# ARGUS investment-judgment example · 2026-10-13 snapshot

This is a frozen illustration of the Skill's output, based on ARGUS public records retrieved on 2026-10-13. It is not a live trade signal or an independent re-audit of every original source. Horizon: 3–6 months. No personal portfolio or current price/valuation supplied.

## 中文示例

**结论：暂不新增 AI 基建主题仓位。研究优先级偏向能以经营现金流覆盖现金资本开支的企业，暂不把债务驱动的扩张当作买入理由。对经营分化判断的信心为中等；当前入场价吸引力尚无法判断。**

| 事实 | 日期与口径 | 对判断的影响 |
| --- | --- | --- |
| 微软 OCF−现金 CapEx 为 +19.639 十亿美元 | 2026-06-30 单季度，合并业务；ARGUS 标记一手核验 | 能覆盖该季度现金资本开支，但不等于 AI 项目的回报或完整偿债能力 |
| Amazon 同口径为 −8.821 十亿美元，前季 −18.171 | 2026-06-30 / 2026-03-31，合并业务；ARGUS 标记一手核验 | 现金资本开支仍超过经营现金流，但缺口缩小，是反对“一直恶化”的证据 |
| 台积电月收入 5118.57 亿新台币，前月 5148.06 | 2026-09 / 2026-08，合并收入、未经审计；ARGUS 标记一手核验 | 不能仅凭单月收入差异认定 AI 订单崩塌，也不能证明 AI 资本回报已实现 |

**推导路径：**ARGUS 对 Damodaran 的导读将“供应商获得收入”与“建设方获得资本回报”拆开：建设方支出成为供应商收入，但只有未来增量现金收益覆盖资本成本，才证明建设创造价值。上述现金流分化说明，同一行业扩张不能作为所有公司增加配置的共同依据。ARGUS 对 Wolf Richter 的导读提示另一条传导路径：高股价支持股权融资，股价下跌可能削弱再融资缓冲。这是风险假设，当前取到的数据不能证明已发生违约。

**最强反对证据：**Amazon 的现金缺口较前季收窄；10 月 12 日报告也判断融资压力尚未传到硬件订单。因此，“全链条立即崩塌、应全面卖出”同样缺乏支持。季度现金流存在季节性，两期比较不足以确认长期趋势。

**行动：**先补齐目标公司的当前估值、最新订单和资本回报证据，再考虑增加配置；研究候选偏向自我融资能力更强的公司。这是研究优先级，不是微软优于 Amazon 的股票收益排序。不据此新增杠杆；未提供持仓和损失预算，不给仓位百分比或现有持仓卖出指令。

**改变判断的条件：**
- 升级为“可条件增加配置”：可核验订单兑现、后续可比期间现金转化改善，并且基于明确现金流/资本成本假设的估值留有缓冲。具体估值门槛须重新计算。
- 转向更谨慎：融资失败导致项目实际延期/取消，并在随后两个有效订单窗口看到采购削减；或供应商利润率、回款同时恶化。这是本次研究采用的观察规则，不是自动交易阈值。
- 下一次核验：目标公司下一次有效财报/订单披露及其正式指引。未核验日程，不编造具体发布日期。

**情景：**基准是需求持续、融资筛选加剧，维持等待与选择性研究；上行情景是订单兑现和资本回报改善，在估值满足条件后再行动；下行情景是融资压力传导到采购和现金回收，降低研究偏好，并结合实际持仓重新评估敞口。

## English example

**Decision: do not add new AI infrastructure theme exposure yet. Prioritize research on firms able to fund cash capital spending from operating cash flow; debt-led expansion alone is not a buy signal. Confidence in operating differentiation is moderate; entry-price attractiveness is unassessed.**

Microsoft's consolidated quarterly OCF minus cash CapEx was +$19.639bn at June 30, 2026. Amazon's was −$8.821bn, improving from −$18.171bn in the preceding quarter. TSMC recorded September consolidated revenue of NT$511.857bn, versus NT$514.806bn in August. ARGUS labels these observations primary-verified; the figures are not AI-only economics, and TSMC's monthly revenue is unaudited.

The ARGUS Damodaran guide separates suppliers' revenue from builders' eventual return on capital: spending becomes revenue elsewhere, but only future incremental cash returns exceeding capital costs establish value creation. The Richter guide proposes that high equity valuations help financing, and falling prices may weaken that buffer. This is a hypothesis, not evidence of a current default.

Counterevidence: Amazon's cash deficit narrowed, and the October 12 report does not yet see financing pressure passing into hardware orders. Neither blanket buying nor an immediate industry-collapse claim follows. Seasonality and only two comparable quarters limit trend confidence.

Action: obtain current valuation, order evidence and return-on-capital assumptions before adding exposure. The financing preference is a research watchlist criterion, not a stock-return ranking. No new leverage, price target or position size follows from this evidence alone.

Upgrade only with verified order conversion, improving comparable cash conversion and valuation supported by stated capital-cost assumptions. Reconsider if financing failures cause actual project delays and procurement cuts over subsequent effective order windows, or margins and collections deteriorate together. Recheck at the next effective company disclosure; do not invent a release date. Base/upside/downside scenarios respectively support waiting, conditional entry, or a portfolio-specific exposure review.

## Sources and coverage

- [Microsoft metric and calculation scope](https://argusx.fun/metrics/msft-cash-after-capex/)
- [Amazon metric and calculation scope](https://argusx.fun/metrics/amzn-cash-after-capex/)
- [TSMC metric](https://argusx.fun/metrics/supplier-revenue/) · [official monthly source](https://investor.tsmc.com/english/monthly-revenue/2026)
- [Damodaran ARGUS editorial guide, 2026-10-08](https://argusx.fun/perspectives/damodaran-ai-earnings-and-cashflows/)
- [Richter ARGUS editorial guide, 2026-10-09](https://argusx.fun/perspectives/richter-ai-debt-and-credit-spreads/)
- [AI infrastructure report, 2026-10-12](https://argusx.fun/research/ai-infrastructure/2026-10-12/daily/)

The guides are ARGUS editorial paraphrases; no claim is made to reproduce or independently confirm the full original essays. This frozen example must be recomputed before use as a current recommendation.