← 市场观察
2026-11-04
高利率、高油价与AI开支:2026年底的美股全景
AI只是其中一条线。看懂整个市场,要先看利率、油价和资金在板块之间怎么流动
一句话结论:今年的美股不是一个“AI行情”,而是高利率、高油价和AI开支三股力量同时作用的结果。短期最重要的变量是10年期美债收益率;长期更适合为几种不同情景做准备,而不是押单一结局。
一、现在是什么样的市场
利率是所有资产定价的锚,而它在近20年的高位。10年期美债收益率在5%上下,9月升破5%,创2007年以来新高;10月一度摸到5.36%,是2002年以来最高。中期选举结果出来后,美债反弹、收益率回落,这是选后股市反转的重要原因之一。
油价和通胀让美联储重新加息。中东冲突让油价长期在每桶90美元上下,通胀预期随之抬升。美联储9月加息25个基点至3.75%–4.00%,是2023年以来首次加息,市场押12月再加一次。
板块分化非常大。到10月初,能源和科技是标普500里表现最好的两个板块,今年以来都涨了约37%;医疗同期只涨了约7%。10月下旬之后,科技经历了一轮明显回调,标普半导体指数的远期市盈率降到14.7倍,已低于标普500的18.9倍。
三股力量合在一起,决定了今年的市场形态:
| 力量 | 推高了什么 | 压制了什么 |
|---|
| 高利率 | 现金和短债的收益 | 高估值成长股、小盘股、房地产 |
| 高油价 | 能源股 | 消费、运输,以及通胀预期下的降息希望 |
| AI资本开支 | 半导体、电力电网、数据中心供应链 | 云厂商自身的自由现金流 |
二、短期(到明年春天):看哪些信号
季节性对股市有利,但今年多了一个加息周期。历史上,中期选举后的六个月(11月到次年4月)是四年总统周期里最强的时段,标普500平均上涨约14%。但历史平均没有考虑加息周期,不能照搬。
短期最值得看的是以下几个信号,市场观察页上方的仪表盘每个交易日自动更新:
- 10年期美债收益率:回落到5%以下,大部分股票都会松一口气;重新冲上5.3%以上,任何反弹都难以持续。
- 油价:决定通胀预期,进而决定美联储会不会继续加息。
- 等权标普(RSP)对标普500(SPY):等权指数跑赢,说明上涨在向更多股票扩散,反弹更健康;只有少数大盘股在涨,反弹就比较脆弱。
- 小盘股(IWM)对标普500:小公司最怕高利率,因为它们的债务多是浮动利率。小盘股领涨,说明市场开始相信利率见顶。
- VIX:市场恐慌程度。急升后回落往往伴随阶段性低点,但不是精确的信号。
三、长期(1到5年):四种情景
与其押一个结局,不如看清几种可能,以及每种情景下哪些资产受益。
| 情景 | 什么情况下发生 | 相对受益的资产 | 相对受损的资产 |
|---|
| AI继续扩张 | 云厂商2027年资本开支逼近1万亿美元 | AI供应链瓶颈公司、电力与电网 | 被AI冲击的软件和服务 |
| 高通胀长期化 | 油价和工资压力不退 | 能源、实物资产、有定价权的公司 | 长期债券、高估值成长股 |
| 增长放缓,利率回落 | 加息把经济压下去 | 长期国债、医疗、必需消费,利率回落后的小盘股 | 周期股、能源 |
| AI开支见顶 | 云厂商削减资本开支 | 防御板块、现金、债券 | 半导体、电力、数据中心链条 |
几点观察:
- 能源和电网是少见的“两种情景都能赢”的板块:同时受益于高通胀和AI用电。但今年已经大涨,估值不再便宜。
- 10年期收益率接近5%,是近20年来第一次能用长期国债锁定这么高的收益。万一经济衰退,债券还能起对冲作用。
- 医疗今年明显落后于大盘,叠加人口老龄化的长期需求,是值得持续观察的防御板块。
- 金融板块受益于高利率下的利差,但新国会可能收紧监管,私募信贷对AI数据中心的敞口也值得留意。
四、主要风险清单
- 10年期收益率突破5.5%:股票估值会被系统性下压,成长股受伤最重。
- 油价再次飙升:通胀反弹,美联储被迫加息更多次。
- AI资本开支见顶:科技、电力、半导体会一起回调,它们之间的相关性比看起来高。
- 私募信贷出现问题:这一轮AI建设有相当部分资金来自私募信贷,一旦违约,风险会传导到金融板块。
五、怎么使用市场观察页
- 宏观仪表盘:六个数字,每个附一句“为什么重要”。不需要每天看,每周看一次变化方向就够了。
- 板块轮动表:11个板块的涨跌、离高点多远,旁边是本站自己的数据——近30天各板块内部人的有效买入和卖出。“股价在跌、内部人却在买”的板块会被标出来,那往往值得进一步研究,但它不是买卖信号。
- 文章:对某个主题的长篇分析,例如AI供应链里哪些公司真正绕不开。
本文仅为研究笔记和思考框架,不构成任何投资建议。文中数据截至2026年11月4日,投资有风险,请独立判断。
资料来源
← 市場觀察
2026-11-04
高利率、高油價與AI開支:2026年底的美股全景
AI只是其中一條線。看懂整個市場,要先看利率、油價和資金在板塊之間怎麼流動
一句話結論:今年的美股不是一個“AI行情”,而是高利率、高油價和AI開支三股力量同時作用的結果。短期最重要的變數是10年期美債殖利率;長期更適合為幾種不同情景做準備,而不是押單一結局。
一、現在是什麼樣的市場
利率是所有資產定價的錨,而它在近20年的高位。10年期美債殖利率在5%上下,9月升破5%,創2007年以來新高;10月一度摸到5.36%,是2002年以來最高。中期選舉結果出來後,美債反彈、收益率回落,這是選後股市反轉的重要原因之一。
油價和通膨讓聯準會重新升息。中東衝突讓油價長期在每桶90美元上下,通膨預期隨之抬升。聯準會9月升息25個基點至3.75%–4.00%,是2023年以來首次升息,市場押12月再加一次。
板塊分化非常大。到10月初,能源和科技是標普500裡表現最好的兩個板塊,今年以來都漲了約37%;醫療同期只漲了約7%。10月下旬之後,科技經歷了一輪明顯回撥,標普半導體指數的遠期市盈率降到14.7倍,已低於標普500的18.9倍。
三股力量合在一起,決定了今年的市場形態:
| 力量 | 推高了什麼 | 壓制了什麼 |
|---|
| 高利率 | 現金和短債的收益 | 高估值成長股、小盤股、房地產 |
| 高油價 | 能源股 | 消費、運輸,以及通膨預期下的降息希望 |
| AI資本開支 | 半導體、電力電網、資料中心供應鏈 | 雲廠商自身的自由現金流 |
二、短期(到明年春天):看哪些訊號
季節性對股市有利,但今年多了一個升息週期。歷史上,中期選舉後的六個月(11月到次年4月)是四年總統週期裡最強的時段,標普500平均上漲約14%。但歷史平均沒有考慮升息週期,不能照搬。
短期最值得看的是以下幾個訊號,市場觀察頁上方的儀表盤每個交易日自動更新:
- 10年期美債殖利率:回落到5%以下,大部分股票都會鬆一口氣;重新衝上5.3%以上,任何反彈都難以持續。
- 油價:決定通膨預期,進而決定聯準會會不會繼續升息。
- 等權標普(RSP)對標普500(SPY):等權指數跑贏,說明上漲在向更多股票擴散,反彈更健康;只有少數大盤股在漲,反彈就比較脆弱。
- 小盤股(IWM)對標普500:小公司最怕高利率,因為它們的債務多是浮動利率。小盤股領漲,說明市場開始相信利率見頂。
- VIX:市場恐慌程度。急升後回落往往伴隨階段性低點,但不是精確的訊號。
三、長期(1到5年):四種情景
與其押一個結局,不如看清幾種可能,以及每種情景下哪些資產受益。
| 情景 | 什麼情況下發生 | 相對受益的資產 | 相對受損的資產 |
|---|
| AI繼續擴張 | 雲廠商2027年資本開支逼近1萬億美元 | AI供應鏈瓶頸公司、電力與電網 | 被AI衝擊的軟體和服務 |
| 高通膨長期化 | 油價和工資壓力不退 | 能源、實物資產、有定價權的公司 | 長期債券、高估值成長股 |
| 增長放緩,利率回落 | 升息把經濟壓下去 | 長期國債、醫療、必需消費,利率回落後的小盤股 | 週期股、能源 |
| AI開支見頂 | 雲廠商削減資本開支 | 防禦板塊、現金、債券 | 半導體、電力、資料中心鏈條 |
幾點觀察:
- 能源和電網是少見的“兩種情景都能贏”的板塊:同時受益於高通膨和AI用電。但今年已經大漲,估值不再便宜。
- 10年期殖利率接近5%,是近20年來第一次能用長期國債鎖定這麼高的收益。萬一經濟衰退,債券還能起對沖作用。
- 醫療今年明顯落後於大盤,疊加人口老齡化的長期需求,是值得持續觀察的防禦板塊。
- 金融板塊受益於高利率下的利差,但新國會可能收緊監管,私募信貸對AI資料中心的敞口也值得留意。
四、主要風險清單
- 10年期殖利率突破5.5%:股票估值會被系統性下壓,成長股受傷最重。
- 油價再次飆升:通膨反彈,聯準會被迫升息更多次。
- AI資本開支見頂:科技、電力、半導體會一起回撥,它們之間的相關性比看起來高。
- 私募信貸出現問題:這一輪AI建設有相當部分資金來自私募信貸,一旦違約,風險會傳導到金融板塊。
五、怎麼使用市場觀察頁
- 宏觀儀表盤:六個數字,每個附一句“為什麼重要”。不需要每天看,每週看一次變化方向就夠了。
- 板塊輪動表:11個板塊的漲跌、離高點多遠,旁邊是本站自己的資料——近30天各板塊內部人的有效買入和賣出。“股價在跌、內部人卻在買”的板塊會被標出來,那往往值得進一步研究,但它不是買賣訊號。
- 文章:對某個主題的長篇分析,例如AI供應鏈裡哪些公司真正繞不開。
本文僅為研究筆記和思考框架,不構成任何投資建議。文中資料截至2026年11月4日,投資有風險,請獨立判斷。
資料來源
← Market watch
2026-11-04
High rates, high oil and AI spending: the US stock market at the end of 2026
AI is only one thread. To read the whole market, start with rates, oil and how money moves between sectors
In one sentence: this year's market is not simply an "AI rally" — it is the combined result of high rates, high oil and AI spending. In the short run the most important variable is the 10-year Treasury yield; in the long run it makes more sense to prepare for several scenarios than to bet on one outcome.
1. What kind of market this is
Interest rates anchor the price of every asset, and they are near a 20-year high. The 10-year Treasury yield is around 5%: it broke above 5% in September, the highest since 2007, and touched 5.36% in October, the highest since 2002. After the midterm results, Treasuries rallied and yields fell — one of the main reasons stocks reversed higher after the vote.
Oil and inflation have put the Fed back on a hiking path. The Middle East conflict has kept oil around $90 a barrel, lifting inflation expectations. The Fed raised rates 25 bp to 3.75%–4.00% in September, its first hike since 2023, and markets expect another in December.
Sectors have diverged sharply. By early October, energy and technology were the two best S&P 500 sectors, each up about 37% for the year, while health care was up only about 7%. Since late October tech has pulled back hard; the S&P semiconductor index's forward P/E has fallen to 14.7x, below the S&P 500's 18.9x.
Together, the three forces explain the shape of this year's market:
| Force | Lifted | Held back |
|---|
| High rates | Returns on cash and short-term bonds | Richly valued growth stocks, small caps, real estate |
| High oil | Energy stocks | Consumer and transport stocks, and hopes for rate cuts |
| AI capital spending | Semiconductors, power and grid, the data-center supply chain | Cloud providers' own free cash flow |
2. Short term (through next spring): signals to watch
Seasonality favors stocks, but this year adds a hiking cycle. Historically, the six months after a midterm election (November to April) are the strongest stretch of the four-year presidential cycle, with the S&P 500 up about 14% on average. Those averages do not include a hiking cycle, so they should not be applied mechanically.
The signals most worth watching are below; the dashboard at the top of the Market Watch page updates them every trading day:
- 10-year Treasury yield: a move back below 5% would relieve most stocks; a return above 5.3% would make any rally hard to sustain.
- Oil: drives inflation expectations, and with them whether the Fed keeps hiking.
- Equal-weight S&P (RSP) vs the S&P 500 (SPY): when equal weight leads, gains are spreading to more stocks and a rally is healthier; when only a few giants rise, it is fragile.
- Small caps (IWM) vs the S&P 500: small companies fear high rates most because much of their debt floats. Small caps leading suggests the market believes rates have peaked.
- VIX: how fearful the market is. A spike followed by a fall often comes with a short-term low, but it is not a precise signal.
3. Long term (1–5 years): four scenarios
Rather than betting on one outcome, it helps to lay out several, and which assets each favors.
| Scenario | When it happens | Relative winners | Relative losers |
|---|
| AI keeps expanding | Cloud providers' 2027 capex approaches $1 trillion | AI supply-chain bottleneck companies, power and grid | Software and services disrupted by AI |
| Inflation stays high | Oil and wage pressure persist | Energy, real assets, companies with pricing power | Long-term bonds, richly valued growth stocks |
| Growth slows, rates fall | Hikes push the economy down | Long-term Treasuries, health care, consumer staples, small caps once rates fall | Cyclicals, energy |
| AI spending peaks | Cloud providers cut capex | Defensive sectors, cash, bonds | Semiconductors, power, the data-center chain |
A few observations:
- Energy and the grid are rare sectors that win in two scenarios: they benefit from both high inflation and AI power demand. But after this year's gains they are no longer cheap.
- With the 10-year near 5%, long-term Treasuries offer the highest locked-in yield in about 20 years, and they can still hedge a recession.
- Health care has clearly lagged this year; combined with long-term demand from an aging population, it is a defensive sector worth watching.
- Financials benefit from wider lending margins at high rates, but the new Congress may tighten regulation, and private credit's exposure to AI data centers deserves attention.
4. Main risks
- The 10-year yield breaks above 5.5%: stock valuations would be pushed down across the board, with growth stocks hurt most.
- Oil spikes again: inflation rebounds and the Fed is forced to hike more.
- AI capital spending peaks: tech, power and semiconductors would fall together; they are more correlated than they look.
- Trouble in private credit: a large share of this AI build-out is funded by private credit; defaults would spread to financials.
5. How to use the Market Watch page
- Macro dashboard: six numbers, each with one line on why it matters. There is no need to check daily; the weekly direction is enough.
- Sector rotation table: each of the 11 sectors' returns and distance from its high, next to this site's own data — effective insider buying and selling in each sector over the last 30 days. Sectors where prices are falling but insiders are buying are flagged. They may deserve a closer look, but this is not a trading signal.
- Articles: long-form analysis of a single theme, such as which companies the AI supply chain truly cannot route around.
This is a research note and thinking framework, not investment advice. Data as of November 4, 2026. Investing involves risk; please make your own judgment.
Sources