Open AI now reportedly getting closer to developing its own custom AI chip.

Reuters reporting that it plans to finalize the design in the next few months.

Deer Tobosa has more on that news for today's tech check.

Hey, D.

Hey, good morning, Leslie.

So Open AI's chip ambitions, they're no secret in Silicon Valley.

That Reuters report had some interesting new details on where they are in the process and the cost,

starting at half a billion dollars for a single version that may not even function.

It helps explain why OpenAI is raising so much money.

It's not just about pushing the frontier.

It's about more vertical integration, making their process more efficient.

Now, the rising stakes of doing so, that's where it gets really interesting.

On the surface, this is a strategic move designed to give them more choice for training and

running their AI models. It's supposed to reduce their reliance on NVIDIA. This is something

that Amazon and Google have been long working on and are now deploying in their own data

centers. Google with its TPUs and Amazon with its Traneum

custom chips. But reading between the lines, this may have just as much to do with Open AIs

evolving and increasingly complicated relationship with Microsoft. When Project Stargate was

announced earlier this year, Open AI and Microsoft tweaked their agreement. Microsoft was no longer the

exclusive cloud provider to Open AI. It moved to a model where Microsoft has a right of first

refusal. So even though it now allows OpenAI to negotiate with other cloud vendors like

AWS or Google Cloud, those companies have no incentive to give Open AI a good price.

...