FBO blurb from another website.
Cost per se is not a big blocker, a multi-mode single clad fiber costs a few $$ per meter. Even with aviation qualification costs, it won't be terribly expensive. And not in itself significantly more expensive than the equivalent electrical wires. And as you mention, fiber optics are not exactly new. They've been around for 40 years, used extensively since at least 20 for data, and used since 15 years for high power applications or applications in hostile environments. I started my working career on high power fiber laser 10 years ago, and high performance fibers were already reliable and cheap back then.
The problem is more that the design organisations are just terribly conservative. Anything designed on an aircraft is tightly constrained by a multitude of rules & regulations, either from certification or internal design rules from each OEM. And the design of every single part on an aircraft calls upon many different specialists within the organisation, with each an own domain of competence. To manage this complexity, the only solution today is basically to do the same things over and over again. If it worked before, it should work again.
Airplane data networks have historically been fragmented point-to-point designs with heavy use of unsophisticated analog and discrete links, or limited buses like Arinc 429. This in turn led to specific design thinking, performance requirements, installation solutions and other various constraints, and this in turn makes it difficult to move to networks with full duplex buses.
What this then implies is that airplanes tend to be designed with a multitude of low data volume point-to-point links rather than a few wires carrying big amounts of data.
But as fiber optics are mainly useful for their capability to carry lots of data, they cannot be used to the best of their abilities in such an architecture, and the business case conclusion is usually "well there is no added value to fiber optics".
In addition, fiber optics has its own installation needs & constraints vs electrical.
For example, electrical connectors work OK as long as some contact is established between the 2 parts. So having lots of electrical connectors is acceptable, even if it's better to minimise the number. However, fiber optics really don't like connectors because light has to jump from one fiber to the other without being disturbed too much. But imagine connecting these things in a dirty FAL...
Other problem : electric-to-optic conversion is roughly 50% efficient. So each time there is a conversion, there is a lot of loss. So such conversions have to be kept to a minimum.
On the + side, fiber optics are low weight and small. So compared to electrical cables, the designers could really benefit by installing dozens or hundreds of fiber optic links all over the place. But that would require a new approach to systems certification. And more simply, it's not how things are done today.
The other big advantage of fiber optics is the interference/EM robustness. Which is directly useful in aircraft like the P1 carrying lots of sophisticated EM sensing devices. But on airliners, the gain would be to reduce the segregation requirements between various routes. But unless the aircraft's systems are completely re-designed to exploit this property, it again doesn't add much value.
So the point is that to fully benefit from the advantages of fiber optics while avoiding their drawbacks, the complete design of at least the data communication networks of the aircraft needs to be re-thought. More likely, the design of all its systems needs to change, including all performance calculations, certification rules & acceptable proof of compliance, test methods, installation rules & design, etc...and more importantly, requiring a change of mindsets, organisation and responsibilities. If designers just try to replace electrical cables with fiber 1-for-1, it'll never work. It's like trying to fit a square in a circular hole.
So not gonna happen any time soon, at least from my perspective. But my deepest wish is to be wrong...