A major change is happening in the AI chip world. Reports say OpenAI is discussing a deal worth at least $10 billion with Amazon. The key point is not only the money. It is also about which chips OpenAI may use to build and train AI systems.
To understand why this matters, it helps to define a few terms.
A chip is a small piece of hardware that does computing work. In AI, chips are used to do heavy math fast. Training means teaching an AI model by feeding it huge amounts of data until it learns patterns. Cloud means using computers that live in big data centers, accessed over the internet.
Nvidia is the best-known company in AI chips. It sells GPUs, which are powerful chips used for AI training. Nvidia also has a major advantage called CUDA (pronounced “koo-duh”). CUDA is Nvidia’s software platform. It is the tools and code system many AI developers already use. This creates a moat, meaning a strong advantage that makes it hard for competitors to take customers.
Amazon is pushing its own chips through AWS, its cloud platform. It uses a three-chip strategy:
- Graviton for general computing
- Inferentia for running AI models (also called inference, meaning the AI is already trained and is now answering questions)
- Trainium for training AI models
Amazon says Trainium can offer better value for the cost in some cases. If a top AI company uses Amazon chips, it can act like “proof” that Amazon’s chips are real competition in certain jobs. It can also make AWS more “sticky,” meaning customers stay because the system works well together.
At the same time, there are risks. Many large tech companies are spending huge amounts on AI. Some of this spending can look circular, where money moves around the same small group of companies. There are also concerns about debt and whether AI spending will create enough profits soon.
Felix Prehn, who runs Goat Academy, frames this moment as both exciting and cautionary. Nvidia still has major strengths, especially its software ecosystem and broad use across many settings. But Amazon and others are building cheaper, more specialized options inside their own clouds. That can slowly shift market power over time.
A balanced view is important. AI demand can keep growing, but markets can also overheat. Diversification means not relying on only one theme or one stock.
Learn more about Felix Prehn’s background here: Felix Prehn Goat Academy
