Apple bets on on-device AI with new M6 and M5 Ultra chips, aiming to stay ahead as Big Tech touts AI PCs

Apple’s chip roadmap is becoming a focal point in the race to deliver AI experiences that do not depend entirely on cloud servers. A recent market-focused write-up on Yahoo Finance argued that Apple’s upcoming M6 and M5 Ultra platforms are designed to narrow the performance gap in on-device AI, while also increasing the capacity to run or support larger AI models locally.

The central theme in the coverage is hardware acceleration, meaning specialized compute blocks in the processor that speed up AI workloads such as natural-language processing and image or video understanding. By increasing on-device throughput, Apple can reduce latency, keep more user data on the device, and potentially lower the cost of AI features that would otherwise require more server time.

The same article tied the chip push to broader competitive pressure from major platform companies, specifically Alphabet and Microsoft. Those rivals have emphasized AI capabilities across their ecosystems, including cloud-hosted model access and AI features that can be used through consumer and business applications. The implication is that the “AI PC” and device-layer experience is becoming just as important as model availability.

Apple’s strategy matters because it can influence how quickly AI features reach consumers and how consistently they work across networks. Local AI performance can be especially important when connectivity is weak or when customers expect real-time responses, such as in voice assistants, text generation, and other interactive features that benefit from faster processing.

For context, Apple’s approach to AI has increasingly leaned on balancing model power with privacy and efficiency, with on-device processing used for tasks where it is feasible. Chips that improve local AI performance expand the range of tasks that can be handled without sending data to a remote system, which can also help maintain responsiveness and user trust.

That said, the coverage did not lay out specific performance benchmarks, production timelines, or concrete details about which AI models will be supported on M6 or M5 Ultra. It also did not provide disclosed information about Apple’s exact neural engine (the AI-specific processing hardware within Apple silicon) configurations or the exact memory or software limits that determine how “large” a model can be for on-device use.

Investors and device developers will likely watch for clearer disclosures around real-world AI capability, including application-level outcomes like responsiveness and how many AI features remain fully functional offline. Apple’s next product announcements and any accompanying technical documentation, if provided, will be the most important indicates for whether the new chips deliver the on-device gains the market commentary expects.