Segment disclosure, not the keynote — except Apple never gave either. Project Titan ran for years with no line in any 10-K. So the public record falls back to patents, and Apple Inc.'s June 2, 2020 grant US10671068B1 is one of the clearest traces: shared sensor data across processing pipelines.

Read the architecture, not the rumor mill. The CPC tags — G05D 1/0088 autonomous control, G01S 17/931 LiDAR-for-driving, G01S 17/08 ranging, G06N 5/04 and G06N 20/00 learning — describe a perception data fabric for a self-driving platform. Apple was engineering at the systems level, not dabbling.

“Sensor data captured at by different sensors may be shared across different sensor processing pipelines. Sensor processing pipelines may process captured sensor data from respective sensors.”— U.S. Patent No. 10,671,068 source

The decoder problem is that none of this reconciled to a number. Apple's R&D line is a single giant figure; autonomy was buried inside it. For an analyst, that is the worst kind of spend — material enough to generate dense IP, invisible enough that you cannot size it from the filings.

This is exactly where patents earn their keep as evidence. When a company refuses segment disclosure, the patent record becomes the only verifiable signal that the program is real and technically serious. This grant says Titan was both.

The honest limit: a sensor-pipeline patent tells you Apple built autonomy infrastructure, not how much it cost or whether a car was ever coming. The 2024 wind-down of the car effort eventually answered the second question; the patents answered the first all along.

The takeaway for the money desk: when a company hides a program inside an aggregate R&D line, read the patents to confirm the program exists and to gauge its seriousness. Apple's autonomy spend was undisclosed, but it was never invisible.