Apple's New M5 Chips Debut
Apple just rolled out its new MacBook lineup, featuring M5 Pro and M5 Max chips with what one report calls "four times AI inferencing performance." The new entry-level MacBook Neo, priced at £599, uses an A18 Pro chip but comes with a controversial, non-upgradable 8GB of RAM. Arm is already breaking down the AI expansions across the new hardware stack.
The M5 family of chips is fabricated using TSMC's third-generation 3-nanometer process, an evolution of the technology used in previous Apple silicon. This manufacturing node allows for increased transistor density and power efficiency. Looking ahead, TSMC has already begun volume production of its 2nm (N2) process, which will introduce gate-all-around (GAA) nanosheet transistors for the first time, promising another leap in performance and efficiency. Architecturally, the base M5 chip features up to a 10-core CPU with four performance cores and six efficiency cores, delivering roughly 15% faster multi-threaded performance than the M4. The more significant gains are in graphics, with up to 30% faster overall GPU performance and a 45% improvement in ray tracing. This is achieved through a new GPU architecture where each of the 10 cores integrates a dedicated Neural Accelerator, boosting AI-centric tasks. For the higher-end M5 Pro and M5 Max, Apple introduced a "Fusion Architecture" that bonds two separate dies into a single SoC using advanced packaging. This multi-die approach allows for a significant scaling of core counts, with the M5 Max featuring up to an 18-core CPU and a 40-core GPU. Early benchmarks show the 18-core M5 Max outperforming the previous 32-core M3 Ultra in multi-core CPU tasks. The entry-level MacBook Neo's A18 Pro chip is built on TSMC's enhanced N3E 3nm process and features a 6-core CPU (2 performance, 4 efficiency) and a 6-core GPU. Its 16-core Neural Engine is capable of 35 trillion operations per second. While powerful for its class, the chip's performance is constrained by the controversial 8GB of non-upgradable RAM, a configuration many argue is insufficient for a 2026 machine, especially given Apple's high prices for memory upgrades. For developers, the transition to new hardware is often accompanied by shifts in software frameworks. Apple's aggressive API deprecation strategy continues, with recent examples including the complete removal of Apple Type Services (ATS/ATSUI) in macOS 14, forcing developers to migrate to modern frameworks like Core Text. Similarly, to encourage multi-window support on iPadOS, Apple has deprecated the legacy UIApplicationDelegate, requiring the use of UISceneDelegate for apps submitted to the App Store. In the broader ecosystem, the Matter smart home standard, heavily backed by Apple, continues to evolve, aiming to unify device interoperability across platforms like Apple Home, Google Home, and Amazon Alexa. The standard, which uses Wi-Fi and Thread for local communication, is updated biannually, with recent versions adding support for new device types and improving security protocols to simplify the experience for both developers and consumers.
