Macs to Get AI-Focused M4 Chips Starting in Late 2024

Original Article ↗ Hacker News Discussion ↗

Apple Silicon vs. PC Laptops

  • Several commenters say current M-series Macs beat comparable PC laptops on battery life, thermals, and screen quality, especially for fanless use.
  • Others counter that high-end Nvidia laptops and desktops are far superior for gaming and heavy AI workloads.
  • Some note modern AMD (Zen 4) laptops can achieve very good battery life and efficiency, arguing software is often the limiting factor, not just hardware.

AI-Focused M4 and On-Device LLMs

  • Many see the M4’s stronger Neural Engine and high unified memory as positioning Macs for local LLMs and “AI at the edge,” enabling private, low-latency assistants.
  • Enthusiasts envision GPT‑4‑class local models (given enough RAM) that see the screen, use the camera, and act as full personal digital assistants.
  • Skeptics say GPT‑4-level local models don’t yet exist for consumers and question whether 512GB–1TB RAM Macs will be affordable or realistic.

Siri, Software Strategy, and Use Cases

  • Several note that despite prior neural hardware, Siri’s capabilities lag; some doubt Apple’s software roadmap.
  • Others point to recent Apple research (e.g., UI-aware models, reference resolution) and strong rumors of a Siri overhaul as evidence Apple has a plan.
  • Expected use cases: system-wide assistance (email, calendar, search), photo/video editing, Safari “browsing assistant,” and deep integration with on-device data.

Hardware Limits: RAM, Pricing, and Upgradability

  • Unified memory is praised for enabling large models on laptops but criticized for cost and lack of upgradeability.
  • Apple’s RAM and storage markups are widely viewed as excessive; historical and current prices are used to argue that multi-hundred-GB or 1TB configurations could be extremely expensive.

Local vs Cloud AI and Hype Level

  • Pro-local: privacy, control, predictable one-time hardware cost, no vendor lock-in or monthly fees; attractive for work where cloud tools are restricted.
  • Pro-cloud / Nvidia: far higher raw performance per dollar for large models, and easy access via GPU instances.
  • Some see AI as transformational and Apple as well-positioned; others invoke the hype cycle and fear intrusive, overbearing “AI everywhere” experiences.