NVIDIA Stock Down 17%. Is The AI Leader Still Expensive Or Undervalued?

On May 14, 2026, NVIDIA stock touched $236.54, a new all-time high. Six weeks later, it sits at $195.74, down roughly 17%, despite the company reporting what may be the most staggering single quarter in semiconductor history: $81.6 billion in revenue, a 25-fold dividend increase, and an $80 billion buyback authorization. 

Jensen Huang himself called the stock's decline a "mystery" in public. That word choice says a lot. A CEO who built one of the most valuable companies on earth does not usually reach for "mystery" when he is calm about the situation. This article unpacks what is actually happening, why good news is not lifting NVIDIA right now, and whether the math at $195 supports the argument that this dip is worth paying attention to.

Let's break down the forces pulling NVIDIA lower, whether they represent a structural threat or a temporary headwind, how the current valuation compares to history, and what a realistic range of outcomes looks like from here.

Why NVIDIA Stock Fell After Record Q1 FY2027 Earnings

The numbers from Q1 FY2027 are not ambiguous. Revenue hit $81.6 billion, up 85% year-over-year. Data center revenue reached $75.2 billion, up 92%. Non-GAAP earnings per share came in at $1.87 against a Wall Street consensus of $1.76. Free cash flow for the quarter was $48.6 billion. These are not metrics from a company in trouble.

And yet NVIDIA's stock fell on every single trading day between May 20, when the results were announced, and May 27. Then on June 5, a broader sector sell-off knocked it further.

This is not unusual behavior for NVIDIA. The phenomenon has a name in market analysis: "buy the rumor, sell the news." NVDA had already risen roughly 13.7% from its February earnings through its May 14 all-time high. By the time the actual results confirmed what the market had already priced in, there was limited marginal buying left. The bar had been raised so high that even a genuine beat registered as "expected."

What makes this instance different is that the selling has not stopped at the post-earnings flush. Something else is adding weight.

Why Is NVIDIA Stock Down 17%? Key Reasons Explained

1. China: A Market Already Lost

On May 21, 2026, Jensen Huang stated publicly that NVIDIA has "largely conceded" China's advanced AI chip market to Huawei. The statement confirmed what was already visible in the earnings data: NVIDIA's Q1 FY2027 data center revenue from China was effectively zero. For context, that same segment generated $4.6 billion in Q1 FY2026, one year earlier. Estimates from Bloomberg suggest Huawei's Ascend chips now command approximately 55% or more of China's domestic AI chip market, while NVIDIA's share, once near 95%, has fallen to the single digits.

The China market, for NVIDIA's purposes, is no longer a near-term revenue story. It is a risk that has already materialized and been absorbed.

2. Export Controls Are Getting Tighter, Not Looser

In early June 2026, the U.S. Department of Commerce moved to close a loophole allowing NVIDIA's Rubin and Blackwell chips to reach Chinese AI companies through offshore subsidiaries. This was not a new ban on the scale of the H20 restrictions from 2025. But direction matters in regulatory analysis. Analysts at JPMorgan and Bernstein have estimated the cumulative impact of export restrictions and domestic Chinese competition could represent a revenue headwind of $5.5 billion to $16 billion over the coming fiscal cycle.

3. The GPU Rental Index: A Signal Most Retail Investors Are Missing

The price to rent a single B200 GPU on major cloud platforms peaked at $6.11 per hour on May 30, 2026. By June 21, that price had fallen to $4.22 per hour, a drop of roughly 31% in three weeks. This metric, tracked via Ornn's compute dashboard and embedded in prediction market contracts on platforms like Kalshi, functions as a real-time proxy for AI workload demand.

Think of it like an index for shipping freight rates. When those rates spike, it tells you global trade is accelerating. When they fall, something has slowed down. GPU rental prices work similarly: rising prices mean AI companies are competing for limited training and inference capacity; falling prices suggest either supply is catching up, or demand from AI model development is temporarily pausing.

A 31% drop in compute rental prices in under a month is the kind of signal that institutional traders track carefully, and it is one reason why NVIDIA has not found a floor yet despite fundamentally strong earnings.

4. NVIDIA Is the Sector's Biggest Laggard This Year

NVIDIA is up around 3.65% year-to-date as of late June 2026. The VanEck Semiconductor ETF (SMH) is up roughly 70% over the same period. Micron is up nearly 284%, AMD is up around 138%, and the sector as a whole has been one of the strongest performing groups in global equity markets. NVIDIA, the company that made AI infrastructure possible, is the worst-performing major semiconductor stock this year.

The explanation is partly valuation math. NVDA entered 2026 already priced at a significant premium to its peers after its extraordinary 2023 and 2024 returns. When the broader sector re-rated upward on AI enthusiasm, NVDA had less room to appreciate because the market had already given it credit for a lot of its future growth. This is not a red flag; it is a feature of any stock that has run as far and fast as NVIDIA has.

The Dual-50 Shift: NVIDIA’s Customer Base Is Getting More Diversified

Twelve months ago, the most serious structural concern around NVIDIA was customer concentration. A handful of hyperscalers, primarily Microsoft, Google, Amazon, and Meta, accounted for more than half of NVIDIA's data center revenue. The fear was straightforward: if any one of those companies paused its AI infrastructure build-out, NVIDIA would feel it in the next quarter's numbers.

That risk profile has changed in a way that is not being discussed enough.

In Q1 FY2027, NVIDIA disclosed that hyperscaler revenue settled at approximately 50% of total data center revenue. The other 50% came from what NVIDIA now calls ACIE: AI cloud providers, industrial companies, enterprise customers, and sovereign AI deployments. Countries including the UAE, Saudi Arabia, and multiple European nations are building national AI infrastructure on NVIDIA hardware. Enterprise adoption of on-premise AI systems has accelerated. AI-native cloud companies like CoreWeave and Nebius are scaling their GPU capacity aggressively.

NVIDIA's demand base is now wider than at any point in its history. The $91 billion guidance for Q2 FY2027 was issued with zero China revenue assumed. The non-China business, on its own, is growing at a pace that supports a new quarterly record. That is a meaningfully different position than where NVIDIA stood a year ago.

NVIDIA Q1 FY2027 Results: Key Numbers Investors Should Know

Source: NVIDIA Q1 FY2027 Earnings Release, May 20, 2026. SEC Form 8-K. NVIDIA Investor Relations.

The Q2 revenue guidance of $91 billion is the number that carries the most analytical weight here. NVIDIA provided this estimate while explicitly stating it does not assume any data center chip sales to China. The implication is that the non-China business, which now includes sovereign AI, enterprise deployments, and AI cloud infrastructure globally, is growing fast enough to carry a new quarterly record by itself.

NVIDIA Analyst Price Targets: What Wall Street Expects

Sources: Benzinga, TipRanks, Yahoo Finance, Mitrade. Data as of June 2026.

Of 37 Wall Street analysts covering NVIDIA as of late June 2026, 36 rate it a Buy or Strong Buy. One rates it a Hold. There are zero Sell ratings in the analyst community. The consensus price target from TipRanks stands at around $309, representing roughly 58% implied upside from the current $195.74 level.

The spread between the lowest target ($215, Deutsche Bank) and the highest ($500, Baird) is unusually wide. That range reflects genuine uncertainty, not about the company's current trajectory, but about what the right multiple to apply to a nearly $5 trillion hardware company with 22x forward earnings looks like in a period of rate volatility and AI investment normalization. The directional consensus is uniformly bullish; the magnitude question is open.

A Quick Valuation Check: Is NVIDIA Stock Cheap? NVDA Forward P/E And PEG Ratio Explained

Standard financial analysis for high-growth technology companies uses two tools together: the price-to-earnings (P/E) ratio and the price-earnings-to-growth (PEG) ratio.

The P/E ratio tells you how much the market is paying for each dollar of current earnings. On its own, a P/E of 22 might seem high for a hardware company. But the P/E ratio does not account for how fast earnings are growing. The PEG ratio adjusts for that. A PEG of 1.0 means you are paying exactly one unit of P/E for every percentage point of annual earnings growth. Below 1.0, the growth rate justifies the valuation multiple. Well above 1.0, the market is expecting you to pay a premium for growth that may or may not materialize.

Source: GuruFocus, StockAnalysis.com, FullRatio.com, FinanceCharts.com, Yahoo Finance, SEC Q1 FY2027 8-K filing. Valuation data as of June 23, 2026.

At around $195, NVIDIA is trading at its cheapest forward earnings multiple in several years. Its 10-year average P/E has been around 54x. Its 5-year average has been approximately 62x. Today it sits at roughly 22x. The stock has not been this cheap relative to its own earnings history since before the AI era began in earnest.

The PEG ratio of 0.44 is what makes the current level analytically interesting. For comparison, AMD is trading at roughly 97x forward earnings according to a Motley Fool analysis from June 25, 2026. The same article posed a pointed question: someone's valuation model is wrong. At 22x forward earnings with a PEG under 0.5, NVIDIA is the cheaper company by any growth-adjusted measure.

One important caveat. The PEG ratio assumes the high earnings growth rate continues. That assumption is what the China headwinds, the compute price decline, and the export control tightening are all putting pressure on. This framework supports the dip thesis mathematically, but it does not eliminate the downside scenario.

NVIDIA Bear Case: Risks That Could Push The Stock Lower

Every strong bull thesis needs a credible counter-scenario. Here is when the current analysis breaks down:

• If B200 GPU rental prices on cloud platforms continue declining toward $3 or below per hour in H2 2026, it would suggest AI model training activity has genuinely plateaued rather than paused for the Vera Rubin upgrade cycle. That would directly challenge the revenue trajectory that underpins NVIDIA's $91 billion Q2 guidance.
• If one or more major hyperscalers, Microsoft, Google, Amazon, or Meta, meaningfully reduces AI infrastructure capex in its next earnings report, the non-China demand story weakens. All four have guided substantial AI spending increases. Any reversal would likely be felt in NVIDIA's next results.
• If the Vera Rubin production ramp encounters supply issues similar to the early Blackwell delays of 2024 and 2025, FY2028 revenue models would need downward revision at precisely the moment markets are pricing in the upgrade cycle.
• If U.S. export controls expand beyond China to restrict Rubin-generation chips to additional countries, the total addressable market for NVIDIA's most advanced hardware shrinks further. The direction of policy is already clear; the magnitude and scope of future restrictions is not.

None of these scenarios are the base case embedded in the current analyst consensus. But they are the precise risks worth monitoring in the 30 to 60 days ahead, because how each of them develops will likely determine whether NVIDIA finds a floor at current levels or continues lower.

Our Take: What NVIDIA’s Stock Fall Means For Investors

At $195, NVIDIA trades at its most attractive forward earnings multiple in years, against a backdrop where the company has already absorbed its single biggest known headwind, China revenue, into its forward guidance. The $91 billion Q2 estimate explicitly assumes zero data center sales to China. Whatever that drag is, it is already in the model.

The PEG ratio of 0.44 is not a headline most investors associate with a stock that has become a synonym for AI infrastructure. It reflects a situation where earnings are growing faster than the market is willing to give the multiple credit for, partly because the China narrative creates a discount, and partly because the stock's previous run leaves it in the curious position of being simultaneously the sector's most recognized name and one of its cheapest on a growth-adjusted basis.

Vera Rubin entering production in H2 2026, with AWS, Google Cloud, Microsoft Azure, and CoreWeave among the first deployers, gives NVIDIA a product cycle catalyst that could shift the narrative before the next earnings report in late August. The demand base, having diversified from hyperscaler concentration to a roughly 50/50 split with sovereign and enterprise AI customers, is structurally more resilient than it was a year ago.

The risks are real and specific: B200 compute prices are falling, export controls are tightening, and the GPU rental index is signaling caution. These deserve monitoring. But at $195, with a PEG of 0.44 and a 37-analyst consensus pointing to $309, the math of the current level is something any investor in global AI infrastructure should understand before deciding how to respond.