The $Invesco QQQ(QQQ)$ fell 1.42 per cent on Thursday and the $S&P 500(.SPX)$ declined 0.30 per cent. $Amazon.com(AMZN)$ closed down 4.25 per cent and fell about another 7 per cent after its results. Microsoft slipped just 0.16 per cent, while Meta gained 1.96 per cent. Companies making large AI investments were no longer trading as one group.
Amazon reported third-quarter revenue of $201.9 billion, up 12 per cent, and earnings of $1.89 a share. AWS revenue grew 39 per cent to $46.0 billion, while operating income rose 55 per cent to $17.8 billion. The issue was cash generation: trailing 12-month free cash flow swung from positive $14.8 billion a year earlier to negative $42.4 billion. Amazon expects about $220 billion of capital expenditure in 2026 and further growth in 2027. The relative resilience of Microsoft and Meta suggests that investors have not rejected AI spending, but they are separating companies by how quickly growth can produce cash.
Google Cloud doubled, yet Alphabet fell 12.64 per cent
$Alphabet(GOOG)$ fell 12.64 per cent in the regular session. Third-quarter revenue was $127.58 billion and earnings were $2.12 a share. Google Cloud revenue increased 103 per cent to $30.71 billion, with operating income of $10.64 billion. That growth was not enough to offset pressure on cash flow.
Quarterly free cash flow was negative $9.07 billion as capital expenditure reached $51.4 billion. Alphabet maintained full-year capital expenditure guidance of $195 billion to $205 billion and expects spending to rise significantly again in 2027. Cloud backlog supports future revenue, but investors still need evidence that it can cover depreciation, computing costs and cash consumption. Backlog is not current-period revenue or realised cash return.
Intel's 23 per cent after-hours fall put the focus on its forecast
$Intel(INTC)$ closed down 3.62 per cent and fell about 23 per cent after results. Third-quarter revenue rose 21 per cent to $16.5 billion, while adjusted earnings were $0.52 a share. Data-centre and AI revenue reached $7.3 billion and increased 78 per cent, showing clear improvement in the current quarter.
The fourth-quarter outlook was weaker. Intel guided revenue to $15 billion to $16 billion and adjusted earnings to $0.25 a share. An October accident at its Israel factory is expected to reduce quarterly revenue by about $300 million and gross margin by roughly 200 basis points. Data-centre growth remained strong, but PC revenue fell 5 per cent, leaving a wider gap between the two businesses. Guidance is a management forecast, not a signed order.
Sandisk delivered strong numbers and still declined
$SanDisk Corp.(SNDK)$ fell 5.32 per cent in the regular session and about another 3.8 per cent after results. First-quarter financial-year 2027 revenue was $11.03 billion, up 23 per cent sequentially. Adjusted gross margin reached 85.1 per cent and adjusted earnings were $49.51 a share. Data-centre revenue rose 61 per cent sequentially to $4.8 billion as enterprise SSD demand benefited from AI infrastructure investment.
The company guided next-quarter revenue to $12.2 billion to $12.6 billion and adjusted gross margin to 84.5 per cent to 85.5 per cent. It also plans an eight-for-one stock split on 19 November. With no obvious shortfall in the results or outlook, the decline looked more like profit-taking after high expectations. Micron fell 4.98 per cent and SK hynix lost 1.58 per cent on Thursday, suggesting that memory shares now require a larger upside surprise to extend their valuation gains.
ServiceNow's 26.95 per cent drop exposed a valuation conflict
$ServiceNow(NOW)$ fell 26.95 per cent, one of the steepest declines among major software shares. Third-quarter total revenue was $4.45 billion, while subscription revenue rose 21 per cent to $4.04 billion and current remaining performance obligations increased 20 per cent. The company also lifted its 2026 AI annual contract value target from $1.5 billion to $1.7 billion.
The figures showed that AI products were generating revenue, but the share-price reaction shifted attention to valuation and the pace of profit conversion. Adjusted earnings of $1.00 a share were slightly below the roughly $1.02 expected. When a valuation already assumes sustained high growth, revenue expansion of about 20 per cent may not be enough. Investors will want AI contracts to lift margins and free cash flow more quickly.
Marvell's 13.24 per cent fall put potential acquisition costs first
Marvell fell 13.24 per cent on Thursday after gaining 3.32 per cent in the previous session. The immediate catalyst was a report that the company was considering an acquisition of Lumentum. Investors focused on the financing requirement, possible dilution, higher debt and integration risk. No final agreement has been announced, so a potential transaction should not be treated as completed.
There is still a strategic case. Marvell's switching silicon and DSP products could be combined with Lumentum's lasers and optical-interconnect portfolio, while Coherent offers a useful industry comparison. The market first priced the cost of the deal and only then considered the possible synergies. The next important variables are the offer price, financing structure and the time management believes it would take to earn an acceptable return.
The sell-off questioned the timing of AI returns, not AI demand itself
The six market reactions pointed to the same change: rapid growth no longer guaranteed a higher share price. Amazon and Alphabet faced questions about cash consumption. Intel's weakness came from its next-quarter outlook. Sandisk had to overcome high expectations, ServiceNow faced a valuation test and Marvell was hit by potential acquisition costs. Microsoft and Meta held up better, showing that investors were still willing to own companies that protected profit and cash flow.
The next test has three parts: whether capital expenditure grows more slowly than revenue, whether free cash flow improves and whether quarterly results validate management's stated return period. AI demand remains strong, but the market now wants growth, profit and cash at the same time.
A note on sources: percentage moves for QQQ, the S&P 500, Alphabet, ServiceNow, Marvell, Microsoft, Meta, Micron and SK hynix use the 29 October US regular-session close. Earnings-reaction exception: Amazon, Intel and Sandisk use extended trading after their results. Company figures come from the quarterly results of Amazon, Alphabet, Intel, Sandisk and ServiceNow. The Marvell section is based on reporting about a potential acquisition; no final agreement has been announced. Company guidance does not equal a signed order and does not allocate fixed revenue to suppliers. The interpretation of valuation, cash returns and market expectations is analysis. This is personal analysis and not investment advice.
💬 【Talking Point】
Which operating metric would convince you that Big Tech's current investment cycle is creating durable value rather than simply increasing capacity?
💰 【Bounty】
Share your view and reasoning in the comments for a chance to earn Tiger Coins! 🎁
🔔 Better shared than saved — tag a friend and join the discussion!
Comments
因为公司花很多钱建设AI数据中心,不代表这些投资最终能够赚钱。
我最关注的5个运营指标
指标 最简单的理解 什么情况才算好?
1. 自由现金流 FCF 公司投资后还剩多少现金 持续增长
2. AI收入增长 AI业务真正赚到多少收入 收入持续增加
3. 毛利率 每赚100元,扣除直接成本后剩多少 稳定或提高
4. 投资回报率 ROI 投入100元,能够带回多少回报 回报逐渐超过资金成本
5. 客户续约及订单 客户是否愿意继续付钱 续约率高、订单持续增加
用三家公司举例
1. Microsoft(MSFT)
我最关注Azure云业务增长,以及AI服务能否带来更多现金流。
如果收入增长、利润率提高、FCF也持续增长,代表AI投资开始产生实际回报。
2. Amazon(AMZN)
AWS增长39%是好消息,但过去12个月FCF转为负424亿美元,值得警惕。
我希望看到未来AWS利润增长逐渐超过AI基础设施投资带来的成本压力。
3. Alphabet(GOOG)
Google Cloud增长103%,非常强劲。
但季度FCF为负90.7亿美元,我会继续观察云业务增长能否逐渐转化为现金。
我的选股判断标准
如果一家公司连续几个季度出现以下情况,我才会更有信心:
AI收入持续增长。
营业利润率提高。
自由现金流改善。
投资回报率超过资金成本。
不需要不断借钱或发行新股维持投资。
最后记住一句话
AI时代,真正值得长期投资的,不一定是花钱最多的公司,而是能够把AI投资变成利润和现金流的公司。
如果只能选一个指标,我选自由现金流(FCF)。
因为EPS可以受到会计项目影响,但现金流更能帮助我们判断公司有没有真正创造现金。
原因很简单:AI收入增长证明市场有需求,但只有新增利润最终转化为现金,才能证明巨额投资创造了股东价值。
以亚马逊为例,AWS收入增长39%,经营利润增长55%,说明云业务需求和盈利能力都在改善。但过去12个月自由现金流转负,意味着投资者还需要等待新增算力投入转化为足够的现金回报。
我会重点跟踪三个指标。
**第一,增量经营利润与增量资本投入的关系。** 如果企业每增加100亿美元资本投入,未来能够持续创造多少新增经营利润?尤其要考虑GPU更新周期、服务器折旧和电力成本。收入增长很快,但资本回报率持续下降,并不是理想的增长模式。
**第二,过去12个月自由现金流的变化。** 单季度自由现金流为负不一定危险,因为数据中心建设具有阶段性。真正需要警惕的是资本开支持续增加,而经营现金流增长逐渐放缓,导致现金缺口不断扩大。
**第三,AI订单向实际收入和现金回款的转化效率。** 云业务积压订单和软件公司的AI合同价值可以证明需求,但不能直接等同于已经赚到的钱。我更希望看到订单兑现、利润率改善和现金回款同步发生。
这也是为什么我认为GOOGL、MSFT和AMZN不能简单按跌幅排序。**谷歌跌12%未必最便宜,微软相对抗跌也未必最贵,关键是未来现金回报相对于当前估值的吸引力。**
对ServiceNow这类软件公司,我会特别关注AI产品是否真正提高客户付费金额和自由现金流利润率,而不只是带来更多合同。
对于芯片和存储公司,还要防止把AI建设高峰期的订单与利润率当成长期常态。
**我的判断是:下一阶段AI行情可能从“谁投入最多”转向“谁的增量资本回报率最高”。**
短期我会继续观察QQQ整体走势,中长期则更愿意研究那些AI收入增长、经营利润扩张和自由现金流改善能够相互验证的公司。
毕竟,**资本开支决定企业能建设多少算力,而资本回报率才决定这些算力最终能为股东创造多少价值。** 🐯