AI Revenue Is Growing. How Much Reaches Shareholders?

Hotter inflation renews pressure on the rebound

The $Invesco QQQ(QQQ)$ fell 0.68 per cent on Tuesday and the $S&P 500(.SPX)$ declined 0.38 per cent. October US CPI rose 0.4 per cent month on month, above the market forecast of 0.3 per cent but below September's 0.6 per cent. Core CPI increased 0.3 per cent, exceeding the 0.2 per cent forecast and matching the previous month.

Headline inflation slowed, but core inflation did not ease further, making disinflation a less convincing basis for extending the technology rebound. Interest-rate pressure can reduce the value investors place on distant profits and increase the burden on companies financing expansion. One CPI release cannot determine the size of a December rate increase, however; a 50-basis-point move is not a settled outcome.

Nvidia's investigation is not evidence of weaker chip demand

$NVIDIA(NVDA)$ fell 5.47 per cent while TSMC gained 2.02 per cent. On 10 November, Nvidia's independent directors announced that they had retained Joon H. Kim and law firm Wachtell Lipton to conduct an independent investigation. The board had previously said it had no reason to believe national security or company secrets had been compromised. Progress with the investigation does not establish wrongdoing.

$Taiwan Semiconductor Manufacturing(TSM)$reported October revenue of NT$586.89 billion on the same day, up 14.7 per cent month on month and 59.7 per cent year on year. That provides actual revenue evidence for foundry demand, but one month's figures cannot establish AI-related profitability. The companies face different tests: Nvidia needs to resolve governance concerns, while TSMC must show that growth can keep translating into profit.

CoreWeave raises guidance, but interest absorbs the operating profit

$CoreWeave, Inc.(CRWV)$ fell 0.51 per cent in regular trading before reporting third-quarter revenue of $3.661 billion after-hours, above the market forecast of approximately $3.62 billion. Its GAAP loss was $1.27 a share and its quarterly net loss was $709 million. The company raised full-year revenue guidance to $13.0 billion to $13.3 billion; the final after-hours quote was approximately 7.90 per cent above the regular-session close.

Revenue growth is materialising, but quarterly operating income was just $207 million against net interest expense of $846 million. Expected full-year capital spending is $37 billion to $39 billion, with the lower end raised from previous guidance. Interest expense for the following quarter is expected to be another $1.05 billion to $1.15 billion. Faster expansion has not removed financing costs, and a revenue beat does not mean the business is profitable.

The company's approximately $132.5 billion revenue backlog includes remaining performance obligations and other expected revenue under committed contracts, subject to delivery and service-availability requirements. It is not cash already collected. Bulls see a larger business; more cautious investors need evidence that profitability improves fast enough to cover debt costs as data centres enter service.

Anthropic's roadshow does not turn funding into cloud revenue

Anthropic formally launched its IPO roadshow on 10 November, proposing to sell 825 million shares at $70 to $78 each. At the upper end, the base offering would total approximately $64.35 billion: about 801.35 million shares would be newly issued and approximately 23.65 million sold by existing holders. Proceeds from the latter do not go to the company. This remains a proposed range, not a final offer price.

For Alphabet and Amazon, the value of investment holdings must be separated from cloud-services revenue. A higher listing valuation could increase investment value, but it cannot be counted directly as Google Cloud or AWS operating revenue. Returns from computing partnerships still depend on actual usage and delivery. New funding can support expansion while requiring investors to assess the returns on larger investment commitments.

T-Head's proposed spin-off still needs operating evidence

$Alibaba(BABA)$ fell 2.52 per cent. In a 10 November announcement, the company said T-Head had confidentially submitted a Hong Kong listing application on 11 September, seeking a separate Main Board listing. Permission to proceed with the spin-off process is not final listing approval, and the offering valuation has not been set.

A separate listing could broaden funding options and make the chip business easier to value. The benefit to Alibaba shareholders will still depend on offering terms, ownership changes and operating profit. Advancing an IPO does not automatically increase Alibaba Cloud revenue; subsequent prospectus disclosures about customers, revenue and costs will be more informative.

Sea reports pre-market, with profit growing more slowly than revenue

$Sea Ltd(SE)$ closed down 1.49 per cent on Tuesday. Third-quarter revenue was approximately $8.6 billion, above the market forecast of about $8.29 billion and up 43.4 per cent year on year. Net income was approximately $502 million, up 34.0 per cent, and GAAP diluted earnings were $0.78 a share.

Growth is supported by profit, but net income is still expanding more slowly than revenue. Whether Shopee's investment improves profitability, and whether the digital-finance business controls credit costs, will influence the quality of that growth. The result is also a reminder that beating revenue forecasts is only one part of assessing earnings; it does not establish that the shares should rise.

Taken together, inflation, earnings and IPO plans reveal a clearer distinction: demand can keep growing without revenue, profit and financing returns moving together. TSMC offers revenue evidence, CoreWeave exposes the interest burden, and prospective listings need further disclosure to justify their valuations. AI investment still offers opportunities, but choosing companies requires closer attention to how much growth becomes profit.

A note on sources: market moves use the 10 November US regular-session close; CoreWeave's approximately 7.90 per cent move is separately identified as an after-hours quote change. Earnings, monthly sales, investigation and spin-off information use company announcements. CPI uses release-day reporting and market forecasts; no rate-hike probability or missing annual reading is inferred. Revenue forecasts come from this edition's Finnhub calendar. EPS estimates with unspecified accounting bases are not directly scored against GAAP actuals. Guidance, expected contract revenue, offering totals and recognised revenue are treated separately. This is personal analysis and not investment advice.

💬 【Talking Point】

What evidence would make you more confident that an AI company's expansion will deliver lasting shareholder returns?

💰 【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!

# Nvidia After-Hours CEO-Change Report: Can the Rally Hold?

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Report

Comment(9)

  • Top
  • Latest
  • Cadi Poon
    ·11-12 16:32
    TOP
    $Taiwan Semiconductor Manufacturing(TSM)$reported October revenue of NT$586.89 billion on the same day, up 14.7 per cent month on month and 59.7 per cent year on year. That provides actual revenue evidence for foundry demand, but one month's figures cannot establish AI-related profitability. The companies face different tests: Nvidia needs to resolve governance concerns, while TSMC must show that growth can keep translating into profit.
    Reply
    Report
  • Lanceljx
    ·11-12 11:58
    TOP
    AI Boom | Strong Demand, but Where Are the Profits? 🐯

    Hotter-than-expected US CPI pressured technology stocks, with $Invesco QQQ(QQQ)$ falling 0.68% and $NVIDIA(NVDA)$ declining 5.47%.

    Meanwhile, $Taiwan Semiconductor Manufacturing(TSM)$ reported 59.7% year-on-year October revenue growth, highlighting resilient semiconductor demand.

    $CoreWeave, Inc.(CRWV)$ raised revenue guidance, but its US$846M quarterly net interest expense far exceeded US$207M operating income. Its US$132.5B backlog offers visibility, not guaranteed profits.

    Anthropic's proposed IPO and Alibaba's T-Head spin-off also raise questions about valuation versus actual earnings.

    My take: AI demand remains strong, but revenue growth alone cannot justify premium valuations. I prioritise free cash flow, sustainable margins and returns on invested capital over order backlogs or IPO hype.

    Reply
    Report
  • Blinkfans
    ·11-12 20:42
    🚀 I give founder leadership a significant amount of weight when choosing a long-term technology investment. A founder who understands the technology, has a clear vision and is willing to innovate can create tremendous value over time.


    However, I don’t invest based on personality alone. I also look at financial performance, cash flow, competitive advantages and the ability to execute.


    For example, leaders like Elon Musk and Jensen Huang demonstrate how strong founder vision can shape entire industries.


    Ultimately, I want to invest in businesses where the founder thinks long term, adapts to change and creates sustainable shareholder value. 🐯
    Reply
    Report
  • mr bean
    ·11-12 10:24
    The key evidence would be profitable growth, not just bigger spending. I’d want to see three things: sustained revenue growth from paying customers, improving operating margins as infrastructure scales, and strong free cash flow after accounting for the enormous capital expenditure.
    I’d also look closely at customer retention, utilisation rates and whether AI demand translates into recurring, high-margin revenue rather than one-off contracts. Ultimately, the winners won't necessarily be the companies spending the most on AI, but those converting that investment into durable competitive advantages, pricing power and shareholder returns.
    Growth gets attention. Cash flow creates long-term shareholder value.
    Reply
    Report
  • D1ane
    ·11-11 18:17
    I would look for evidence that AI expansion is translating into sustainable revenue growth, improving profit margins and stronger free cash flow. Rising customer demand and repeat business would also be important, rather than relying on short-term hype or large spending commitments. Most importantly, the company should demonstrate that returns from AI investments are growing faster than the cost of building and maintaining the infrastructure. Growth is exciting, but profitable growth is what creates lasting shareholder value.
    Reply
    Report
  • 靖润
    ·11-11 16:51
    真正能让我确信一家AI公司的扩张能给股东带来持久回报的证据,绝不是发布会上的跑分测试,也不是账面营收的爆发式增长,而是极其冷酷的“资本回报率(ROIC)”与“自由现金流(FCF)”的双重闭环。


    在最新公布的10月核心CPI(环比0.3%)再次超预期、长端利率死死卡在5%以上的宏观重压下,资本市场正在经历极其残酷的估值重估。以下三个铁证,才是识别真伪AI印钞机的核心标尺:


    第一,ROIC(投入资本回报率)必须显著高于其WACC(加权平均资本成本)。
    这是斩断伪AI泡沫的第一刀。当前无风险利率逼近5%,这意味着企业的资本成本极高。如果一家AI公司为了扩张,疯狂砸钱建数据中心、买GPU,但其算力租赁或应用变现带来的ROIC无法覆盖高昂的借贷成本和折旧,那么这种扩张本质上是在“毁灭股东价值”。你可以去扒财报,只要它的增量投入换不来更高的边际利润率,所谓的“高增长”就是个吞噬现金流的黑洞。


    第二,用真正的自由现金流支撑回购与分红,而不是靠稀释股权续命。
    看赚了多少钱没用,要看最后兜里剩了多少。如果一家公司宣称AI收入大增,但经营现金流减去资本开支(CapEx)后依然是深不可测的负数,甚至还要靠发高息债或者增发新股来维持运转,那它就是一台纯粹的碎钞机。真正值得托付的标的,是那种能用AI业务产生的强劲FCF,去大规模回购注销股票、提高股息的公司。真金白银的回报,永远比PPT上的愿景更诚实。


    第三,拥有不可替代的定价权,毛利率能扛住价格战。
    目前的AI模型层已经卷入了惨烈的价格战(比如某些龙头API被迫降价)。如果一家公司的营收是靠“以价换量”拼出来的,说明它根本没有护城河,随时会被大厂的降维打击碾碎。我要看到的证据是:在算力成本高企的背景下,它的AI相关业务依然能维持极高位且稳定的毛利率。只有垄断了核心技术或生态位,才能把成本压力顺畅地转嫁给下游客户。


    回看当下的盘面,CPI数据已经让美债收益率再次蠢蠢欲动,高估值的科技股随时面临估值分母扩大的绞杀。


    在这个阶段,不要被“AI星辰大海”的宏大叙事洗脑。去锁定那些拥有极强造血能力、能在高息环境下靠自有现金流实现内生性扩张的绝对龙头。至于那些靠借钱买卡、靠烧钱续命的所谓“高增长算力新贵”,让它们在信贷收紧的寒冬里自行出清。保留充足的战略流动性,作壁上观,等待真正的错杀机会出现。
    Reply
    Report
  • Jerry Lam
    ·11-11 15:33
    🐯 让我对AI公司扩张更有信心的,不是订单金额有多大,而是增长能否持续转化为自由现金流和股东回报。

    我主要看三个证据:

    ① 订单兑现率:积压订单能否按期交付、确认收入并最终收到现金?尤其是长期算力合同,签约金额不等于实际利润。

    ② 投入资本回报率(ROIC):新增数据中心、GPU和研发投入能否创造高于资本成本的回报?如果营收增长50%,但折旧和利息增长更快,扩张未必创造价值。

    ③ 自由现金流与股东回报:企业能否在维持必要资本开支后产生现金,而不是长期依赖举债、增发来支持增长?

    CoreWeave就是值得观察的案例:收入和订单增长强劲,但高额利息费用提醒我们,规模扩张不等于股东价值增长。

    相比之下,我更青睐具备定价权、稳定毛利率和健康资产负债表的AI龙头。

    🚀 最能改变我判断的信号,是连续几个季度自由现金流改善,同时ROIC高于资本成本。AI需求决定增长空间,现金流和资本回报率才决定股东最终能赚到多少! 🐯

    Reply
    Report
  • 吉3186
    ·11-11 14:33
    我认为,判断AI公司是否值得长期投资,不能只看收入增长或订单数量,更重要的是看公司能否把AI需求转化为持续利润和现金流。

    我会重点观察五个方面:

    1. 收入是否来自真实需求
    客户是否真正使用AI服务?是否愿意持续付费和续约?订单增长必须逐渐转化为实际收入。

    2. 利润是否跟得上收入
    如果收入增长40%,净利润只增长10%,说明成本可能增加得更快。长期应关注营业利润率和净利润是否持续改善。

    3. 自由现金流是否增加
    AI数据中心需要大量资本开支,包括GPU、机房和电力。如果公司收入很高,却不断烧钱,长期发展仍有风险。

    4. 债务是否过高
    重点观察利息支出是否吞噬营业利润,以及公司是否需要不断借钱或增发股票维持扩张。

    5. 是否拥有竞争优势
    例如英伟达的GPU生态、台积电的先进制程,以及谷歌和亚马逊的云计算平台。竞争优势越强,越有机会维持长期盈利能力。

    我的投资判断:

    长期最值得关注三个指标:

    - 营业利润率:是否持续改善?
    - 自由现金流:是否稳定增长?
    - 每股盈利(EPS):是否长期上升?

    总结来说,AI需求决定公司能做多大的生意,盈利能力决定公司能创造多少价值,而每股盈利和现金流,才更接近股东真正获得的回报。

    对于新手投资者,不要只追逐AI热门概念,更应该选择能够持续赚钱、负债合理、现金流健康的企业。

    Reply
    Report
  • keke006
    ·11-11 14:42
    Gross margin is the filter here. AI revenue that stays above 60% margin and converts to free cash flow is what actually reaches shareholders
    Reply
    Report