Airbound Labs Patent Portfolio Analysis

The money arrives thirty months later

Airbound was founded in India in 2020. Its last public patent filing is dated 8 February 2024. The $37M Series A from Greenoaks, Lightspeed India Partners and DoorDash landed thirty months after that date. Patent applications publish roughly eighteen months after filing, so everything the company has filed since mid-2024 is invisible. The visible record and the funding event are separated by a gap wide enough to hide an entire second wave of filings.

What is visible is specific. Two families, both titled for a blended wing body tailsitter unmanned aerial vehicle. Filed in EP and WO, which is a route chosen by people who intend to enter multiple national markets and want to defer the decision about which. Claims sit in B64C, the aircraft group. This is an airframe portfolio, not a logistics-software portfolio, and it was laid down before the money.

Nothing granted, everything still pending

The granted side of the record is empty. Zero grants, two published and pending applications. First grants are expected around 2027. Until then there is no enforceable right anywhere in the portfolio, in any jurisdiction, on any claim. What exists is a priority date and a disclosure, which are not nothing but are also not a fence.

The skew is normal for a company that started filing in 2024. Two years of prosecution is ordinary. What matters is that the claims which eventually grant will be narrower than the claims as filed, and nobody outside the file knows yet by how much. In a portfolio of two, the difference between a broad grant and a narrow one is the difference between a position and a footnote.

In a portfolio of two, scope is the entire story.

Two filings against seventy-four, on one scale

The same CPC groups have drawn Textron Innovations with 74 families over five years, Beta Air with 54, Sz Dji with 34, Archer Aviation with 33, Jetzero with 29 and Porsche with 29. Airbound sits on that chart at two. The aircraft classifications Airbound chose are among the busiest in aerospace, populated by incumbent airframers, electric aviation entrants and the largest drone manufacturer in the world.

Count is the wrong metric here and the chart makes that clear by being so lopsided. A large filer in B64C covers many configurations shallowly. A single company filing twice on one configuration, the blended wing body tailsitter, is making a concentrated bet that a particular airframe geometry is the thing worth owning. Whether that bet holds depends entirely on what survives prosecution, not on how many neighbours share the shelf.

Five citations, and who they came from

Applications this young are rarely cited at all. Airbound's have been, by Milvus Usa, Newman Glen, Torc Robotics, Davis William and Fox Rabinovitz Joseph R, one citing publication each. Two of those are organisations, three read as individual inventors. Torc Robotics works in autonomous ground vehicles, which means the disclosure is being read across a boundary, not only by people building aircraft.

Forward citations on a pending application are a weak signal taken alone. What they do establish is that examiners and drafters in adjacent areas found the disclosure relevant enough to place it in the prior art of their own work. That is early evidence the technical content sits somewhere people are actually building, rather than in a corner nobody visits. Five is a small number. It is not zero.

What the record does not tell you

Two published families does not mean two families exist. The publication lag makes the last thirty months unreadable, and a company that raised $37M has had reason to file. Zero grants does not mean zero position, because pending claims still carry a 2024 priority date. And a crowded classification does not mean the space is taken, because scope in B64C is defined by configuration, not by headcount of filers.

One inference does hold, and it is the uncomfortable one. Naman Pushp is named on both families. The entire public inventive record attaches to one person. Meanwhile the hiring is 36 open roles, of which 6 are software and AI and 10 are operations and finance, against a claim set that is entirely airframe hardware. The company is staffing for something the published claims do not describe.

The filings are hardware. The hiring is not.

The gap between filed and built

Airbound filed twice in 2024 on an airframe, then went quiet in public while raising from Greenoaks, Lightspeed India Partners and DoorDash and opening 36 roles weighted toward software, operations and field work. The published moat is a wing shape. The company being assembled looks like a delivery operator. Those can be the same company, but only if there are filings nobody has seen yet.

The thread through all of this is timing. A two-family portfolio with one inventor and no grants is exactly what a hardware company looks like eighteen months into filing, and also exactly what a company looks like if it stopped. The record as published cannot distinguish between the two. What resolves it arrives on its own schedule, as publications, starting roughly eighteen months after each filing that has already happened.