NVIDIA — Business & Moat

NVIDIA sells data-center AI accelerators (GPUs) and the software ecosystem that runs on them. The thesis is simple: if AI training/inference demand is a structural shift in computing rather than a passing boom, the company holding the key component of that shift benefits. [Investing Notes.md]

For balance: the risks (bear case) below get equal weight with the bull case. An analysis with only one side isn't analysis — it's cheerleading.

The real moat is less the chip's raw performance and more the CUDA software ecosystem. Countless AI frameworks and libraries are built on CUDA, so even when a competitor ships comparable hardware, the switching cost for developers is high. Hardware can be matched; ecosystems erode slowly. [Investing Notes.md]

• AI infrastructure spend is backed by the enormous capex of the major cloud providers.
• As demand expands from training into inference, it shifts from one-off to recurring.
• Ecosystem lock-in supports pricing power (margins).

[Investing Notes.md]

1. 1.Valuation. Expectations are already high; even a small growth slowdown can hit the stock hard.
2. 2.Customer concentration. A few large clouds are a big share of revenue — if their capex turns, it's a direct hit.
3. 3.In-house chips. Customers are building their own ASICs — possible long-term demand erosion.
4. 4.Cyclicality. Semis are sensitive to inventory/macro cycles. Inventory-correction risk after a boom.

[Investing Notes.md]

Data-center revenue growth, cloud capex guidance, inventory levels, competitor accelerator shipments. Of these, cloud capex guidance is the most leading signal — demand direction shows up here first. [Investing Notes.md]

This growth lens is the opposite of Coca-Cola's dividend/defensive lens. The trade-off between them is in Growth vs. Dividend.