- Apple is extending its private AI infrastructure beyond its own data centers for the first time
- The move marks a notable shift for Apple’s privacy-first cloud strategy, pairing Apple-controlled software with Google Cloud
- Analysts say the decision likely reflects a mix of technical integration with Gemini-based features and the economics of scaling AI infrastructure
Apple has built a brand around privacy, starting first with its devices before expanding to its own private cloud infrastructure in 2024. Now, though, it’s bringing a third party into the mix: Google.
Tekonyx Founder and Chief Research Officer Sid Nag told Fierce that the private cloud collaboration between the two isn't the big news. "The headline is that Apple is attempting to export its privacy architecture beyond its own infrastructure," he said. In a nutshell, Nag believes Apple's Private Cloud Compute (PCC) offering "is evolving from a consumer privacy feature into a strategic cloud architecture."
Here's what's happening: Apple is expanding its PCC infrastructure beyond its own data centers and will begin running Apple Intelligence workloads on Google Cloud.
“The wider industry has been working to provide a set of confidential inference primitives that could theoretically be combined to reach the security level of PCC. However, until today, those primitives have never been integrated into a comprehensive, end-to-end confidential inference pipeline capable of operating at global scale,” Apple wrote in a blog. “That’s what we’ve done with PCC on Google Cloud.”
While PCC previously ran exclusively on Apple silicon, the implementation on Google Cloud will use a combination of Nvidia Confidential Computing with Nvidia GPUs, Intel CPUs with TDX and Google's Titan chip, Apple said. The company added that it retains “complete control” over its private cloud software and that “Apple devices will only trust PCC software that is cryptographically approved by Apple.”
This bit is what caught Nag's attention.
"Apple is effectively arguing that trust in AI systems should come from cryptographic and architectural guarantees rather than trust in the cloud provider itself," he told Fierce. "This represents the emergence of a new AI infrastructure design pattern. Historically, enterprises were forced to choose between privacy and access to large-scale AI compute. Apple is demonstrating a third option: portable trust."
News of the Google Cloud move came as Apple unveiled its third generation of foundation models, which includes two on-device models and three server-based models. AFM 3 Cloud Pro, in the latter category, will run on Google Cloud.
Nag added that if Apple's move is successful, it's infrastructure model "could become a blueprint for how enterprises deploy AI workloads across sovereign clouds, hyperscalers, and third-party AI infrastructure without exposing sensitive data."
Other reasons for the move
Looking beyond Nag's take, Moor Insights and Strategy VP and Principal Analyst Jason Andersen told Fierce that the move could be a function of Apple’s collaboration with Google on its new foundation models.
“I think it has as much to do with integration as anything else. The Apple Intelligence features are a derivative of Google Gemini, and I also would have to assume that a lot of Apple device-specific integrations are there,” he said. “There is a good chance that to make this new stack perform well, Apple needed to deploy it closer to the Gemini models.”
“Secondarily, there is economics and expertise. It may just make more economic sense for Apple to outsource the infrastructure to someone who is better equipped from a supply chain and data center expertise perspective,” Andersen concluded.