Hardware costs a ludicrous amount of money to design. It costs a ludicrous mount of money for each prototype run, of which there will be several. It costs a ludicrous amount of money to have PCBs fabbed. It costs a ludicrous amount of money to design a nice case that everything fits into and that doesn't destroy the radio signal. It costs a ludicrous amount of money to hire RF engineers for anything. It costs an above-average amount of money (for HN) to develop software and firmware for hardware. It costs a ludicrous amount of money to package for consumers, to store, and to ship. It cost a ludicrous amount of money to convince stores to carry your product. It costs a ludicrous amount of money to support the product and the customer after it has shipped to market.
I would be shocked if $32 million covered the development and first run of product (assuming first run >100,000 units).
I think John Sullivan is grossly underestimating how much it costs to develop a phone, let alone one that is totally open. Convincing companies to completely open source their drivers requires tons of effort, and there has to be a benefit to the vendor beyond just what you believe to be "right". In some cases that benefit is money, or purchases ("we'll only use your product if..."), but I don't think this fundraiser has near enough leverage via either of those means to both build a high quality device and manage to make it totally free.