Nvidia Is Being Mispriced Again: Blackwell Demand Looks Worth More Than the Market Assumes

The market focused on the miss, not the full quarter

Nvidia still delivered revenue of $46.7 billion and non-GAAP diluted earnings per share of $1.05. The clear miss was data center revenue: $41.1 billion versus about $41.3 billion expected. Shares fell roughly 2% after the report, which looks more like a reaction to mixed signals than evidence of a demand break.

That reaction is easy to understand. A small segment shortfall can overshadow beats on the top and bottom lines, especially when the prior quarter's $4.5 billion H20 charge was still fresh. In that kind of setup, investors often fixate on the most visible risk and treat everything else as secondary.

Still, the operating picture was not as clean a warning as the stock drop suggested. Blackwell Data Center revenue grew 17% sequentially, and China's impact on the quarter looked more like a disruption than a broad demand problem.

Why the reaction looked behavioral

The market often wants a simple verdict when a giant like Nvidia misses on one key number: either the AI spending story is intact, or it is cracking. This quarter, the report was messier. A prior China-related hit, weaker-looking margins, and a small data center shortfall sat beside stronger overall revenue, EPS, and Blackwell growth. That kind of mixed signal can push investors toward the more negative frame.

Margins were distorted by the prior H20 hit

The clearest example is gross margin. Nvidia reported non-GAAP gross margin of 61.0%, but that was before excluding the prior H20 damage. Excluding that $4.5 billion charge raises the figure to 71.3%. That gap says more about how one-time policy damage can distort first impressions than about Blackwell's underlying economics.

China stayed in the story, even without H20 sales

The same anchoring is visible in how investors are reading China exposure. Last quarter, Nvidia took the $4.5 billion charge after export rules changed and missed out on an additional $2.5 billion of H20 revenue. This quarter, there were no H20 sales to China-based customers. At the same time, management said Blackwell Data Center revenue grew 17% sequentially. The China issue stayed in the headlines, but the larger engine kept moving.

Client concentration is a watchpoint, not proof of a break

About half of data center revenue came from large cloud service providers. Bulls can read that as evidence that enterprise-scale AI buildouts are still happening. Bears can read it as added risk if hyperscaler spending slows. Both readings are reasonable. The problem comes when investors treat that mix and customer concentration as permanent traits rather than variables that could change.

That is the core mispricing argument here. Variable policy risk and mix questions are being weighed more heavily than current demand. But Blackwell's sequential growth is the more direct read-through on what is happening now.

What to watch into the next earnings report

This is a watchlist setup, not a blind ownership case.

The next checkpoint is the Aug. 26 earnings release

The timing is less important than the signal. Nvidia's next earnings report is confirmed for Aug. 26, and a prior earnings preview noted a 6.5% stock price range in either direction in options pricing. The useful question is not simply whether the company beats or misses estimates. It is whether Blackwell demand continues to look durable enough to outweigh policy noise.

Management has already pointed to a Q3 revenue projection of $54.0 billion plus or minus 2%. If that guidance holds up, it becomes harder to treat the quarter as evidence of a demand break.

A rerating likely needs three things

A stronger case for mispricing would be supported by all three of the following: - Blackwell demand remains evident in reported revenue, not just in management commentary. - The revenue guide stays credible without needing special adjustments. - China-related issues stop dominating the interpretation of results that are otherwise strong.

The risk is getting clever too early

That is a high bar. If Blackwell demand looks temporary, or if margins need constant clean-room adjustments to look healthy, then the market may be underestimating the risk rather than the opportunity.

So the setup is selective, not automatic. For now, the argument is that the stock reaction overemphasized a mixed quarter. Whether that translates into a rerating depends on whether the next report confirms that Blackwell demand is still outrunning the noise.