称，在某个特定时刻，倘若市场崩盘，用于购买芯片的资金便会消失殆尽。
“这着实令我忧心不已，”他坦言。“这次，我笃定英伟达的财报数据依然亮眼——（英伟达）以高得离谱的价格出售所能制造的所有产品，这倒也无妨，只要它能一直这般蒙混过关。”但他补充道，从更广泛的市场角度来看，大规模的人工智能数据中心建设“不可能永远持续下去”。
弗罗伊德指出，这正是黄仁勋实际上正着力引导投资者，将关注焦点从以数据中心为中心的视角转向英伟达业务其他板块的原因，包括其汽车和机器人业务：“这便是他当下的策略，即随着人工智能从数据中心走向现实世界，如何让投资者转向以更全面的视角看待人工智能。”（*）
译者：中慧言-王芳
Nvidia’s earnings aren’t just about Nvidia anymore. The $4 trillion chipmaker’s quarterly financials have become a litmus test for the AI boom—and, by extension, for the whole stock market. Constituting 8% of the market-cap-weighted S&P 500 index and with an unrivaled grip on the chips that power generative AI, Nvidia sees its results treated more like a macroeconomic indicator by Wall Street than as a report card on a single company. The earnings announcement has even become a cultural phenomenon complete with watch parties.
Investors are bracing for the company’s latest quarterly results due after Wednesday’s market close, with trading in Nvidia options implying expectations that the stock will move 6%, up or down—equal to a $260 billion change in Nvidia’s market value.
In the three months since the company last gave investors a quarterly update, back in May, Nvidia’s stock has surged 35%. But the tension surrounding what is already the most closely watched earnings event of the season has been ratcheted up by recent jitters over what some worry is a dangerous financial bubble in AI-related stocks. And uncertainty about Nvidia’s China business continues to loom large.
Wall Street analysts are looking for Nvidia’s Q2 revenue to surge 53% year over year to $46 billion, at the high end of Nvidia’s guidance, with earnings per share of $1.01. Data center sales, the crux of Nvidia’s business, are expected to come in close to $40 billion. But with Nvidia’s shares having gained so much in recent months, a miss on Wednesday, or cautious guidance tied to China restrictions, could send the stock plummeting.
Nvidia in the U.S.-China crosshairs
Nvidia may remain one of the greatest beneficiaries of the generative AI boom, but a critical part of the company’s business has also become a geopolitical football as the U.S. and China compete for technological dominance. In April, Washington began requiring export licenses for the company’s H20 chips—stripped-down versions of Nvidia’s top-of-the-line AI chips that were specifically designed to comply with the U.S. export controls that took effect in late 2022 and were tightened again in 2023. Those tighter export licenses forced the company to take a $4.5 billion charge in Q1 tied to unsold inventory and purchase commitments.
From there, things only got more complicated for Nvidia’s China business. After Nvidia CEO Jensen Huang visited President Trump at Mar-a-Lago, the White House said it would permit the company to sell H20s after all. Nvidia applied for export licenses but faced extensive delays, thanks to the tougher U.S. stance and Chinese buyers hesitating to commit to purchasing. Then, earlier this month, Nvidia and AMD struck a deal with the Trump administration to grant licenses in exchange for a 15% revenue-sharing arrangement on China chip sales.
But as shipments of H20 chips resumed, China began discouraging companies from buying them, expressing concerns that the information Nvidia was asking customers to submit for U.S. government review could contain sensitive information. The Chinese government also reportedly claimed it had found evidence that Nvidia’s chips might contain back doors that would allow U.S. spy agencies to extract data on how they were being used. In addition, comments from U.S. Commerce Secretary Howard Lutnick about providing China with Nvidia’s “fourth-best chips” were considered “deeply insulting” by Chinese officials, according to the Financial Times.
Finally, last week Huang announced in Taipei that Nvidia has started winding down production of the H20 chip and begun work on a more powerful successor, saying the company was working on offering a “new product for AI data centers,” modified to reduce some of its performance, as required by the United States. He said he was seeking the Trump administration’s approval to sell the chip.
“It’s up to, of course, the United States government,” Huang said. “And we’re in dialogue with them, but it’s too soon to know.”
As a result of all the uncertainty, analysts predict Nvidia will not allude to China revenue in the earnings report.
“I suspect they will not count, nor forecast China revenue; there’s too much uncertainty involved,” said Karl Freund, founder and principal analyst at Cambrian-AI Research.
Jack Gold, founder and principal analyst at J. Gold Associates, told Fortune that Nvidia now has two primary groups to please: stockholders and the Trump administration. “They’re caught between a rock and a hard place,” he said. “It’s a really strange situation we’re in now where the government in the U.S. actually has their hands into the pockets, into the wallets of these companies.”
AI bubble trouble
Beyond geopolitics, Nvidia faces another challenge: growing unease that the AI boom is starting to look like a bubble. This would strike at the heart of Nvidia’s business and its stratospheric valuation—the company trades at more than 40 times its projected earnings—which rely on ever-growing demand for its powerful GPUs. Nvidia’s growth is heavily concentrated in a handful of cloud giants, including Meta, Amazon, Google, and Microsoft, as well as highly funded AI startups like OpenAI. If those companies slow spending, Nvidia could suddenly lose its biggest buyers.
“I do believe that everyone’s concerned about an AI bubble,” said Freund, though he added that those concerns have lasted for three years already. He did not, he emphasized, think it would pop now. “I think there are still two to five years of growth left,” he said.
Gold agreed, saying there were “at least several quarters, if not a couple of years of good profits” for Nvidia, but he cautioned, at some point, if the market crashed, that money spent on chips would go away.
“It concerns me,” he said. “This time, I’m sure the earnings will still be great—[Nvidia is] selling everything they can build at ridiculously inflated prices, which is fine, if you can get away with that.” But from a broader market perspective, he added, the massive AI data center build-outs “can’t go on forever.”
That’s why, said Freund, Huang is actually working to get investor attention to shift from the data-center-centric view to other areas of Nvidia’s business, including its automotive and robotics work: “That’s his game right now, how to get investors to shift to a more holistic view of AI as it moves out of the data center and into the real world.”