Tsukanta's Individual Stock Analysis ② NVIDIA (NVDA)｜つかんた

The stock for this momentous second individual stock analysis is
NVIDIA (NVDA).

“Is NVIDIA already too expensive?”

On October 2, NVIDIA (NVDA) stock price rose to a high of $237.88, setting a new all-time high.
It has surpassed the high set in May for the first time in about four and a half months.
The closing price was $233.95, and the market capitalization is approximately $5.65 trillion.

Just three months ago, in July, voices were everywhere saying, “The AI market is over.”
This was a phase where the stock had fallen 16% from its May high, and about $1 trillion in market capitalization had vanished.

So, what has changed in these three months?
And how far into the future is the current stock price pricing in?

This time, I will break down NVIDIA one by one based on primary information such as the earnings press release announced on August 26, the CFO commentary, the quarterly report (10-Q), and statements made during the earnings call.

The conclusion and the A-D rankings are summarized at the end of the article.

First, let’s look at the latest earnings.
To start with the conclusion, the numbers were flawless.

The results for the second quarter of the fiscal year ending January 2027 (May–July 2026) are as follows.

Revenue: $96.2 billion (+106% year-over-year, +18% quarter-over-quarter)
Data Center Revenue: $89.0 billion (+117% year-over-year)
Gross Margin: 75.0%
Operating Income: $63.7 billion (+124% year-over-year)
Earnings Per Share (Adjusted): $2.22

Market expectations were around $92.1 billion in revenue and $2.10 in earnings per share, so both were exceeded.
This is the 14th consecutive quarter that the company has exceeded its own guidance.

However, what I want you to look at most in this earnings report is the “content” rather than the size of the revenue.

Starting this quarter, NVIDIA has begun disclosing data center revenue by splitting it into two categories.

The first is Hyperscale.
For major cloud providers and giant internet companies, it was $48.7 billion (+102% year-over-year, +13% quarter-over-quarter).

The second is ACIE (AI Cloud, Industrial, and Enterprise).
For emerging AI clouds, sovereign AI for various governments, and general enterprises, it was $40.3 billion (+138% year-over-year, +25% quarter-over-quarter).

The growth rate is higher for ACIE.
CEO Jensen Huang explains that while a year ago one AI research lab was driving investment, now multiple cutting-edge labs are expanding their scale in parallel.

In other words, NVIDIA’s revenue is shifting from “dependence on a few giant IT companies” to a “stage where the customer base is broadening.”
If customers are diversified, the damage when a specific company cuts back on investment becomes smaller.
For long-term holders, this is a positive change.

However, ACIE also includes emerging clouds with weak financial foundations.
I will touch on this point in detail in the risk section in the second half.

Next is the company’s guidance for the third quarter (August–October 2026).

Revenue: $108.0 billion (±2%)
Gross Margin: 74.0% (±0.5 percentage points)
Revenue from data center computing products for China is assumed to be zero

The market expectation was approximately $104.2 billion, so it has already exceeded that at the forecast stage.
And the next-generation “Vera Rubin” is expected to account for approximately 20% of data center revenue in the third quarter.
The company explains that growth in the third quarter will be driven primarily by ACIE, and that hyperscale demand will re-accelerate from the fourth quarter as the supply of Vera Rubin increases.

Finally, shareholder returns.
In the second quarter alone, approximately $26 billion was returned to shareholders through a combination of share buybacks and dividends.
The quarterly dividend was also significantly increased from $0.01 to $0.25 per share this past May.

In fact, what moved the stock price the most in this earnings report was not the second-quarter figures.

It was the outlook provided by CFO Colette Kress during the earnings call, projecting revenue growth of approximately 70% for the next fiscal year (fiscal year ending January 2028).
Moreover, this is explained as a figure that “incorporates supply constraints.”
She even stated that if customer plans were simply aggregated, the growth rate would be on a pace to double.

Prior analyst expectations for the next fiscal year were around +44-47%.
Since the revenue forecast for the current fiscal year is approximately $396 billion, a 70% increase would put the next fiscal year at a scale of approximately $670 billion.

Following this statement, earnings per share estimates for the next fiscal year were raised sharply.
From $12.82 to $15.68 just 30 days ago.
There were 42 upward revisions and zero downward revisions.

Let’s also confirm the figures behind this.

The CFO indicated a view that capital expenditures by the top five hyperscalers will reach approximately $800 billion in 2026 and approximately $1.3 trillion in 2027.
Furthermore, she explained that with the Vera Rubin generation, NVIDIA’s revenue opportunity per gigawatt of data center capacity will expand to approximately $40 billion.
This is because the business model has shifted to selling “entire factories,” including not just GPUs, but also CPUs, networking, and inference-specific chips.

So far, this is the good news.
And here is something you must keep in mind: the gross margin.

The company expects the gross margin to fall to 74% in the third quarter and 71-72% in the fourth quarter, bottoming out there, and then settling at 72-73% for the next fiscal year.
The main reason is the surge in memory prices.
The CFO explained that the rise in memory prices has exceeded expectations and will rise further heading into next year.
The plan is to recover that portion through price increases on products that have already been implemented.

In other words, NVIDIA is entering a phase where it will earn through “volume,” even if it means slightly sacrificing profit margins.
This means that even if revenue grows by 70%, profit growth will be slightly smaller than that.

This is not so much bad news as it is a plan the company has already explained.
The important thing is whether the fourth-quarter gross margin will fall below the “71%” the company mentioned.
This will be the yardstick for measuring the reliability of the 70% growth story for the next fiscal year.

There is one more small but important point.
Starting this fiscal year, NVIDIA has changed its adjusted earnings to include “stock-based compensation expenses.”
This is a conservative way of presenting figures, resulting in lower reported profits than before.
The earnings per share and P/E ratios in this article are all calculated using this conservative standard.

Here, let’s look back at the stock price trend.
NVIDIA is currently in a position where it has moved past the “period where the stock price was left behind despite earnings growth.”

May 14: Closing price of $235.47, the highest at the time.
June 5: Semiconductor stocks plummeted following Broadcom’s earnings. NVIDIA also fell about 6%.
Early July: Fell 16% from the high, with market capitalization down about $1 trillion. Expected P/E ratio hit 18x, the lowest since 2019.
Late July: Global semiconductor stocks fell across the board due to selling driven by doubts about the recovery of AI investments.
August 26: Earnings announcement. After falling as much as 3% in after-hours trading, it rebounded on the statement of 70% growth for the next fiscal year.
August 27: Rose 8.7%. The largest gain since April 2025.
September 16: The Fed raised interest rates by 0.25% (policy rate is 3.75-4.00%).
September 28: Added $150 billion to the share buyback program.
October 2: Hit an all-time high of $237.88 during the day. A recovery of about 25% from the late July low.

First, what you need to know is that the decline in July was not due to deteriorating performance.
Analyst earnings estimates were actually continuing to rise.
The market share for server GPUs also remained high at approximately 97% as of the end of 2025.
What was happening was a “rotation within the AI market,” where funds were moving from NVIDIA to memory stocks and others.

There is one more fact you should remember.
Before the August earnings, NVIDIA’s stock price had fallen after earnings in 6 of the previous 8 quarters.
Good earnings had become the norm, making it harder to surprise the market.
In August, that trend was broken by “future figures” in the form of the next fiscal year’s outlook.
This is a very clear example that what moves the stock price of a giant company is not “past performance” but “future outlook.”

Next, the market environment.

On September 16, the Fed implemented its first interest rate hike since 2023.
This was the first policy change under Chair Warsh, and many participants expect another rate hike within the year.
The next meeting is October 27-28.

A phase of rising interest rates is normally a headwind for growth stocks.
Even so, NVIDIA hit a new high.
This means that the speed at which earnings estimates are being raised is outweighing the weight of interest rates.

Then, on September 28, they announced an additional share buyback program of $150 billion, the largest in history.
The remaining authorization is $235 billion, which is planned to be executed during the fiscal year ending January 2028.
A policy to further increase dividends was also indicated (the amount and timing are yet to be determined).

Against a market capitalization of $5.65 trillion, $235 billion accounts for approximately 4%.
This is a sufficiently large scale to act as a floor for the stock price.

So, is the current stock price overvalued?
To conclude, looking at next year’s earnings, it is by no means expensive.

The calculation is simple.

Stock price: $233.95 (closing price on October 2)
Current fiscal year (FY ending Jan 2027) EPS forecast: approx. $9.3, P/E ratio approx. 25x
Next fiscal year (FY ending Jan 2028) EPS forecast: approx. $15.5–$15.7, P/E ratio approx. 15x

With the S&P 500’s overall forward P/E ratio exceeding 20x, a 15x multiple on a next-year basis is quite conservative.
For a company that claims its own sales will grow by 70%, this is a fairly modest valuation.

Let’s try to ‘reverse-calculate’ here.

Paying only 15 times next year’s earnings suggests that the market has to some extent priced in the possibility that ‘next year will be the peak of earnings, and growth will either stagnate or decline thereafter.’
AI investment will also eventually hit a peak.
The stock price is already anticipating about half of that peak.

Conversely, the moment the market begins to believe that growth will continue beyond the year after next, there remains room for the P/E ratio itself to expand.

Based on this, I will consider the stock price 12 months from now using three scenarios.

Bullish scenario (25% probability)
Supply constraints ease, and growth for the next fiscal year exceeds 70%.
EPS $17 × P/E 22x = approx. $374 (+60%)

Neutral scenario (50% probability)
70% growth as per company guidance, with gross margins landing at 72–73% next year.
EPS $15.5 × P/E 18x = approx. $279 (+19%)

Bearish scenario (25% probability)
Gross margins fall below 70%, and ACIE customers face funding bottlenecks.
EPS $13 × P/E 14x = approx. $182 (-22%)

Weighting the three by probability, the expected value is approximately $279.
This is around +19% from the current stock price.
Since the average target price from 61 analysts is $327.70, I am being more cautious than that.

The point is that even in the bearish scenario, the calculation does not result in the stock price being cut in half.
When comparing risk and return, the upside potential is greater.
The low range from late July (the $190 level) serves as one benchmark when considering the worst-case scenario.

From here on is the part I most want to convey this time.
In the back of the financial statements, there are numbers that are not often covered in the news.

First, the increase in accounts receivable.
Accounts receivable increased to $63.1 billion, and the Days Sales Outstanding (DSO) extended from 45 days in the previous quarter to 60 days.
The company explains that this is due to extended payment terms for large, multi-quarter contracts with investment-grade major customers.
As a result, operating cash flow was halved from $50.3 billion in the previous quarter to $24.1 billion.
Profits are increasing, but cash inflows are slowing down.
This is something I definitely want to check in the next earnings report.

Second, the surge in supply commitments.
Commitments for the purchase of components and production capacity more than doubled from $119 billion in the previous quarter to $279 billion.
Much of the content is for securing memory.
Inventory has also increased from $25.8 billion to $31.6 billion in preparation for the launch of Vera Rubin.
As long as demand continues, this is a strength as ‘preemptive securing’.
However, if demand collapses, it will turn directly into a burden.

Third, guarantees to support customers.
NVIDIA has undertaken credit enhancements of up to $105 billion for a data center site being developed by SB Energy in Ohio (approx. 4.25 gigawatts, a 20-year lease for OpenAI).
Including guarantees for AI cloud rental obligations, the maximum amount of guarantees is $108.5 billion.
The company explains that this single site alone represents a sales opportunity of $150 billion to $200 billion per GPU generation.
In other words, it is a structure where NVIDIA itself supports customer financing to generate sales.
The reason it is criticized as ‘circular investment’ is precisely because of this.

Fourth, China.
The quarterly report (10-Q) clearly states that they are effectively shut out of the Chinese data center market.
Shipments of Hopper products to China are less than 1% of data center revenue, and the company’s outlook also assumes zero.
This is a risk, but at the same time, if exports are resumed, it will be an immediate upside factor.

Fifth, competition.
Like Google’s TPUs, semiconductors designed in-house by major cloud providers are steadily increasing.
Just as semiconductor stocks plummeted following Broadcom’s earnings in June, the movements of these proprietary chip camps are factors that can shake stock prices.

Sixth, interest rates.
The Fed has set its 2026 inflation forecast (PCE) as high as 3.7%, leaving the possibility of additional rate hikes.
Since many AI data centers are built with borrowed money, rising interest rates directly impact customers’ investment capacity.

None of these are stories about ‘earnings collapsing immediately’.
However, it is a fact that one should be aware of that NVIDIA’s growth is, in some respects, being pulled forward using the balance sheets of both its customers and itself.

The next earnings report is scheduled for mid-November.
If it follows the usual pattern, it will be after the U.S. market close on Wednesday, November 18th, which is early morning on the 19th in Japan time (the company has not yet made an official announcement).

Before that, there is the FOMC on October 27-28.
Depending on the direction of interest rates, there is a possibility that the stock price will fluctuate significantly before the earnings report.

There are five points to watch in the earnings report.

First, how much revenue exceeds the company’s forecast of $108 billion (upper limit $110.2 billion).
Since it is a company that has consistently exceeded its own forecasts, hitting the upper limit might be perceived as ‘no surprise’.

Second, gross margin.
Will the third quarter land around 74%, and will the fourth quarter forecast fall within the 71-72% range?

Third, the ramp-up of Vera Rubin.
Is it going according to the plan of approximately 20% of data center revenue?
And are there signs that hyperscale demand is re-accelerating?

Fourth, the quality of cash.
Will the days sales outstanding (60 days) shrink or extend further?
Will operating cash flow catch up to profits?

Fifth, the 70% growth forecast for the next fiscal year.
Is it being maintained, or is it being raised?
This will be the biggest focus for the stock price.

Conversely, please consider that even if revenue exceeds expectations, if the gross margin floor falls below 71%, or if the tone of the next fiscal year’s forecast weakens, the stock price is likely to fall.
As proven in August, what moves this company’s stock price is not ‘this set of numbers’ but ‘the next set of numbers’.

Also, NVIDIA’s earnings are directly linked to Japanese semiconductor stocks.
The day after the August earnings report, NVIDIA’s rise pushed up the entire U.S. semiconductor sector.
For those holding Advantest or Tokyo Electron, this is an event that cannot be overlooked.

Finally, the conclusion.

NVIDIA has:
Revenue that has doubled from the previous year,
The company itself expects about 70% growth for the next fiscal year,
And yet, the P/E ratio based on next year’s earnings is about 15x.
Added to that is the largest share buyback in history.

On the other hand:
Gross margins are declining toward the fourth quarter,
Accounts receivable, supply commitments, and customer guarantees are expanding significantly,
And interest rates are rising.

It is a fact that we have entered a phase where we should verify not just the ‘quantity’ of growth, but the ‘quality’.
Even so, when comparing the strength of demand, the breadth of customers, and the undervaluation of the stock price relative to earnings, I judge that the value of holding it for the medium to long term is high.

Tsukanta’s rating is Rank A.

Rating Guide
A: Can be a core holding for the medium to long term
B: Consider buying with conditions
C: Wait and see
D: Avoid

However, the stock price is at an all-time high.
Rather than buying all at once, the basic approach is to buy in increments over time, spanning the FOMC and the November earnings report.
If the gross margin or cash figures collapse in the earnings report, I will review this rank.

*This article is for informational purposes only and does not recommend the buying or selling of any specific stock. Please make your own investment decisions.