Apple unveiled the next generation of MacBooks in a second spring product wave, introducing an M5-equipped MacBook Air and new MacBook Pros powered by M5 Pro and M5 Max. The headline technical gains are dramatic: a reworked M5 architecture that decouples CPU and GPU dielets using TSMC’s SoIC‑MH packaging, yielding up to fourfold peak GPU AI throughput versus the prior generation and meaningful multithreaded CPU improvements.
Alongside the silicon leap, Apple has quietly reshaped its product mix. Base storage capacities have been raised — the M5 MacBook Air now starts at 512GB and the 14‑inch M5 Pro MacBook Pro at 1TB — while the company simultaneously discontinued lower‑capacity SKUs. Retail prices have climbed: the MacBook Air’s entry price is up $100 to $1,099 (RMB 8,499 in China), and the 14‑inch MacBook Pro’s floor is $2,199, a $200 bump, with equivalent yuan increases for local pricing.
Apple is explicit about holding memory upgrade pricing steady even as it lifts base storage. RAM tiers and their upgrade costs remain the same as the M4 generation, and higher GPU counts are restricted to the M5 Max, consolidating the company’s high‑end graphics and AI performance behind its premium SKU. The company also improved SSD performance, claiming sequential read/write speeds roughly twice those of the previous generation and offering up to 4TB on Air models for the first time.
This product-and-price combination reads as a tactical response to a broader industry problem: sustained increases in NAND and DRAM prices. By eliminating the cheapest storage options and pushing the baseline upward, Apple makes a higher entry price less conspicuous while raising average selling prices and insulating margins from component cost inflation.
Technically, the M5’s adoption of TSMC’s SoIC‑MH marks a pivotal shift. Where prior "Pro" and "Max" chips were effectively multi‑chip modules built from stitched standard dies, M5 blends chiplets into a more integrated package that lets Apple rebalance CPU and GPU resources more flexibly. The outcome is materially better on‑device AI performance — a capability Apple is increasingly touting as a differentiator as generative and inference workloads move to endpoints.
For consumers, the trade‑offs are clear. Buyers gain far stronger local AI and GPU performance and faster storage, but they pay more for those gains even if they do not need the extra capacity. For would‑be upgraders, Apple has preserved predictable memory pricing but funneled high‑end GPU options into the priciest configurations, nudging heavy users toward the top of the price ladder.
Industry‑wide, Apple’s moves have ripple effects. NAND and DRAM suppliers will welcome a product lineup that tolerates higher component pricing, while PC rivals face pressure to offer comparable on‑device AI performance without a similar ability to shift SKU mixes and pricing strategy. At the same time, TSMC’s SoIC‑MH wins yet another marquee design partner, underscoring the foundry’s centrality to the evolving chip ecosystem.
Investors will watch whether Apple’s repositioning staves off margin erosion or simply accelerates larger ticket purchases that compress upgrade cycles. For enterprises and developers, the M5’s improved on‑device inference capacity reduces dependence on cloud GPUs for certain workflows, altering cost calculus for edge AI applications and potentially encouraging more macOS‑centric AI tooling.
In short, Apple’s M5 MacBooks combine substantive technical advances with a deliberate commercial tweak: raise the base, keep upgrade math unchanged, and corral higher‑margin demand into premium SKUs. It is a calculated move to protect profitability amid commodity price pressure while continuing to sell the narrative of superior on‑device AI capability.