2026, Amazon's year of self-validation

I have to admit that in the past two years, the stock-holding experience of Amazon $Amazon(AMZN.US) has been worse compared to a few others.

This afternoon, while checking Amazon's progress in the robotics field, I coincidentally saw that Friday's stock price of 227.3 was exactly the same as on December 6, 2024. 😅, but over the past year, Amazon:

EPS growth: +36.4%
Revenue: +13.4%
Profit margin: +38.2%

2026 is crucial because computing power is defined as the water, electricity, and coal of AI, which means its price cannot be too high. The upper-layer enterprises of AI applications need to make money, meaning the business model must be viable and validated, for AI applications to flourish and become widespread. Below are the current computing power prices of the three major clouds ☁️:

▸ Google GCP (A3-high): $3.0/card·hour (lowest in 2025Q4, no premium)
▸ Amazon AWS (P5 instance): $3.9/card·hour
▸ Microsoft Azure (NCH100 v5): $6.98/card·hour

From a price perspective, Amazon has a significant cost advantage, and it can continue to decrease, thanks to its self-built data centers and self-developed chip computing power. Microsoft currently has no large-scale self-developed chip deployment, and its computing power costs rely entirely on the price of NVIDIA $NVIDIA(NVDA.US) cards. Given the price advantage, it's foreseeable that AWS's growth rate will continue to rise.

As for electricity, the three companies handle data center power differently. Amazon, like building its logistics system, is constructing its own power plants, with power generation and energy storage centers near its data centers. 😅, this is a capital-intensive move, and the risk is borne by itself. Google and Microsoft have chosen to cooperate with existing power plants to secure electricity. The market may not approve of Amazon's approach, as the investment is too large and significantly impacts the income statement.

However, as secondary market investors, we need our own perspective. Mr. Duan said: "Buying stocks is buying companies. Do the right things, and do things right." I think this makes a lot of sense. I believe Amazon is doing the right thing, regardless of how the market views its capital investments. Microsoft's CEO mentioned in a podcast interview that they have a large number of NVIDIA cards sitting idle, not because they can't sell them, but because of a lack of electricity. This highlights two issues: first, AI demand is huge, and power supply can't keep up. Returning to Amazon's self-built power plants, it's actually solving its own power shortage problem, ultimately aiming for pricing power in computing. So, in the long run, it's about doing the right things right, and infrastructure has always been Amazon's strength.

Regarding progress in robotics, Amazon's robots 🤖 are designed to improve efficiency and reduce costs, used only in specific scenarios at scale, making it hard to evaluate the gap with $Tesla(TSLA.US)'s Optimus general-purpose robot. So, let's wait and see. 😂. The autonomous driving Zoox is also for specific scenarios. Amazon seems uninterested in general-purpose designs, focusing instead on perfecting certain scenarios or directions, which is also a good business model. 😀

I remain optimistic about Amazon in 2026. 👋👋, good night, everyone.