Could Macs Handle More Enterprise AI Workloads?

Industry observers note that as organizations expand AI deployments, recurring inference costs have become a growing concern alongside data governance and security. Cloud GPUs are expensive to rent, every AI query consumes compute, and those costs grow as deployments move from pilot projects into production.

On its fiscal third quarter earnings call, held on July 30, Apple argued that more enterprise AI workloads can run on Macs instead of relying entirely on cloud infrastructure, reducing recurring inference costs for some use cases.

Apple's Enterprise AI Pitch

Apple is making the case for a different AI deployment model. The usual setup runs through a GPU cluster, into the cloud, out through an API, and back as inference, with a cost attached to every step. Apple positions its approach as an alternative for certain workloads: a Mac, local inference, lower cloud spend, and better privacy. If adopted more broadly, the approach could influence how enterprises decide where different AI workloads should run.

The enterprise AI discussion came during a strong quarter in which Apple reported revenue of $109.4 billion, up 16% year over year, with Mac revenue rising 29%.

Apple CEO Tim Cook laid out the reasoning behind Apple's hardware strategy during the earnings call. He said the company began preparing for AI-driven experiences when it introduced the Neural Engine in 2017, and has since built out chip design, system architecture, and memory capacity around AI performance.

Cook described Mac as continuing "to be the ultimate AI powerhouse," pointing to its strength in on-device inference across a broad range of AI workloads. The remarks position Apple silicon as capable of handling some AI workloads that many organizations currently process in the cloud, though Cook did not specify which workloads are best suited to local inference.

Cook also said customers are "deploying clusters of Mac Studio systems to run frontier class models locally," extending Apple's strategy beyond individual laptops and desktops to small-scale enterprise AI infrastructure.

Through the customer examples and Cook's remarks, Apple positioned Macs not simply as employee devices, but as infrastructure capable of running AI workloads closer to where company data is created.

Practicing the Strategy

Kevan Parekh, Apple's Chief Financial Officer (CFO) cited several organizations that are already using Apple hardware as part of their AI deployments, saying that "more companies are choosing Mac for on-device AI advantages, including lower costs, better performance, and enhanced privacy and security."

French retail bank Crédit Agricole was highlighted as using on-device AI on MacBook Pro to support regulatory workflows. According to Parekh, the deployment is "reducing manual processing time by over 80%."

Disney was presented as another example. Parekh said the studio's creative teams "are increasingly turning to Mac for on-device AI workflows that reduce overall cloud token costs and keep their IP secure."

Morgan Stanley illustrated a different part of Apple's enterprise strategy. Parekh said the bank "has deployed over 20,000 iPhone 17 devices globally" as it shifts employees from personal devices to corporate-owned hardware for liability and security reasons.

While the three deployments address different business problems, together they reinforce Apple's broader argument that enterprise AI can increasingly run on managed endpoint devices instead of relying exclusively on cloud infrastructure.

Uncertainty Over Cloud Costs Remains

Even with these examples, analysts on the call wanted to know what Apple's own AI costs would look like. Bank of America analyst Wamsi Mohan asked directly whether "the capital intensity of Apple will change in the future because of Siri AI."

Cook said Apple uses "a hybrid model," splitting requests between third-party cloud services and its own data centers. His comments indicate that Apple sees local inference as complementing cloud infrastructure rather than replacing it entirely.

On being asked how usage data from the recent iOS 27 public beta of Apple Intelligence is shaping Apple's view of compute costs, Cook responded that "it's obviously early going for us," adding that he did not want to say "that we have a complete plan for that."

A Broader Enterprise AI Trend

Apple's argument fits within a broader industry push toward local AI inference, as hardware vendors increasingly promote endpoint devices as a way to reduce recurring inference costs and keep sensitive enterprise data closer to where it is created. 

Apple's contribution to that discussion is its argument that Apple silicon has become powerful enough to run more enterprise AI workloads locally, allowing organizations to reserve cloud infrastructure for workloads that still require it.

Apple did not disclose enterprise AI revenue or broader adoption figures beyond the customer examples discussed during the call. Instead, it used those deployments to argue that more enterprise AI workloads can run on Macs and other Apple devices, while continuing to rely on cloud infrastructure for workloads that require greater scale.

>

>