M1 completes Steve Jobs' vision of Apple making “the whole widget” for all the devices in the company's product line
Even as Apple's final event of 2020 gradually becomes a speck in the rearview mirror, I can't help continually thinking about the new M1 chip that debuted there.
Context & Ripple Effects
Apple's M1 is the endpoint of a long integration arc: the company had already moved through PowerPC and x86 before building the A-series iPhone chips that proved it could design its own processors at scale, a history traced in Apple's CPU lineage from PowerPC to x86 to A-series. With M1 running across the entire Mac line, Apple no longer depends on a merchant chip vendor for any product category — the 'whole widget' Jobs described is now literal.
What followed shows the strategy compounding rather than plateauing: Apple stacked two M1 Max dies into an M1 Ultra combining two Max dies into one package, pushed transistor counts higher with M2, and by the M3 generation was spending on the order of $1B in tape-out costs for three chips — evidence that the whole-widget approach scales up only for companies willing to fund it.
First-order effects
- Intel loses the Mac socket entirely, and the CPU-history record shows the same playbook already beating Qualcomm and other merchant suppliers — Apple now sets its own silicon cadence instead of waiting on theirs.
- Every device in Apple's lineup shares one architectural base, so features built for iPhone chips carry to Mac without negotiating with an outside vendor's roadmap.
Second-order effects
- Rivals that sell chips or build devices around them are pushed toward their own custom-silicon programs, since competing against vertically integrated machines with off-the-shelf parts means competing on someone else's schedule.
- Unified memory and die-stacking let Apple position desktop-class machines like Mac Studio against workstation incumbents, pulling performance-per-dollar comparisons away from raw clock speeds toward system integration.
Third-order effects
- If tape-out costs near $1B per chip generation hold, custom silicon becomes a scale game only a handful of device makers can play, splitting the industry between integrated players and those renting merchant chips.
- The durable advantage shifts from any single chip to control of the full stack — hardware, OS, and silicon designed together — making it structurally harder for component-vendor-based competitors to match feature timing.
The trend: Consumer-device competition is consolidating around vertically integrated, workload-specific silicon, with rising tape-out costs narrowing the field of companies able to make the whole widget themselves.