The number is in the filing or it isn't — and a patent is not a filing. That's the first rule of this desk. Across our recent coverage we've treated grants like Rivian's battery-architecture patent, GM's reference-electrode diagnostics, and NVIDIA's LiDAR-perception method as signals — but signals of intent and direction, not of financial results. This piece is the discipline behind that.

Rule one: a patent shows where capital is being pointed, not where it landed. A battery-architecture or inverter grant tells you a company is investing in a cost lever. Whether the lever moved cost-of-revenue is a question only the cash-flow and income statements answer. Treat the grant as a hypothesis the filings later test — and be willing to find the hypothesis didn't pan out.

Rule two: read the claims and the CPC codes, not the title. The CPC classifications tell you what a patent is actually about — H01M and B60L for battery and propulsion, B62D for steering, G01S and B60W for sensing and autonomous control. A grant classified in pack protection is a manufacturability-and-warranty story; one classified in LiDAR perception is an ADAS-cost story. The codes route the patent to the right financial line.

Rule three: distinguish the assignee. A supplier's patent (BorgWarner, DENSO, ZF, Murata) is a supply-chain and content-per-vehicle signal; an automaker's patent is an in-house capability signal; a non-automaker's (Apple, NVIDIA) is a competitive-positioning signal about who captures a layer's value. Same document type, three different financial reads — and the assignee decides which.

Rule four: never headline a number that isn't filed. A patent contains no revenue, no margin, no reserve. When we tie a grant to economics, the economics come from the SEC filings, surfaced through the filing evidence index, and the patent supplies only the direction. The patent is the hypothesis; the filing is the receipt. Confuse the two and you're writing the press release the company would have written.

The honest summary: automotive patents are genuinely useful to a financial reader as soft, early, directional signals — and dangerous if read as anything more. Use them to know where to look in the next filing. Then go look.