The Real Keyword at the Power Battery Conference Isn't Solid-State

"Electric drive for everything" is more worth listening to than "solid-state batteries." At the conference in Yibin in early September, over 800 industry chain players gathered. Topics ranged from solid-state batteries, AI, ships, and eVTOLs, with the focus no longer just on cars. Power batteries are becoming a common foundation relied upon by shipping, low-altitude economy, robotics, and energy storage.

I recently spent two weeks running materials through an AI agent and agent harness, and also browsed open-source projects on Hacker News. The feeling was very direct. When AI enters the battery industry, the first thing to land won't be mass production of all-solid-state, but breaking down R&D, production lines, quality, and supply chains into verifiable processes. The conference mentioned that an electric ship can easily require tens of thousands of cells, with over 7,000 control points on the production line, achieving defect rates at the parts-per-billion level. This is where digitalization creates value. No matter how smart the model is, without task boundaries and acceptance criteria, it only generates pretty conclusions.

From a consulting perspective, this looks like a reshuffling of the value chain. Benchmarking against overseas cases, battery factories have long stopped selling just cells; they turn manufacturing data, recycling systems, and carbon footprint platforms into customer stickiness. Domestically, launching carbon footprint information platforms, solid electrolyte R&D, and smart labs together is the right direction. But solid-state still requires patience. The issues of sulfide electrolytes being "fast-conducting" yet "stable" haven't been fully resolved. While the inaugural year of semi-solid mass production sounds exciting, implementation will still get stuck repeatedly on interfaces, continuous manufacturing, and costs.

My advice: don't rush to slap the label of "disruptive" on AI projects. Start small with one scenario—material screening, anomaly detection, equipment maintenance, supplier compliance. Whichever has clear acceptance criteria, start there. Otherwise, you'll just end up with another chat box, not another production line.

When batteries become the base for everything, manufacturers won't just sell cells; they'll also sell the system that reduces uncertainty.