--- title: "WWDC 2020: Apple's silicon bet." date: 2020-06-24 url: https://remiam.co.uk/notes/apple-silicon-the-bet tags: [Apple, Apple Silicon, Industry, Hardware] read_time_minutes: 9 description: "WWDC 2020 reaction — Apple announces the Mac transition to Apple Silicon. What it means for studios, for Intel, and for the next decade of Mac hardware." --- # WWDC 2020: Apple's silicon bet. *Published 2020-06-24 · 9 min read · by Liam (Remiam)* Apple just announced it is moving the Mac off Intel and onto its own chips. The keynote was confident. The two-year transition window is ambitious. Here is what it means for studios, for the laptop market, and for the next decade of computing — plus the transition history that explains why we think it works. Apple announced the architecture transition on Monday. The Mac is moving off Intel and onto Apple's own ARM-based silicon. The first machines ship by the end of the year; the transition is supposed to be complete within two. Tim Cook called it 'a historic day for the Mac'. He's right — but the historic part isn't the announcement itself. It's that Apple has done this exact thing twice before, and both times it worked. ## Apple's transition history | Year | From | To | Duration | Outcome | | --- | --- | --- | --- | --- | | 1994 | Motorola 68k | IBM PowerPC | ~5 years | Successful, opened up the Mac platform | | 2006 | IBM PowerPC | Intel x86 | ~1.5 years | Wildly successful, accelerated Mac growth | | 2020 | Intel x86 | Apple Silicon (ARM) | 2 years planned | TBD — early signs are extraordinary | *Apple's three Mac architecture transitions.* ## Why this is actually a big deal - Apple has been designing world-class chips for the iPhone and iPad for a decade. The A-series is already faster than most laptop CPUs by per-core measures. - The Mac has been thermally constrained by Intel's roadmap for years. The MacBook Pro fan noise is a chip-design problem, not an Apple-design problem. - Unified memory architecture (CPU and GPU sharing one pool) changes the kind of workloads laptops can handle. - Vertical integration — Apple owns the chip, the OS, the dev tools, the apps. They can optimise across boundaries that nobody else can. - Performance per watt — the killer metric. Intel's mobile chips are years behind Apple's A-series by this measure. - Apple controls the timing. No more 'waiting for Intel to ship a faster part'. ## What about the existing Mac stack? - Rosetta 2 is the translation layer. Apple says it's good. Apple usually means it when they say that — but real-world performance is the test. - Xcode produces 'Universal Binaries' so apps can target both architectures from one project. A familiar move from the PowerPC-to-Intel transition. - Most developer tooling — Node, Docker, Homebrew, language runtimes — needs to be recompiled. Expect a rough six months. - Boot Camp goes away. Running Windows natively on a Mac stops being a thing. Parallels via virtualisation will be the new answer. - iOS apps will run natively on macOS — a side-effect of the shared architecture. This is a smaller deal than it sounds in practice. > If the chip is half as good as the rumours suggest, the Mac is about to have a five-year head start over the rest of the laptop industry. The industry has no answer to vertical chip design at this scale yet. ## What we'll do at the studio - Hold the existing Intel Macs for another year. We've just refreshed. - Buy one DTK (developer transition kit) to start testing our toolchain. - Buy a real M1 Mac whenever it ships, primarily to flush out edge cases in our build pipeline. - Avoid committing to a single-architecture stack until we see how Rosetta 2 performs at production scale. - Test every client production build on both architectures during the transition window. - Keep at least one Intel machine around for the long tail — there'll be legacy tooling that takes years to migrate. ## Risks to the transition - Pro app vendors (Adobe, Autodesk, Avid) move slowly. If Adobe takes two years to ship native Apple Silicon, half the creative industry stays on Intel. - Docker performance — running Linux containers under emulation is the obvious pain point. - Gaming. The Mac gaming story was thin under Intel; ARM makes it thinner before it gets better. - x86-only enterprise software with no ARM build will take years to migrate (or never will). - Initial M1 chips may be excellent for laptops but underwhelming for desktops — the Mac Pro story is the last unsolved problem. It's the most ambitious thing Apple's announced since the iPhone. If the chip is half as good as the rumours suggest, the Mac is about to have a five-year head start over the rest of the laptop industry. We'll see by December. Until then, the smart play for studios is curious-but-cautious: watch the launch, test the chip, hold the migration. ## References 1. [Apple's announcement of the Mac transition to Apple Silicon (WWDC 2020 keynote)](https://www.apple.com/newsroom/2020/06/apple-announces-mac-transition-to-apple-silicon/) 2. [Anandtech — Apple A-series chip analysis (years of reporting)](https://www.anandtech.com/tag/apple)