Just want to voice my interest in this area and say thank you to Jon and Gil for their work.
This is a very, very worthwhile endeavor and I really hope to see robust rhino/GH/ArchiCAD interoperability come to fruition.
I also want to address the perception that grasshopper is for complex organic geometries. While it can handle such geometries very well, it is also very useful for run-of-the-mill projects. The nodal, graphical algorithm interface is incredibly useful for a lot of standard architectural work, too. Factories, warehouses, school buildings, spec office space, data centers, convention centers, airports, many types of health care buildings are ideal candidates for modeling in grasshopper. Any building type which has a high degree of repetitive components and systems can greatly benefit from the efficiencies of model reuse during the concept design phase from modeling in a nodal interface --- specially when such models, once created, are reusable with only tweaking from project to project.
Currently, I am exploring GH's potential with respect to heavy industrial building types. It is a great way to evaluate design options. Revit has similar functionality, but so far my experience is that Revit's menu-based interface makes it more difficult to "see" the underlying algorithm and that its system is not as flexible and feature-rich as grasshopper's system (mostly because so many people are developing tools for grasshopper). Autodesk is addressing this by building dynamo --- a nodal interface for vasari and revit. I have not tried it so I cannot comment. But back to grasshopper and why there should be great interoperability with ArchiCAD.
First of all, using GH is a fast way to make adjustments to parameters and have then entire model update accordingly. Imagine a data center layout where adding one more structural bay leads the slab, walls, roof, and openings to update automatically, as well as respacing/adding downspouts or skylights, or egress walkways, etc. Or imagine if changing column types makes the slab perimeter offset and the location of exterior walls to the column grid centerlines automatically update. Or imagine if switching from an EPDM roof to a standing seam metal roof (with different slope) automatically makes the parapet heights adjust. I am not a super savvy user but after a little while with grasshopper, this is getting easy, even for me.
Second, grasshopper is very flexible and powerful in letting users identify, import and use, and then extract useful information from the models. Information may be exported for use in Excel or other CAD applications and used for budget estimating and scoping.
Third, grasshopper has an evolutionary solver (Galapagos), which means --- depending how far you refine the model --- you can for instance optimize floor to floor height based upon some matrix of lighting levels, heating/cooling systems specifics, percent openings in the envelope, and envelope performance metrics. Galapagos is fun to use and narrows down the design space to the variations which optimize the selected parameters. This can also be done by exporting data to CSV's, opening in excel, and reviewing the outputs of alternatives.
Fourth, the nodal workflow is outstanding, flexible, and powerful and now I even work out some details in grasshopper just so I can quickly run through a range of parameters and look at the implications of various sizes and placements of components. In summary, grasshopper can be a superb conceptual tool and a superb sketching tool.
Fifth, rhino/grasshopper are not competing with ArchiCAD. Rhino's biggest weakness, in my opinion, is that its documentation tools remain weak, even in Version 5. Conceptual modeling/sketching in rhino/grasshopper needs to be paired with a good design and documentation application. Once the design does move into documentation, changes and refinements will inevitably continue to be made and at that point, it is still important for the documentation tool to have all of the modeling capacities that ArchiCAD has. So these two applications really could have a very symbiotic relationship.
Fifth, as Jon pointed out, rhino and grasshopper are ubiquitous in architecture schools. And with good reason. McNeel is a great, fair, company with great products overall. But they also make rhino very attractive for students ($195 per perpetual academic license) and all academic licenses automatically convert to full commercial licenses upon graduating from / leaving school, at which point the student is on the commercial upgrade path. This means that over the last ten or so years, armies of current and future designers have been (often unknowningly) purchasing their first commercial design software and will start their careers with rhino+grasshopper as their go-to, affordable CAD/modeling tool. Also, Grasshopper is still free as of now, and rhino by itself is one of the best conceptual modeling tools around. In fact, in my case, i found it so fast and easy to learn and so powerful that i have never gone back to sketchup.
So ArchiCAD integration with rhino would tap into a huge market of curent and future designers. Currently, it appears to me that Autodesk and Bentley haven't taken much more initiative than Graphisoft seems to have taken in integrating with rhino/grasshopper (though it seems autodesk is starting to focus on rhino/gh integration now in the last year +/-). But, importantly, Autodesk's licensing and subscription policies are expensive enough that very few students will acquire personal copies of autodesk software when they leave school. Then there is the their large, heavy softwares that often require several expensive plug-ins purchased in addition to the Autodesk software in order to optimize work flows. And their resource-hungry tools often require more powerful hardware to run efficiently --- so there are a lot of additional costs associated with going the Autodesk route which makes it unlikely that young practitioners will purchase Autodesk tools as their personal complement to rhino/grasshopper. So there is a void space for some vendor to come in and claim the position of the design and documentation package most interoperable with rhino/grasshopper that is a reasonable reach for practitioners to purchase as their first legitimate commercial documentation tool. ArchiCAD is a good upfront value when compared to the other major players and their subscription service is also more reasonable.
So I hope to see this interoperability develop because it would be useful, practical, powerful, and it seems that currently there is a market void to be filled.