谷歌云业务第三季度的亮眼增长被其他令华尔街失望的财务业绩所掩盖，尤其是谷歌搜索广告收入增长乏力。谷歌股价在盘后交易中下跌近10%。
负90亿美元自由现金流
财报显示，Alphabet三季度营收同比增长25%至1275.8亿美元，略低于华尔街预期；净利润同比下滑25%至261.7亿美元。由于在人工智能领域投入巨资，三季度自由现金流为负90多亿美元，而去年同期则为正240多亿美元，这也是连续第二季度为负。
那么钱都去哪里了。本季度资本支出514亿美元，大部分用于AI技术基础设施建设，其中65%是服务器和网络设备、35%用于数据中心。值得一提的是，谷歌TPU系统销售库存挤压及2018年欧盟罚款52亿美元现金支付影响，也在影响现金流情况，第三季度经营现金流423亿美元。
Alphabet CFO阿纳特·阿什肯纳齐在财报会上指出，公司2026年全年资本支出指导区间1950亿至2050亿美元保持不变，预计2027年资本支出将显著增加。这使得外界更加关注谷歌这笔钱以及未来花的钱究竟值不值，而本季财报中唯一亮点的是谷歌云业务，而不是谷歌搜索。
一方面，Alphabet的基本盘搜索业务只增长了12%至633.5亿美元，也是自2025年第二季度以来该业务的最低增速。这一结果与其宣称的“人们比以往任何时候都更频繁使用谷歌搜索”相矛盾。Alphabet CEO桑达尔·皮查伊表示，谷歌的搜索查询量同比增长率是五年来最高的，而Alphabet首席商务官菲利普·辛德勒则解释，AI Mode中的搜索量受到了影响，导致现阶段AI Mode的盈利能力在下降。这一说法意味着，AI确实在改变用户的搜索行为，但怎么商业化还并不清晰。
另一方面，谷歌云的亮眼则主要来自于蓬勃发展的TPU业务。本季度谷歌云营收翻了一番多，超过300亿美元，而营业利润更是增长了两倍，达到111.2亿美元。这一营收数字仅略低于微软Azure同期330亿美元的营收。
不过，谷歌并未披露本季度TPU的具体销售额，但表示TPU是“收入增长的重要贡献者”。Alphabet管理层向分析师发出预期，预计明年TPU销售额将对谷歌云产生更大的影响。同时本月初，TD Cowen一份报告中预估，TPU在今年下半年将创造94亿美元的销售额。
从外部来看，大量客户对TPU的需求正在激增，其中包括OpenAI、Anthropic，前者一直是英伟达AI芯片的主要客户，而后者则已经宣布计划在谷歌云上投入1145亿美元。
最强模型同日登顶
就在财报发布几小时前，谷歌推出Gemini 4 Carbon，在Artificial Analysis智能指数以59分登顶，压过Claude Opus 5.5和GPT-6 Astra。这是一个月内第二个主要模型——近一年没发旗舰模型的谷歌，突然连发两款，似乎急于把市场份额从OpenAI和Anthropic手里抢回来。
谷歌给Carbon的定位是“网络安全与长程任务的最强模型”，谷歌表示，Carbon在网络安全能力方面实现了“重大飞跃”，其生成安全代码的能力几乎是Argon的两倍。Carbon的发布正值网络安全危机爆发之际，此前，一些不法人工智能程序利用了企业和政府的漏洞。
与Argon一样，谷歌表示将首先向部分网络安全专业人士发布Carbon，然后再向其Fairwind项目的合作伙伴发布。谷歌还表示，正在积极参与美国政府的预发布模型测试项目。
但目前来看Carbon的商业姿态很拧巴：没有开放公开API、只向少数网络安全防御者开放、普通开发者开放时间未定。同时，在定价方面，则继续按照“低价进场、到期翻倍”（输入2美元/输出10美元，优惠期后翻倍）。
科技巨头的融资转向
作为今年早些时候宣布的“市场发行计划”的一部分，Alphabet三季度出售了价值124亿美元的股票：1200万股A类股筹43亿美元，2400万股C类股筹81亿美元。这是6月宣布的900亿美元股权融资计划的一部分，截至9月30日仍有276亿美元可用。
Alphabet在过去一年间已通过六大主要货币市场累计发债超过850亿美元，债务总规模突破1000亿美元。进入2026年后，发债节奏明显加快：2月发行200亿美元多期限美元债，5月先后发行90亿欧元与85亿加元债券，随后又发售约36亿美元的日元武士债。这种“股债并举”的融资策略，与其过去多年偏好回购股票、保持低负债的形象形成了鲜明反差。
事实上，不只是Alphabet，随着AI基础设施建设投入的不断增加，各大科技公司一直在竞相发债发股以筹措资金。Meta本周二表示，将计划通过股票发行筹集975亿美元用于AI投资，高于此前宣布的950亿美元目标，同时，Meta三季度自由现金流同样转负，为负21.2亿美元。
当前，多家万亿市值的科技巨头仍在不断追加资金砝码，这场尚不明朗的AI竞速仍在继续。（本文首发于钛媒体APP，文｜TechHorizon）
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
The brightest growth in the third quarter of Google Clouds operations was overshadowed by other disappointing financial performances on Wall Street, particularly the weak growth in Google search advertising revenues. Google stock prices fell by almost 10 per cent in post-business transactions.
Negative $9 billion in free cash flows
According to the press, Alphabet earned an increase of 25 per cent to $12,758 million over the same period in the third quarter, slightly below the Wall Street projection; net profits declined from 25 per cent to $26.17 billion over the same period. As a result of significant investment in artificial intelligence, free cash flows in the third quarter were negative of over $9 billion, compared with more than $24 billion in the same period last year, which was negative for the second consecutive quarter.
So where's the money? Capital expenditure this quarter was $51.4 billion, mostly for the construction of AI technology infrastructure, 65 per cent of which was for servers and network equipment and 35 per cent for data centres. It is worth mentioning that the pressure on stocks from Google TPU sales and the effect of the EU fine of $5.2 billion in 2018 on cash payments are also affecting cash flows, which are running at $42.3 billion in the third quarter.
At the press conference, Alphabet CFO Anat Ashkenazi noted that between 1950 billion and 20.5 billion dollars remained constant throughout 2026, and capital expenditure is expected to increase significantly in 2027. This has led to increased attention being paid to the value of Google's money and future spending, while the only highlight in this season's financial statements is Google Cloud Business, not Google Search.
On the one hand, Alphabet ' s basic disk search operation grew by only 12 per cent to $63.35 billion, the lowest rate since the second quarter of 2025. This result contradicts its assertion that “the people are using Google search more frequently than ever”. Alphabet CEO Sandard Pichai states that Google ' s search search queries grew at the highest rate in five years, while Alphabet ' s Chief Commercial Officer, Philip Schindler, explained that the search volume in AI Mode was affected, leading to a decline in AI Mode ' s profitability at this stage. This statement means that AI is indeed changing the user ' s search behaviour, but it is not clear how commercialization will take place.
On the other hand, the bright eye of Google Clouds comes mainly from the flourishing TPU business. The harvest of Google Clouds more than doubled this quarter to over $30 billion, while operating profits tripled to $1112 billion. This collection figure is only slightly lower than Microsoft Azure's $33 billion collection over the same period.
However, Google did not disclose the specific sales of TPU during the quarter, but indicated that TPU was “an important contributor to income growth”. Alphabet management sent the analyst the expectation that next year's TPU sales would have a greater impact on Google Clouds. At the same time, earlier this month, a report by TD Cowen predicted that TPU would generate $9.4 billion in sales in the second half of the year.
Externally, demand for TPU by a large number of clients is on the rise, including Open AI, which has been a major client of the British Wida AI chip, and Anthropic, which has announced plans to invest $114.5 billion in Google clouds.
The strongest model goes to the top of the same day.
Just a few hours before the newspaper publish, Google launched Gemini 4 Carbon, peaking at 59 points in the Artificial Analysis Intelligence Index, crushing Claude Opus 5.5 and GPT-6 Astra. This is the second major model in a month — Google without a flagship model for almost a year — and it suddenly re-emerges two pieces, which seem eager to take back market shares from Open AI and Anthropic.
Google's positioning of carbon is “the strongest model of cybersecurity and long-range missions”, and Google indicates that carbon has achieved “a major leap” in cybersecurity capabilities, with almost twice as much capacity as Argon to generate security codes. Carbon's publish came at a time when the cyber-security crisis had been triggered by a number of irregularities that had exploited the loopholes of businesses and Governments.
Like Argon, Google indicated that it would first address some of the network security professionals, publishCarbon, and then publish, a partner in its Fairwind project. Google also indicated that it was actively participating in the United States Government ' s pre-publish model testing project.
But at the moment, the business stance of Carbon has been very twisted: it is not open to API, it is open to only a few cyber-security defenders, and it is open to ordinary developers. At the same time, in terms of pricing, the “low entry, double the maturity” (input $2/output $10, double the preferential period).
The technology giant's financing is shifting.
As part of the “market distribution scheme” announced earlier this year, Alphabet sold $12.4 billion worth of stocks in three quarters: $12 million in A and $4.3 billion in C and $8.1 billion in C. This is part of the $90 billion equity financing scheme announced in June, with $27.6 billion available as at 30 September.
Alphabet had accumulated over $85 billion over the past year through six major currency markets, and the total size of the debt had exceeded $100 billion. After entering 2026, the pace of debt disbursements accelerated significantly: over $20 billion in United States dollars in February, €9 billion and €8.5 billion in bonds in May, followed by approximately $3.6 billion in Yen. This “bond-to-stock” financing strategy is in stark contrast to its preference for stock buy-backs over the past years and its low-debt image.
Indeed, not just Alphabet, as AI infrastructure investment continues to grow, leading technology companies have been competing for debt-financing. Meta indicated this Tuesday that it would plan to raise $97.5 billion for AI investments through equity issues, above the previously announced target of $95 billion, while free cash flows in the third quarter were equally negative at $2.12 billion.
At present, the multibillion-dollar technology giants continue to generate additional funding, and this undefined AI race continues. (This post was launched on titanium medium APP, Man-TechHorizon)
Auto-fetched and translated by AI Atlas. All rights belong to the original author.