In partnership with
The Bottom Line Upfront 💡
$NVDA ( ▼ 1.51% ) NVIDIA is one of the greatest businesses ever built, powered by a CUDA software moat that took decades to construct. But at $210-$230, the stock already prices in years of near-perfect execution. Great company, demanding price.
Sponsorship
Your Competitors Just Hired an Entire GTM Team. No Humans.
AI agents are doing the work of full GTM teams. Pipeline, content, customer service. The startups using them aren't waiting for headcount approvals.
Get the free Practical Guide to Agentic GTM for Startups and see exactly where to start.
Strata Layers Chart
Layer 1: The Business Model 🏛️
NVIDIA started in 1993 making graphics chips for video games. Then it accidentally invented the backbone of modern AI.
Here is the core insight: a GPU (graphics processing unit) is built to do millions of simple calculations simultaneously. That parallel processing power, originally designed to render pixels, turns out to be exactly what you need to train AI models. NVIDIA figured this out early, built a software platform called CUDA in 2006 to let developers tap into that power, and has been printing money ever since.
Today NVIDIA sells two things:
Compute and Networking (92% of revenue): Data center GPUs, CPUs, networking chips, and the full software stack. Think of it as selling the entire engine room of the AI economy, not just one part.
Graphics (8% of revenue): GeForce RTX gaming GPUs and workstation cards. Still profitable, still growing, but clearly the supporting act now.
The real moat is CUDA. Over 7.5 million developers worldwide write code for CUDA, and switching to a competitor means rewriting everything. That is not a moat, that is a castle with a crocodile-filled moat and a drawbridge.
Key Takeaway: NVIDIA sells the picks and shovels of the AI gold rush, and it also owns the only mine shaft that most developers know how to use.
Layer 2: Category Position 🏆
NVIDIA powers over 78% of the world's top 500 supercomputers. Its Blackwell Ultra chips are so in demand that customers are signing multi-year supply agreements and paying deposits just to get in line. That is not a competitive market, that is a waiting list.
Competitors exist, but none are close:
AMD is the most credible challenger with its MI-series AI accelerators, gaining some share but still a distant second.
Intel re-entered discrete GPUs and has struggled to gain traction.
Hyperscalers (Google, Amazon, Microsoft, Meta) are building custom chips for their own internal workloads. This is the real long-term threat: if your biggest customers become your biggest competitors, that is a problem worth watching.
Huawei is building alternatives for the Chinese market, which NVIDIA has been effectively locked out of due to U.S. export controls.
The China situation deserves a flag. Export restrictions have cut China from roughly 7% of revenue to a rounding error, and NVIDIA openly admits this has helped competitors build ecosystems there. That is a real, ongoing cost.
Key Takeaway: NVIDIA dominates AI compute with a software ecosystem that is genuinely hard to displace, but the hyperscaler custom-silicon threat is the one risk that could actually matter over a five-year horizon.
Layer 3: Show Me The Money! 📈
Revenue by segment (H1 FY2027):
Compute and Networking: $162.9B ↗️ (+101% YoY)
Graphics: $15.0B ↗️ (+52% YoY)
Revenue by customer type (Q2 FY2027):
Hyperscale (Amazon, Google, Microsoft, Meta): $48.7B ↗️ (+102% YoY)
AI Clouds, Industrial, and Enterprise: $40.3B ↗️ (+138% YoY)
Edge Computing (gaming, workstations): $7.2B ↗️ (+27% YoY)
Geography: The U.S. accounts for 62% of revenue. Taiwan (home to key customers and TSMC) is 28%. China is now just 8% and falling, down from 35% of international revenue in prior periods.
Margins are exceptional. Gross margin hit 75% in H1 FY2027, up from 66.6% a year ago. Operating margin is 66%. Net margin is 66.4%. For context, most semiconductor companies would celebrate a 50% gross margin. NVIDIA is lapping them.
Cash generation is staggering. Operating cash flow was $74.4B in just the first half of FY2027. The company returned $39.8B to shareholders via buybacks and raised its quarterly dividend from $0.01 to $0.25 per share in May 2026. It also issued $25B in new debt in June 2026, not because it needed cash, but to lock in long-term financing for its ecosystem investment strategy.
One wrinkle: NVIDIA has committed to $279B in supply and capacity obligations through FY2032, plus $108.5B in guarantees (including a $105B guarantee backing OpenAI's data center buildout in Ohio). These are big numbers. If demand softens, those commitments become a headache.
Key Takeaway: NVIDIA is generating cash at a rate that makes most companies look like lemonade stands, but it is also making enormous forward bets on sustained AI infrastructure demand.
Layer 4: Long-Term Valuation (DCF Model) 💰
The Verdict: Fairly Valued to Modestly Overvalued
Scenario | Fair Value | vs Current Price ($210-$230) |
|---|---|---|
Conservative (14.5% WACC) | $63-$83 | -65% to -70% |
Base Case (10% WACC) | ~$197 | roughly flat |
Optimistic (9% WACC) | ~$274 | +20% to +30% |
FMP Model Estimate | $248 | +8% to +18% |
The wide range here is not a bug, it is the feature. NVIDIA's valuation is almost entirely a bet on your discount rate assumptions. Use a textbook CAPM rate and the stock looks wildly expensive. Use a market-implied rate that reflects NVIDIA's franchise quality and it looks roughly fair.
Key assumptions driving the spread:
At $210-$230, the market is pricing in roughly a 9-10% discount rate and 3-3.5% terminal growth, which implies NVIDIA becomes a durable infrastructure utility.
The base case assumes revenue grows at roughly 28% annually for five years before normalizing. That is a lot to ask of a company already doing $177B in six months.
Gross margin compression is a real risk as competitive pressure builds and packaging costs stay elevated.
One-line take: NVIDIA is a phenomenal business trading at a price that requires things to keep going very right for a very long time.
Layer 5: What Do We Have to Believe? 📚
Bull Case 🚀
The AI infrastructure buildout is a multi-decade capital cycle, not a two-year fad, and NVIDIA stays the default choice throughout.
The CUDA ecosystem is sticky enough that even as hyperscalers build custom chips, they keep buying NVIDIA for the workloads their custom silicon cannot handle.
Robotics, autonomous vehicles, and physical AI (via the DRIVE and Isaac platforms) add a meaningful new revenue layer by FY2029-2030.
Bear Case 🐻
Hyperscaler custom silicon (Google TPUs, Amazon Trainium, Microsoft Maia) gradually displaces NVIDIA for a growing share of training and inference workloads, compressing both volume and pricing power.
The AI capex cycle peaks sooner than expected, leaving NVIDIA holding $279B in supply commitments and a demand cliff.
Export controls expand beyond China, cutting off additional markets and handing competitors a permanent geographic advantage.
The Bottom Line: NVIDIA is one of the most impressive businesses ever built, with a software moat that took decades to construct and financial results that look like a typo. The question is not whether the company is great. The question is whether the stock at $210-$230 already prices in all the greatness, plus some. For long-term investors with high conviction in the AI infrastructure thesis, the risk/reward is acceptable. For everyone else, patience for a better entry point is not the worst idea.
Layer 6: What to Watch 👀
Hyperscaler capex guidance each quarter. If Amazon, Google, Microsoft, or Meta signals a slowdown in data center spending, NVIDIA's order book gets complicated fast.
Rubin architecture ramp (H2 FY2027). The next-generation platform promises a 10x cost-per-token improvement over Blackwell. A smooth ramp sustains momentum. A stumble creates an air pocket.
Custom silicon adoption rates. Watch for any public disclosures from hyperscalers about the percentage of AI workloads running on internal chips versus NVIDIA hardware.
China policy developments. Any easing of export restrictions is a meaningful upside catalyst. Any tightening is a headwind. The regulatory environment here changes fast.
Gross margin trajectory. If margins slip below 70%, it signals either pricing pressure or product mix deterioration. Both matter for the long-term earnings story.
AI-written, human-approved
Disclaimer: This guide is for informational purposes only and does not constitute financial advice, investment recommendations, or an offer or solicitation to buy or sell any securities. The information contained in this report has been obtained from sources believed to be reliable, but StrataFinance does not guarantee its accuracy, completeness, or timeliness.