More companies are building their own AI chips instead of relying on Nvidia

In short: Big tech companies are increasingly designing their own AI chips to reduce how much they depend on Nvidia.

What's going on

Nvidia has been the main supplier of chips used to run and train many AI systems for years. These chips are the specialized parts that do the heavy math needed for AI, like a powerful engine built for one kind of job.

Now, more companies are trying to use chips they design themselves. TechCrunch reports that OpenAI recently shared plans for “Jalapeño,” a custom inference chip built with Broadcom. “Inference” is the step where an AI model answers questions or creates output after it has already been trained (like using a finished recipe, not testing ingredients).

OpenAI is not alone. Google and Apple have been working on their own chips for years, and SpaceX is also part of the group moving in this direction. The point is not always to stop buying from Nvidia completely. Instead, it is often a hedge, meaning a backup plan that lowers the risk of relying on a single supplier.

One reason companies do this is control. A custom chip can be tuned for the company’s specific workloads, which can improve speed or cost. TechCrunch compares this to Apple switching away from Intel in its Macs, which gave Apple more say over how its computers perform.

What to watch

If more large buyers shift even part of their spending to custom chips, Nvidia could face slower growth in some areas. At the same time, chip partners like Broadcom could benefit, and AI services may get cheaper or faster if these chips reduce operating costs.

Source: TechCrunch AI