It's easy to write off as a failure, as doing so requires little to no engagement with the internal structure and struggle of the project, which is the largest in informatics and neuroscience that I am aware of. If you don't agree, please cite something else on the same scale.

I'm involved with HBP and other non-European and European national neuroscience projects (though I am neutral here, just a university-employed engineer), and the science in the recent phase (starting in April this year) is a complete break with preceding funding phase and on-par with other projects. The people leading the science subproject are actually leading PIs on the topics. For comparison, let's name Allen Institute, though Allen is more a monolith, whereas HBP is more like micro-services, with a high cost in coordination, but now a focus on "FAIR" aspects like open-ness and reusability.

The infrastructure side is also coming along, and will provide an AWS of neuroscience (6 national HPC sites, multi-zone OpenStack/OpenShift, 2 neuro-morphic platforms, neuro-robotics platform, federated IAM, metadata and object storage stack, dedicated clinical infrastructure, etc) , but the failure or success will depend a lot more on the larger community getting convinced to engage their workflows on the platform. If people collectively agree that it isn't worth using, then the infrastructure teams never get to justify the whole thing, and it will flop. In that sense, the marketing aspect is not a facade, but a necessary condition for HBP to become a EU RI.

Of course I have no chickens in the fight; if it flops I can keep writing my shitty works-on-my-computer codes for one off unreproducible research projects.