We’re still in the middle of the rollout of Apple’s M4 chips, but today we got our first big peek at its successor, the M5. The report comes from reliable analyst Ming-Chi Kuo, who posted on X today and spilled some juicy technical details about what’s coming.

The M5 family of chips will be manufactured on TSMC’s N3P node, the next step up from the N3E on the M4. Kuo says the new node “entered the prototype phase a few months ago,” but it’s the first time we’re getting three generations of chips in a row that use the 3nm node.

The M5 Pro, Max, and Ultra, specifically, will use “server-grade” 2.5D packaging, with the explicit purpose of improving production yields and thermal performance. This will allow the CPU and GPU to use separate designs, which is a major change from previous generations that use a conventional system on chip. Kuo also mentions that these high-end M5 chips will be “better suited” for AI inferencing.

There’s a lot we don’t know about in terms of exactly how these changes will affect the end product, but clearly, Apple is giving its higher-end chips some new capabilities and features

The timeline provided by Kuo doesn’t deviate from previous generations, with the M5 kicking things off in the first half of 2025, followed by the M5 Pro/Max in the second half. Lastly, the M5 Ultra is scheduled for 2026. This would return the most powerful Mac desktops to a yearly update cycle after having skipped the M3 Ultra.

We’re still waiting on the M4 Ultra, of course, which is scheduled to pop up in a refreshed Mac Studio and Mac Pro sometime in 2025.