Take

To sum it up: Intel has been embarrassing itself for the past few years, Apple has been building up an impressive silicon know-how for the past decade, and ARM architecture chips have been inherently power-efficient since they were invented.

The M1 versions kick the Intel Core versions around the block a few times in most workloads, and the integrated GPU cores have been receiving tuning and tweaking as part of pipelines targeting Metal on iPad and iPhone for years. There's also no reason to doubt the statements about power efficiency and performance-per-watt increases. And having a completely fan-less MacBook Air in a way that is not just a dangerous fool's errand is a remarkable achievement.

There are still open questions, though:

• The M1 leans into UMA, a "unified memory architecture" where the memory is shared between the GPU and CPU. That's great most of the time, but how is performance affected when both the GPU and CPU draw lots of resources?

  Will future M chips allow off-SoC memory in the first place, or off-chip GPUs? (eGPUs are noted incompatible with M1. Update: I was mistaken; only the MacBook Air M1 is incompatible, probably because of the lack of thermal envelope. Update: Apple's page was mistaken, as now no M1-powered models are listed at all.) Being the fastest integrated GPU is impressive, but it's also like walking the fastest; there are plenty of chips out there that can run.

  The M1 tops out at 16 GB now and the previous Mac mini offered up to 64 GB; how far north will the included RAM go in the future?

• Mac mini loses out on two additional Thunderbolt/USB-C ports and 10 Gbit Ethernet. Will this type of functionality simply drop off the radar?

• Both M1 MacBooks use the ISP to clean up the image, but still use 720p webcams? What year is it? Has no one heard of "garbage in, garbage out", or are they betting that extra chip functionality will solve literally every problem?

The M1 is the first step of the transition; the opening salvo instead of the crescendo. There will be surprises and revelations ahead, and I look forward to seeing genuinely new designs within the next year, including next spring and during WWDC.

Don't forget: The Macs that were revved today were from the bottom part of the lineup. For all we know, the M2 or M1X containing both faster cores and the additional infrastructure to do fuller system architecture are still being finalized. It takes time to build out the chip capabilities from iPhone/iPad-level chips to something capable of handling all subsystems in all Macs. PCI Express (used by Thunderbolt/USB4) and virtualization support are two examples of features already present in the M1 that no previous Apple SoC has needed to support.