01 — 🏢 Understanding the Business

NVIDIA has changed dramatically.

It is no longer simply a company selling GPUs.

Its Data Center platform combines GPUs, CPUs, networking, systems and software, with CUDA and the surrounding ecosystem creating a significant competitive advantage.

And Data Center is now overwhelmingly the business.

In FY2026, NVIDIA generated approximately $193.7B of Data Center revenue, compared with:

🎮 $16.0B Gaming
🖥️ $3.2B Professional Visualization
🚗 $2.3B Automotive

Data Center therefore represented roughly 90% of total revenue.

The concentration has increased even further.

During the first six months of FY2027, Data Center revenue reached approximately $164B, with $89B in Q2 alone.

The NVIDIA investment thesis is therefore increasingly an AI infrastructure thesis.

And NVIDIA continues to strengthen its position with every new product generation.

Blackwell is still ramping, while Vera Rubin has already entered production shipments.

Jensen Huang also highlighted something I found particularly interesting on the earnings call:

The revenue opportunity per gigawatt of AI infrastructure has increased from approximately:

➡️ $18B with Hopper
➡️ $25B with Grace Blackwell
➡️ $40B with Vera Rubin

This means NVIDIA isn't necessarily dependent on simply selling more GPUs.

If each new AI factory contains more NVIDIA technology, the revenue opportunity per data center can continue increasing.

There is also meaningful growth outside the traditional hyperscalers.

In Q2, approximately:

🤝 $49B of Data Center revenue came from hyperscalers
🤝 $40B came from AI clouds, industrial and enterprise customers

That second group grew 138% YoY.

It includes AI startups, neoclouds, enterprises and sovereign AI projects.

NVIDIA expects its neocloud partners to exit 2026 with approximately 8 GW of installed capacity, compared with around 3 GW at the end of 2025.

That's important because NVIDIA's future isn't entirely dependent on Microsoft, Amazon, Google and Meta continuing to increase CapEx at today's extraordinary rate.

02 — 💰 Analyzing Financial Health

The financial statements remain one of the strongest parts of the NVIDIA story.

Q2 revenue increased 106% to $96.2B.

Operating income increased 124% to $63.7B.

Net income increased 126% to $59.7B.

Diluted EPS was $2.46, up 128%.

Gross margin was 75%.

These are extraordinary numbers.

But I don't think today's margins should simply be projected forever.

NVIDIA currently expects gross margin of approximately:

📉 74% in Q3
📉 71–72% in Q4
📉 72–73% in FY2028

One of the pressures is memory pricing. Management said memory costs have increased significantly more than previously expected.

Even a 72% gross margin would be exceptional.

The point is simply that today's 75% shouldn't automatically become our long-term assumption.

💪 Balance sheet

As of July 26, 2026, NVIDIA had approximately:

💵 $22.4B cash
💵 $34.1B marketable debt securities
💵 $42.8B marketable equity securities
📥 $63.1B accounts receivable
📦 $31.6B inventory
💳 $33.4B debt

Total assets were approximately $320.3B, with shareholders' equity of approximately $229B.

The financial position is extremely strong.

There are nevertheless some things I want to keep watching.

Accounts receivable has increased substantially, reaching $63.1B.

NVIDIA reports Days Sales Outstanding of 60 days and explains that some large investment-grade customers have received extended payment terms because of the size and timing of their data-center purchases.

I don't consider this a problem by itself.

But when a company is growing this quickly, I want to make sure reported revenue continues converting into cash.

Inventory also increased to $31.6B, from approximately $21.4B at FY2026 year-end.

NVIDIA says part of this increase is related to preparing for the Vera Rubin launch.

More interestingly, supply and capacity commitments increased from approximately:

⚠️ $119B → $279B

in just one quarter.

That's an enormous increase.

It makes sense given the current demand environment. NVIDIA wants to secure the manufacturing capacity and components necessary to satisfy customers.

But there is another side to this.

These commitments are based on expectations for future AI demand.

If demand remains strong, securing supply is a competitive advantage.

If demand eventually slows unexpectedly, the company could face excess inventory, higher costs or unused capacity.

For now, the cash generation gives NVIDIA a very large cushion.

During the first six months of FY2027, NVIDIA generated approximately:

💵 $74.4B operating cash flow
💵 $69.9B free cash flow

FY2026 free cash flow was approximately $96.6B.

That's an incredible amount of cash for any company, let alone one that is still growing this quickly.

China is a different issue.

NVIDIA says H200 shipments to China represented less than 1% of Data Center revenue in Q2.

Its current outlook assumes no China Data Center compute revenue.

So the immediate revenue impact is relatively small.

The longer-term concern is competition.

If NVIDIA cannot participate fully in China's AI market, Chinese competitors have more opportunity to develop their own hardware and software ecosystems.

03 — 📊 Perform DCF Valuation

This is where I deliberately want to be realistic.

NVIDIA's recent growth rates are extraordinary.

Management expects approximately 70% revenue growth in FY2028, and says that outlook is still constrained by supply.

But I don't believe it makes sense to assume anything close to that growth rate for the next decade.

At some point, NVIDIA becomes too large.

So my DCF assumes strong growth for several years, followed by a gradual normalization.

I estimate approximately $391B of revenue in FY2027.

From there:

📈 FY2028: +35% → $528B
📈 FY2029: +20% → $633B
📈 FY2030: +12% → $709B
📈 FY2031: +8% → $766B
📈 FY2032: +5% → $804B

These are my assumptions, not NVIDIA guidance.

And even this is a very bullish outcome.

I'm effectively assuming NVIDIA eventually becomes an $800B annual-revenue company.

For free cash flow, I assume margins gradually decline:

💵 FY2027: 42% → $164B FCF
💵 FY2028: 41% → $216B
💵 FY2029: 40% → $253B
💵 FY2030: 39% → $277B
💵 FY2031: 38% → $291B
💵 FY2032: 36% → $289B

Again, these are my assumptions.

I'm not assuming NVIDIA becomes an average semiconductor company.

I'm assuming it remains one of the most profitable businesses in the world.

I'm simply not assuming today's economics continue unchanged forever.

Using:

🔹 10% WACC
🔹 3% terminal growth

I get:

Enterprise Value: ~$3.46T

After adding approximately $66B of net liquid financial assets:

Equity Value: ~$3.52T

Using approximately 24.1B shares:

🎯 DCF value: ~$146/share

NVIDIA is currently around $217.55/share, implying a market capitalization of approximately $5.25T.

The sensitivity is also important:

At 9% WACC → ~$171/share

At 10% WACC → ~$146/share

At 11% WACC → ~$128/share

The market is clearly assuming a better outcome than my base case.

And the market could absolutely be right.

NVIDIA could grow faster than I expect.

Margins could remain higher.

AI infrastructure spending could remain enormous for much longer.

But at today's valuation, there isn't much room for disappointment.

04 — 👔 Assess Leadership & Capital Stewardship

NVIDIA's management has an exceptional execution record.

The company has successfully moved through several major product generations while simultaneously expanding into networking, CPUs, systems and software.

The compensation structure is also heavily linked to performance.

Approximately 96% of the CEO's target compensation was performance-based in FY2026.

That's something I generally like to see.

Capital allocation is becoming more interesting as NVIDIA gets larger.

During Q2 alone, NVIDIA returned approximately:

💵 $20B through share repurchases
💵 $6B through dividends

That's approximately $26B returned to shareholders in one quarter.

During the first half of FY2027, NVIDIA repurchased approximately $39.8B of its own shares.

But NVIDIA is doing more than buybacks.

As of July 26, 2026, NVIDIA had approximately:

💰 $99B of equity investments
💰 $25B of equity investment commitments

It has also made approximately $36B of cloud-service commitments, generally over six years.

And NVIDIA has entered preliminary arrangements intended to mobilize more than $500B of third-party capital for AI infrastructure.

There is also significant credit support connected to major AI infrastructure projects, including a potential aggregate obligation of up to approximately $105B related to the SB Energy/OpenAI project, subject to specific conditions.

Strategically, this makes sense.

NVIDIA benefits when more AI infrastructure gets built.

Helping fund or support the ecosystem can therefore create future demand for NVIDIA's products.

But this is also an area where I want to see the actual returns.

NVIDIA is increasingly using its balance sheet to shape the AI ecosystem around itself.

That could be an excellent capital allocation decision.

It also introduces additional risk that wasn't nearly as important when NVIDIA was primarily selling chips and buying back shares.

05 — 🧠 Build the Investment Thesis

NVIDIA doesn't need to become a bad company for the stock to disappoint. It only needs to grow somewhat less than the market expects.

My DCF assumes NVIDIA eventually reaches approximately $804B in annual revenue and generates almost $300B in annual free cash flow. That's already a very strong outcome. Yet at a 10% discount rate, I arrive at approximately $146/share, versus roughly $217 today.

The company can absolutely outperform my model. Enterprise AI could grow faster, inference could create another huge wave of demand, and Rubin and future architectures could increase NVIDIA's share of each AI data center.

But as a fundamental investor, I don't want to pay today's price based on the assumption that the extraordinary growth of the last few years will continue indefinitely.

NVIDIA remains an exceptional business. At the current valuation, however, I would want a better margin of safety.