China names AI phones, robots, smart homes a national consumption priority

Beijing just put AI hardware on the consumer shopping list. The Commerce Ministry and six other departments published an "implementation opinion" on Monday that elevates smart phones, PCs, wearables, robots, and smart home gear to a national consumption priority, with new "AI + consumption" pilot zones, retail experience centers, and a 60-trillion-yuan retail-by-2030 target backing the push. The same day, a domestic supernode vendor won a $100 million-plus China Mobile AI compute contract — proof that the supply side is racing to keep up with the demand Beijing is now engineering.

The Commerce Ministry, NDRC, MIIT, the Ministry of Finance, the Agriculture Ministry, the Culture and Tourism Ministry, and the State Administration for Market Regulation jointly issued the "Implementation Opinion on Promoting the Expansion and Upgrade of Commodity Consumption" — a 20-clause plan filed August 13 and released publicly today. The document sets a target of about 60 trillion yuan in total retail sales of consumer goods by 2030 and calls for cultivating "hundred-trillion-yuan markets" in green, smart, and health consumption. Item 14 — labeled "smart consumption" — is the AI hook: build national AI application pilot bases for consumer goods, "launch a batch of internationally leading, high-potential intelligent consumer products," and accelerate the rollout of AI phones and PCs, smart wearables, intelligent robots, and smart home devices. The plan also mandates "AI + consumption" cluster zones, collection stores, and dedicated application experience centers as consumer infrastructure — effectively treating AI retail the way China once treated appliance subsidies and EV charging.

This is the demand-side companion to the supply-side push we covered this morning — China's MIIT sets hard targets for an AI service-provider army — but the two policies are doing very different work. MIIT is building the integrator bench; Commerce is trying to create the buyer. The plan also extends AI to aging-friendly products (exoskeleton robots, smart mattresses, health monitors entering elderly homes), to children's smart care devices, and to standards for smart-home interoperability. The intent is unmistakable: lift domestic consumption out of its post-property slump by making AI gadgets the next big retail category, and back the move with finance and tax incentives for individuals and merchants.

What to watch: whether the "AI + consumption" pilot zones actually become the appliance-style subsidies of the late 2020s, and whether the 60-trillion-yuan target survives a property drag that has yet to break.

Domestic vendor Yangtze Computing, a unit of Fiberhome, won 71.43 percent of China Mobile's latest AI supernode equipment tender — a contract worth more than 700 million yuan ($100 million) including tax, the company confirmed on Monday. The bid covers China Mobile's Hohhot intelligent computing center expansion for 2026 and the supplementary procurement under its 2026–2027 AI supernode framework, both central to the operator's domestic compute build-out. Yangtze Computing is now Fiberhome's flagship ARM-server line; it has stacked more than 4 billion yuan in operator-side AI contracts this year, including a 1.6-billion-yuan Telecom server tranche and earlier Mobile PC-server wins. China Mobile is one of three carriers building out the "Eastern Data, Western Compute" national network, and these supernode awards determine which silicon ends up anchoring China's hyperscale AI training and inference fleets.

The order matters less for the dollar figure than for what it signals: an ARM-based, domestic-supplier stack is winning large chunks of the country's largest carrier AI tender — exactly the kind of substitution Beijing's industrial policy has been funding since 2023. If Fiberhome's Hohhot expansion performs on schedule, it sets a template for the rest of the carrier build-out.

With AI now a designated consumer category, do you think retail subsidies will buy the adoption curve the labs need? Tell us in the comments.

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re

Read more

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Amazon's new Bee devices listen all day — for $11.99 a month

Perplexity open-sources its agent sandbox after 118 failed escapes