Apple’s Siri just handed Nvidia another win

Apple said, 'Let someone else handle the heavy lifting'

Apple loves a good homegrown solution. Custom chips, tight vertical integration, the whole fancy ecosystem thing. So when the company reportedly decided to route select Siri workloads through Nvidia Blackwell B200 GPUs on Google Cloud, that’s not just a quirky backend choice — it’s a neon sign flashing Nvidia still owns the AI plumbing.

Why this matters for Nvidia

The spicy part here isn’t that Apple is using cloud GPUs. It’s whose GPUs it’s using. Nvidia keeps showing up anywhere real AI inference has to happen at scale, even when the customer is a company with deep pockets and its own silicon ambitions. If Apple needs Nvidia hardware to make Siri feel fast enough in production, that tells you a lot about where the market is today.

And yes, Google, Amazon, Microsoft, and Meta are all still building their own custom AI chips. But this is the annoying-for-competitors part: custom silicon hasn’t kicked Nvidia out of the ring. It’s just added more seats to the table.

The bigger picture

A few things jump out:

• Apple reportedly tested its own environment and found bottlenecks for real-time use.
• The workload was shifted to Nvidia Blackwell GPUs inside Google Cloud.
• That means Nvidia’s hardware is sitting in the path of a consumer-facing AI feature, not just a dusty data-center benchmark.

For investors, that’s important because every new AI feature shipping in the wild keeps pulling more compute into the market. And right now, Nvidia is still the company cashing the checks.

Big picture: Apple didn’t just make a backend tweak. It accidentally sent a pretty loud message about who still runs the AI infrastructure show.