The primary inherent advantage of an Open Source unit, which is also modular and designed to use common, widely available parts, is that it can be economically maintained over the long term. All components used are easy to replace and can always be sourced, in the near or distant future, from many different suppliers.
Because the parts can be sourced from suppliers other than the original manufacturer/designers, the incentive to engage in planned obsolescence is broken.
Additionally, there is no need for the company to stay around for the very long term in order for the unit to continue operation very long term. 70% of companies in the business of home improvement products are out of business within 10 years, and 50% are gone within only 3 years.
Long warranties are therefore nearly useless anyway.
It is important that building elements be effectively maintainable over the long term to get a good return on investment. Buildings are a long-term proposition.
If a unit lasts 30 years instead of 5 years, that means a naive reckoning indicates it's average cost of operation is 1/6th as much, per year. This has a major impact on return on investment. The OpenERV has only one wearing part, the bearings, and they can be replaced easily. The other parts can also be replaced indefinitely.
The use of additive manufacturing and open source components is no philosophical matter. In order for the custom mechanical components to be easily and economically obtained into the distant future, they are both necessary. No machine of this kind can be produced without a few custom mechanical parts. The only practical way to make such custom parts widely available from someone other than the original manufacturer is to use common digital manufacturing processes and publish the digital manufacturing files.
Fortunately, for electronics, the ecosystem for components has a lot of good commodity parts that are widely available.
Note: there are some IP issues with the heat exchanger (there is a patent pending which sucks), that part should not be considered "pure" open source, however patents don't impede the ability to produce one for personal use/maintenance/repair/replacement, only to sell them. In the manual I describe a simple approach to make a reasonably good although not quite as good heat exchanger using clever slicing/infill from the CAD file, which is free of IP encumberance (you can increase density or make it longer for higher efficiency at some cost to flow rate, you could again compensate for this by making two machines or using four fans per machine instead of two. The original heat exchanger is more elegant but you can get to the finish line with either approach.). (executive summary: Grid infill, not top or bottom, suitable density and road thickness = a cylinder with parallel square channels of the right size and density. PLA actually makes an excellent heat exchanger material. Coating it with sorbent is unfortunately far more complex and requires adhesives and chemicals, a grinder etc., describing that will have to wait for another day, it is highly unlikely anyone will use such an approach anyway, even I knowing exactly how to do it hesitate to set up again in order to do it to make prototypes.)
Note: You are welcome to build your own but I honestly think most people don't have enough respect for manufacturing, it is not that easy, there is much more to it than the blueprints/source code, unfortunately, and I cannot make a complete instruction manual or a textbook on printing and electronics to go with the design, especially because it is very rare anyone makes one. Some people have complained it is not easy enough to make, however no other project you will find of such complexity is going to be any easier, really. It's not supposed to be a thing you can easily make yet, it's not done and secondly the purpose of the source code is to primarily allow for maintenance and modification/hacking/aftermarket parts that are compatible etc. rather than diy manufacture. DIY manufacture would entail simplifying it in many ways which would compromise performance, and I would have to make instructions etc. and again it's very rare anyone try to make such a machine, in my experience, so there is little point in paving that road.