Nvidia Review: Sequential Revenue Decline In The U.S. (Rating Downgrade)

Funding Thesis

NVIDIA Company’s (NASDAQ:NVDA) Q2 earnings report on the floor confirmed the tech firm firing on all cylinders, with revenues and EPS surpassing each prime and bottom-line expectations, respectively. The chip large’s non-GAAP EPS of $0.68 beat estimates by $0.04, whereas income reached $30.04 billion, above expectations by $1.31 billion. Nvidia’s Information Heart income hit a file $26.3 billion, up 154% yr over yr. The corporate additionally issued steerage for FY Q3 2025 with income anticipated to be $32.5 billion, which is barely above the consensus of $31.75 billion​.

Regardless of what seem like robust figures, Nvidia’s shares fell after hours Wednesday, reflecting what, I believe, is the market sniffing outgrowing issues which have been simmering.

For individuals who have not adopted my protection of Nvidia, I’ve develop into more and more skeptical concerning the chipmaker since I began writing about them final fall. My issues (first identified in November) on competitors appear to be coming true. The Blackwell chip delay appears to be the catalyst that precipitated the market share disruption I’ve been warning about all alongside.

I believe the market is adjusting for potential challenges that Nvidia is dealing with below the hood, similar to delayed plans to ship billions of value of Blackwell chips and issues about income peaking in america (as famous of their 10-Q and on the earnings name).

Whereas Nvidia continues to be assured concerning the This fall supply of Blackwell chips, the market’s skepticism is certainly greater than it was pre-earnings. There is no doubt in my thoughts that Nvidia will work out these chips and scale manufacturing. The query is that if these chips will ship within the quantity they estimate on the This fall timeline they plan. What’s attention-grabbing concerning the admissions from the earnings name is that not one of the issues a couple of “three-month delay” like I lined earlier than earnings appeared to have been dispelled. I (from my analysis) have stated that Blackwell chips would ship late this Fiscal yr. Administration confirmed this on the decision. However administration was obscure.

With out Blackwell, Nvidia will possible discover it harder to compete in opposition to Superior Micro Gadgets, Inc. (AMD) given their MI 300X chips at the moment are ramping manufacturing in key use circumstances.

Nvidia is closely depending on Blackwell’s success going ahead. I believe traders have to weigh the dangers and contemplate the influence of a protracted delay.

With this, Hopper gross sales (the rapid income resolution to this delay) really noticed their gross sales peak in america (the place the plurality of worldwide Information Facilities are situated) sequentially 2 quarters in the past. Whereas general Hopper gross sales elevated, the sequential decline in income on this key market (I believe) runs fully opposite to the Nvidia narrative that demand for Hopper is powerful the place the massive prospects are. Most of their massive Hyperscaler prospects (Massive tech) are in america. They (on the entire) are shopping for much less.

Regardless of my bearish stance on Nvidia heading into earnings, I really anticipated barely stronger Hopper efficiency this previous quarter, given their (I am going to admit) robust momentum in AI and information heart. I didn’t count on to see the sequential drop-off in Hopper gross sales simply but.

With this, whereas some noticed the quarter as a beat, I believe we even have some structural points beneath. I’m downgrading my view on Nvidia from a promote to a powerful promote.

Why I am Doing Observe-Up Protection

Final month, I wrote pre-earnings protection that highlighted issues I had about Blackwell chip delays and cautioned traders about potential income disruptions stemming from delays in Nvidia’s upcoming Blackwell chips. With the discharge of their Q2 outcomes, my issues had been partially mirrored in commentary on the decision.

Regardless of administration’s admission that the Blackwell chip design needed to be reengineered (the muse of why the chips had been delayed), administration supplied restricted readability on the exact size of the delays, which I believe left many traders with lingering issues. Administration talks about billions of of income from Blackwell chips in This fall FY 2025, however didn’t specify a quantity. Is that this $3 billion? $8 billion? This chip is the way forward for Nvidia in some ways. The shortage of readability was irritating.

With this, they’ve deflected issues by speaking about Hopper gross sales. I’ll go into this later, however I’m not certain concerning the sturdiness of this demand.

As with earlier than earnings, I proceed to be a contrarian to the market with my protection. I do know I’m severely diverging from the bullish consensus on Wall Avenue. I believe the little particulars add up right here. The purpose of my follow-up protection is to point out why I’m now extra bearish and to offer traders one other viewpoint.

What Involved Me About The Quarter

As with most quarters for Nvidia over the past two years, one of many key highlights of their Q2 outcomes was the file Information Heart income of $26.3 billion, which grew by 16% (on the entire) from the earlier quarter and 154% from the prior yr.

We have now to dive into these particulars to get a greater view of those gross sales, nonetheless.

Of their most up-to-date 10-Q submitting, it is now clear that gross sales in North America have really begun to say no, which could possibly be attributed to potential market saturation, ready for the Blackwell sequence, and elevated competitors inside AI and information facilities (see my analysis on AMD). That is in contrast sequentially to FY 25 Q1.

When requested about this on the decision, Nvidia’s CFO responded that the place a buyer’s income is booked is probably not the place the client finally ends up utilizing the product (as a deflection for why US income declined sequentially):

That is not essentially the place the product will ultimately be, and the place it might even journey to the end-customer. These are simply transferring to our OEMs, our ODMs, and our system integrators for essentially the most half throughout our product portfolio.

What’s notable about this quote, nonetheless, is that she is referring to her Authentic Tools Producers (OEMs) & ODMs (Authentic Design Producers). Assume firms like Tremendous Micro Pc, Inc. (SMCI) and Dell Applied sciences Inc. (DELL). This doesn’t seem like the massive tech firms we consider which have been among the greatest consumers similar to Microsoft Company (MSFT) or Meta Platforms, Inc. (META).

So, a sequential income decline can solely be partially defined by OEMs and ODMs shifting purchases to non-US areas. I, personally, assume ODM demand is much less sustainable than direct-to-user demand like that of Massive Tech. ODMs are technically resellers of Nvidia chips. In the event that they see demand loosen up, they may lower orders to maintain stock lean. Massive tech has already indicated (from their earnings calls) that they may overbuild for now.

Word as properly that Nvidia’s CFO on the decision supplied no assertion to analysts on whether or not this sequential drop can be only a blip or a part of a brand new development. My bearish perception (therefore this text) is that this isn’t only a blip. She additionally implies this sequential gross sales drop is because of GPU/datacenter gross sales (vs. automotive or different class gross sales) as a result of she stated it was firms who construct information heart servers that had been shifting their invoices.

Moreover my issues about Hopper gross sales, the Blackwell chip delays had been touched on in the course of the earnings name, and not one of the executives had been particular concerning the nature of the defect.

As an alternative, CFO Kress & Huang said:

We executed a change to the Blackwell GPU mass to enhance manufacturing yields. Blackwell manufacturing ramp is scheduled to start within the fourth quarter and proceed into fiscal yr ’26. -Kress

The change to the masks is full. There have been no purposeful modifications mandatory. And so we’re sampling purposeful samples of Blackwell — Grace Blackwell in quite a lot of system configurations as we converse -Huang.

It is bullish for Nvidia that they’re determining easy methods to scale the Blackwell chip. What shouldn’t be bullish (for my part) is the extent of vagueness on the magnitude of gross sales we’ll see from Blackwell this yr.

Valuation

Regardless of the selloff following earnings, Nvidia shares proceed to command a valuation far exceeding the sector median. Nvidia’s ahead P/E ratio is roughly 41.33, which represents a 73.91% premium over the sector median ahead P/E of 23.77.

Whereas progress general is powerful, I actually assume their prolonged P/E a number of over the sector median actually raises issues now concerning the sustainability of the place the inventory is. That is very true in an atmosphere the place progress expectations are being scrutinized by analysts and traders extra rigorously. I, personally, discovered the analyst questions on this earnings name to be way more probing and particular than earlier earnings calls.

Given Nvidia’s key efficiency within the US, I believe there’s a robust case to be made for a extra conservative strategy to valuing their shares. A extra sensible stance, I believe, would advocate for Nvidia buying and selling at only a 25% premium to the sector median ahead P/E.

If Nvidia’s ahead P/E had been adjusted to a 25% premium, this might suggest a ahead P/E of roughly 29.72, in comparison with the present ahead P/E of 41.33. Particularly, if the market adopted by and re-rated shares for this adjustment, it could lead to a draw back of roughly 28.10% for shares on prime of the post-earnings drop. I really imagine the market might have overpriced Nvidia’s progress prospects relative to the sector​.

Bull Thesis

Through the earnings name, Huang emphasised that over $1 trillion value of AI server infrastructure must be upgraded. He famous that as servers transfer away from CPUs to accelerated computing powered by GPUs, there is a rising demand for extra environment friendly and highly effective computing options​. He stated on the decision:

…the world is transferring from basic function computing to accelerated computing. And the world builds about $1 trillion ’ value of knowledge facilities — $1 trillion ’ value of knowledge facilities in a number of years will likely be all accelerated computing. Prior to now, no GPUs are in information facilities, simply CPUs. Sooner or later, each single information heart can have GPUs. And the rationale for that’s very clear is as a result of we have to speed up workloads in order that we will proceed to be sustainable, proceed to drive down the price of computing in order that once we do extra computing our — we do not expertise computing inflation.

We’re at the start of our journey to modernize a $1 trillion ’ value of knowledge facilities from general-purpose computing to accelerated computing.

If the chance is as huge as instructed, it begs the query of why america, which homes extra information facilities inside its borders than some other nation, had income that dropped, not to mention slowed notably final quarter.

I get that some ODM income may need been booked abroad, however within the US it simply does not make sense that income would even come near peaking if the chance was so robust. GPU demand is seasonality agnostic (so this isn’t a seasonal fluctuation) and in keeping with Huang, we’re at the start of a pc revolution.

With this, whereas the potential $1 trillion alternative is undoubtedly large (and possibly I could possibly be incorrect it could possibly be this massive), the corporate’s present market capitalization is roughly thrice the dimensions of this complete alternative. Often, the whole addressable market, or TAM, of your market is larger than the online current worth of future income from the chance (the market cap).

Do not get me incorrect: the AI revolution is right here and Nvidia will play a giant half in it. I simply assume that Nvidia’s shares seem overvalued regardless of the alternatives in AI.

The discrepancy between the potential market and Nvidia’s present valuation suggests (to me) that the inventory value already costs at an unrealistic degree of success, leaving little room for error or market disruptions.

Backside Line

Whereas beating estimates on the highest and backside strains, Nvidia Company’s valuation stays a priority to me, particularly given the sequential decline of their United States income. With no Blackwell gross sales anticipated for this present quarter, the near-term outlook seems bearish (and this might compound into the valuation).

The mix of those components – doubtlessly peaking US demand, and delayed new product contributions – reinforces my view that Nvidia’s shares are overvalued, prompting my robust promote perception.