Exein Raises $270M at $1.7B to Secure Physical AI

Italian startup Exein has raised a $270 million funding round led by Headline at a $1.7 billion valuation.

The company issued the announcement from Rome on September 15, 2026. The round was significantly oversubscribed, according to the company. The $1.7 billion valuation is more than double its prior $770 million valuation, as reported by the Financial Times.

Headline is a San Francisco-based venture capital firm. It led the round. That math matters. An oversubscribed raise at more than twice the prior valuation signals sustained investor appetite for infrastructure around embodied systems, even at a time when late-stage funding remains selective.

Exein was founded in 2018 and is based in Rome. It describes itself as providing a cybersecurity layer for physical AI, a category that covers robots, drones and autonomous machines, according to the company. The framing is deliberate. The target is not cloud workloads or enterprise endpoints. It is machines that act in the world.

Its product, Photon, is a runtime security solution designed to prevent attacks at the kernel level of a device, as reported by TechCrunch. For practitioners, the distinction is familiar. Most IoT and embedded defenses focus on secure boot, signed updates or network policy. Runtime enforcement at the kernel sits underneath application logic. It is intended to catch malicious behavior during execution, including privilege escalation and tampering that static controls miss.

Exein claims to have secured more than 2 billion connected devices. Asia Pacific currently drives half of its revenue. Both points help explain the geographic logic of the round. The installed base is already broad and heavily weighted toward APAC manufacturing and deployment pipelines.

The company said it intends to use the fresh funding for M&As and hiring to accelerate expansion in the U.S. and Asia Pacific. The plan is expansion, not a product pivot. Acquisitions would extend reach or capability around the existing Photon base, while hiring supports go-to-market and engineering in two regions where deployment of connected and autonomous hardware is concentrated.

The broader context here is why kernel-level runtime is getting renewed attention. Physical AI collapses the usual separation between compromise and consequence. A breached server leaks data. A breached actuator moves incorrectly. Update cycles are slower, device lifetimes are longer, and connectivity is intermittent. That combination makes persistent, low-overhead enforcement attractive, particularly if it can run across heterogeneous silicon without rewriting application stacks.

In my view, worth flagging is the operational question that follows any claim at this scale. Securing 2 billion devices is a deployment statement, not a uniformity statement. Heterogeneity is the norm in embedded fleets, with different kernels, board support packages and lifecycle constraints. The technical test for Exein will be how consistently Photon policy holds across that variance, and how it handles false positives where a blocked syscall stops a machine rather than a process. Investors appear willing to fund that work. Operators will judge it on stability and overhead as much as on prevention.

If the approach holds, the upside is practical. Secure runtime primitives could let robotics and drone vendors ship faster without rebuilding safety and security logic per platform, and could give enterprise buyers a common control point across fleets that otherwise age unevenly. That is the long-arc benefit of infrastructure of this kind, less visible than the machines themselves but decisive for whether they can be trusted outside the lab.