5月20日收盘后，NVIDIA公布了Q1 FY2027财报。从任何客观标准来看，这份财报都是出色的。
| 指标 | 实际值 | 华尔街预期 | 超出幅度 |
|---|
| EPS | $1.87 | $1.76 | +6.3% |
| 营收 | $816亿 | $792亿 | +3.0% |
| 数据中心收入 | $380亿（超大规模客户） | — | 环比+12% |
| AI云收入 | — | — | 同比增长超3倍 |
在电话会议上，黄仁勋说了一句将被引用很多个季度的话：
「需求已经进入抛物线状态。Agentic AI已经到来。」
然后股价下跌了。
这个矛盾——一份全面超预期的财报之后出现抛售——才是当前唯一值得读懂的故事。
为什么超预期之后股价反而跌？
这不是NVIDIA出了什么问题。这是一个经典的市场机制，叫做「买预期、卖事实」（Buy the Rumor, Sell the News）。
来看具体发生了什么：
- 财报前几周，市场已经将NVIDIA超预期的概率计入价格。股价提前上涨，隐含波动率（IV）处于高位。
- 财报公布的瞬间，不确定性消失。支撑高IV的恐惧溢价在一夜之间崩塌。
- 「只是超预期」已经不够了。 对于一家高估值公司，市场需要的不是刚好超过预期——而是指引必须远超所有人的想象。
这就是高期望值的宿命。每一次超预期，都在抬高下一次的门槛。NVIDIA没有让市场失望，是市场的期望已经跑在了现实前面。
两个不能忽视的背景
背景一：超大规模客户资本支出正在爆发
四大超大规模云客户——Alphabet、Amazon、Microsoft、Meta——对2026年的合计资本支出指引约为7,250亿美元，较上年增长77%。NVIDIA的订单能见度已超过1万亿美元，延伸至2027年。
这不是投机性需求。这是全球最大公司们的确定性基础设施投入。它们都在押注同一件事：AI算力将决定未来十年的竞争优势。
背景二：中国市场被定价为零——这是一份隐藏的期权
NVIDIA的业绩指引中明确假设中国市场收入为零。45亿美元H20库存减值已经一次性处理。中国市场已被完全排除在所有前瞻性预测之外。
这意味着什么：任何中美关系的缓和——哪怕是出口限制的部分放松——都将释放不存在于任何分析师模型中的收入增量。这是一个纯上行的免费期权，没有任何人在定价。市场已经充分定价了中国市场的最坏情况。任何偏离最坏情况的变化，都会直接转化为上行空间。
财报已过，现在怎么做？
当前：IV正在快速下降——评估现有仓位
财报后的IV crush正在全面展开。如果你在财报前持有的卖Put仓位已接近50%利润，IV crush正在加速你的盈利兑现。这是评估提前平仓的时机——不是继续死扛的理由。
操作：审查所有未平仓位。利润达到50%的，认真考虑平仓或滚仓。
下一步：等IVR重新达标再进场
IV crush结束后，重新评估隐含波动率排名（IVR）。IVR低于50%时不要进场。
等待下一个催化剂推高IV：
- GTC大会公告
- 新品发布
- 超大规模客户资本支出更新
- 任何地缘政治噪音
每一个事件都可能短暂推高IVR，为有纪律的期权卖方创造丰厚的权利金窗口。
时间线：NVDA下次财报——8月26日
从现在到7月底，你有一个DTE 30–45天的可操作交易窗口。在下次财报前平掉所有仓位，彻底消除跨财报风险。
新仓位进场条件
| 参数 | 规则 |
|---|
| IVR | ≥ 50% |
| Delta | 0.20–0.25 |
| 单仓规模 | ≤ 组合的5% |
| 止损 | 收取权利金的2倍 |
| 进场时DTE | 30–45天 |
| 财报重叠 | 无——8月26日前平仓 |
核心结论
黄仁勋说需求是抛物线。但抛物线的股价不需要你追。
等它调整。等IVR达标。系统性进场。NVIDIA的机会不会消失——但最好的风险收益比来自于耐心，而不是FOMO。
After market close on May 20, NVIDIA reported its Q1 FY2027 earnings. The results were, by any objective measure, exceptional.
| Metric | Actual | Wall Street Estimate | Beat |
|---|
| EPS | $1.87 | $1.76 | +6.3% |
| Revenue | $81.6B | $79.2B | +3.0% |
| Data Center Revenue | $38B (hyperscalers) | — | +12% QoQ |
| AI Cloud Revenue | — | — | >3x YoY |
On the earnings call, CEO Jensen Huang delivered a statement that will be quoted for quarters to come:
"Demand has entered a parabolic state. Agentic AI is here."
Then the stock dropped.
That contradiction — a blowout quarter followed by a sell-off — is the only story worth understanding right now.
Why Does a Stock Drop After Beating Expectations?
This isn't a sign that something is wrong with NVIDIA. It's a mechanical pattern known as "Buy the Rumor, Sell the News."
Here's what happened, step by step:
- Weeks before earnings, the market priced in a high probability that NVIDIA would beat estimates. The stock rallied. Implied Volatility (IV) climbed to elevated levels.
- The moment earnings dropped, uncertainty vanished. The fear premium that was sustaining high IV collapsed overnight.
- "Just beating estimates" is no longer enough. For a company trading at premium valuations, the market demands guidance that shatters every expectation — not merely meets them.
This is the curse of high expectations. Each beat raises the bar for the next one. NVIDIA didn't fail the market. The market's expectations had simply outrun reality.
Two Background Factors You Can't Ignore
Factor 1: Hyperscaler CapEx Is Exploding
The four largest hyperscale cloud customers — Alphabet, Amazon, Microsoft, and Meta — have guided approximately $725 billion in combined capital expenditure for 2026, a 77% increase year-over-year. NVIDIA's order visibility now exceeds $1 trillion, extending into 2027.
This isn't speculative demand. This is committed infrastructure spending by the largest companies on Earth, all betting that AI compute capacity will determine competitive advantage for the next decade.
Factor 2: China Is Priced at Zero — And That's a Hidden Option
NVIDIA's forward guidance explicitly assumes zero revenue from China. The $4.5 billion H20 inventory write-down has already been taken as a one-time charge. The Chinese market has been surgically removed from every forward projection.
Here's why that matters: any improvement in U.S.-China relations — even a partial easing of export restrictions — would unlock revenue that exists in no analyst model. It's a pure upside option that nobody is pricing in. The market has fully priced the worst-case scenario for China. Any deviation from that worst case flows directly to the upside.
Earnings Are Over. What Now?
Right Now: IV Is Collapsing — Evaluate Your Positions
Post-earnings IV crush is in full effect. If you held short put positions before earnings and they're approaching 50% profit, IV crush is accelerating your gains. This is the time to evaluate early closure — not a reason to hold on stubbornly.
Action: Review open positions. If profits have reached 50%, seriously consider closing or rolling.
Next Step: Wait for IVR to Reset Before Re-Entry
Once IV crush completes, reassess Implied Volatility Rank (IVR). Do not enter new positions when IVR is below 50%.
Wait for the next catalyst to push IV higher:
- GTC Conference announcements
- New product launches
- Hyperscaler CapEx updates
- Any geopolitical noise
Each of these events can temporarily spike IVR, creating fat premium windows for disciplined options sellers.
Timeline: Next NVIDIA Earnings — August 26
From now through late July, you have an actionable window for DTE 30–45 day trades. Close all positions before the next earnings report to eliminate cross-earnings risk entirely.
Entry Conditions for New Positions
| Parameter | Rule |
|---|
| IVR | ≥ 50% |
| Delta | 0.20–0.25 |
| Position Size | ≤ 5% of portfolio |
| Stop Loss | 2× premium received |
| DTE at Entry | 30–45 days |
| Earnings Overlap | None — close before 8/26 |
The Bottom Line
Jensen says demand is parabolic. But a parabolic stock doesn't require you to chase it.
Wait for the pullback. Wait for IVR to reach your threshold. Enter systematically. The opportunity in NVIDIA is not going away — but the best risk-adjusted entries come from patience, not FOMO.
