Apple's $299 HomeView ships in November — and Siri AI still needs your iPhone
Apple turned its long-rumored smart home hub into real hardware today, and the whole line now runs through Siri AI — while a Chinese memory maker rattled chip markets and NVIDIA rounded up local-AI apps ahead of Friday's RTX Spark launch.
Apple launched the HomeView, its first smart display, and put Siri AI across its home line. The $299, 7-inch device ships November 13 with an A18 chip, a 12-megapixel camera that can recognize up to six household members, and Apple Intelligence features — personal context, text descriptions for security-camera feeds — that lean on a connected iPhone. That tether has edges: every household member needs an iPhone on the latest OS to get those features, and the device isn't sold in the EU at all because Siri AI isn't available there yet. The rest of the line follows the same pattern. The $149 Apple TV 4K gains four far-field microphones so you can talk to it without the remote, though answers come back as on-screen text; the $149 HomePod Mini 2 ships October 30 but won't support Siri AI until a "later" software update. Apple's answer to Echo Show and Nest Hub is finally real — but Siri AI still outsources much of its intelligence to the phone in your pocket, which is both the launch's headline feature and its biggest caveat. We saw this hub in employee testing before launch — Apple's Siri AI home hub is in employee testing, eyeing October.
CXMT's new memory architecture wiped billions off chip stocks. The Hefei-based Chinese maker said over the weekend it had made progress on a denser "4F square" cell architecture and plans to ship a server-memory product built on it by the end of 2026, and markets reacted as if the memory order is about to change: SK Hynix closed 10% lower in Seoul, Samsung fell 7.4%, and Micron was down 3.7% premarket, with SanDisk, Seagate and Western Digital all in the red. CXMT only listed in July — shares jumped 466% on day one and still carry a $520 billion market cap — and its share of global DRAM has gone from roughly 4% to around 10% in a year, per an HSBC note. The panic is a pricing signal: incumbents' margin cushion assumes CXMT stays behind, and memory is currently the chokepoint for AI buildouts.
NVIDIA says the local-AI app ecosystem is ready for RTX Spark. Laptops built on the 128GB unified-memory, one-petaflop superchip ship Friday, October 16, and NVIDIA's launch-eve checklist includes the Nemotron 3.5 Bolt 74B model (74 billion parameters, 9 billion active, 2-bit quantized into 24GB, fully open source including training data) arriving October 15, one-click local-model setup in GitHub Copilot, Hermes, OpenClaw, Perplexity and others, and OS-level sandboxing for agents. NVIDIA also claims local LLM downloads already run 5x higher than all of last year. Treat the software numbers as vendor claims, but the launch date and independently reported pricing — Microsoft's Surface Laptop Ultra starts at $2,599 — are solid.
What to watch: whether HomeView preorders sell out or sit, and if Apple speeds up the iPhone requirement for Siri AI.
Is an AI smart display that needs an iPhone for every family member really a smart-home hub? Tell us in the comments.
