Apple forced to use Nvidia chips for AI despite reluctance

Apple confirmed at WWDC on June 8 that its cloud AI runs on Nvidia Blackwell B200 GPUs hosted in Google Cloud, marking a shift from its reliance on Apple silicon. The deal uses Nvidia's confidential computing framework to protect user data, while Apple develops its own in-house chip, Project ACDC, expected by late 2026.

Apple forced to use Nvidia chips for AI despite reluctance Cupertino quietly confirms it's running its cloud AI on Nvidia's Blackwell GPUs via Google, in a deal nobody expected Apple spent years telling itself it could do everything with Apple silicon. Then Siri happened. At WWDC on June 8, Apple acknowledged for the first time that its Apple Foundational Model Cloud Pro, the backbone of its most advanced cloud AI capabilities, runs on Nvidia GPUs hosted inside Google’s cloud infrastructure. The arrangement puts three of the most powerful companies in tech inside the same deal. Apple brings the software, the privacy architecture, and a billion-plus device install base. Google brings the cloud infrastructure. Nvidia brings the hardware muscle, specifically its Blackwell B200 chips, which will be powering a meaningfully upgraded version of Siri starting September 2026. Why this is a bigger deal than it sounds The specific technology Apple adopted is Nvidia’s confidential computing framework, which allows sensitive workloads to run in a third-party cloud environment without exposing user data. Apple needed a way to run powerful AI in the cloud without handing Google or Nvidia visibility into what users are actually doing. Nvidia’s confidential computing stack was apparently the piece that made the deal acceptable to Apple’s privacy team. Critically, Apple is not buying Nvidia hardware directly. The arrangement runs through Google Cloud, meaning Nvidia’s chips sit inside Google’s infrastructure and Apple accesses them as a service. What this means for Nvidia and the broader AI chip market For Nvidia, landing Apple as a customer, even at arm’s length through Google Cloud, is a meaningful signal to the rest of the market. Nvidia’s stock reaction to the announcement was muted, which makes sense given that the company is already priced as the dominant AI infrastructure provider. The competitive read here cuts both ways for Google. On one hand, Google Cloud gets to claim Apple as a major enterprise customer. On the other hand, Google is essentially providing the pipes that make a better Siri possible, which is a rival to Google Assistant. Apple’s longer game Apple is not abandoning its own silicon ambitions in parallel. The company is actively developing what it’s calling Project ACDC, an in-house data center inference chip aimed at reducing dependence on third-party GPU providers. The expectation is that those chips arrive in the second half of 2026, which would give Apple a proprietary path forward for at least some of its cloud AI workloads. The picture that emerges is one of a company bridging a capability gap. Nvidia fills the immediate need. Apple silicon fills the long-term ambition. The risk is that Project ACDC slips its timeline, which would extend Apple’s dependence on Nvidia and Google longer than anyone at Cupertino would prefer. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .