China's robot industry books a record 165.5B yuan first half

China's robotics sector just posted its strongest half on record as the 2026 World Robot Conference opened in Beijing — and the executives on stage are already rethinking what the business model looks like once the machines can actually do the work.

China's robot industry hit 165.5 billion yuan in H1 revenue, up 24.5% year over year. At the Aug 19 opening of the 2026 World Robot Conference in Beijing, MIIT vice minister Xin Guobin said the country's robotics industry is now scaling fast: full-year 2025 revenue passed 300 billion yuan, the sector has grown more than 20% a year for five straight years, and the first half of 2026 alone reached 165.5 billion yuan (about $23 billion) on 24.5% growth. The ministry said its next move is to push "intelligent" robots as flagship products — steering the industry from building machines that merely move toward machines that reliably perform jobs in factories, farms, hospitals, and hazardous sites. For a sector that spent the last decade proving it could make a robot walk, the new mandate is explicitly about throughput, not demos.

Star Era's CEO says embodied AI's real business will be selling tokens, not robots. Gao Jiyang, founder and CEO of embodied-AI startup Star Era (Xinghai Tu), argued at a conference forum that the core business model for embodied intelligence will shift to "token sales" as the dominant form. His framing traces a three-stage arc: first hardware sales to validate scenarios, then subscription-based solutions, and finally selling the tokens — the units of inference — that let a robot act in the physical world, much like cloud AI charges per token today. It's a provocative bet: that the durable margin in robotics won't come from the steel, but from metered access to the brain that drives it. Whether buyers pay per-action for a warehouse robot the way they pay per-token for a chat model is still unproven, but it signals where the leading labs think the money is heading.

What to watch: whether MIIT's "flagship product" push translates into procurement, and whether the token-metering thesis shows up in actual robotics pricing this year.

Is "robots-as-a-token-metering business" a real shift or just conference-stage framing? Tell us in the comments.

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

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

Read more

AI 101 — How do AI video generators work?

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

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