The natural word for that is "capacity" rather than "capital".
For what's it's worth we targeted between 1 and 10 million units as a one-time bid (without them having to continue to guarantee availability ), and also targeted a ridiculously long fulfillment period, like a year - which would be enough to build any capacity they want, to fulfill that one bid. With the understanding that in the meantime (while the bidders' wait the year or however long) a newer version could be announced, no harm no foul and the bidders would not get that, and also that this is a way for the bidders to support the foundation, without getting support (such as supplier support) in return.
Finally, we thought that Broadcom might not want to flood the market with cheap chips, so we positioned it as a genuine set of orders around 1000 pieces, by entrepreneurs. (For real.) This would then lock these small-time entrepreneurs in with the Broadcom family (there are other ARM suppliers), and Broadcom could therefore upsell them on the rest of their full production solutions, since the small sellers would not have unlimited access to more chips and if their small marketing takes off they will want to place large orders for chips that are actually available, on their custom PCB.
At any rate, whatever the issue was, it wasn't capital.
I suspect the reason is quite obvious. They are losing money on every unit produced.
This is a "loss leader" strategy to get devs committed to the Broadcom platform only to find out "actually $5 computers don't exist, they are really $35".
Go to Mouser / DigiKey and put in the part number for the Broadcom chip, choose a very high number as quantity - look at that price.
Further contact a PCB and Assembly outfit and get a quote from them for a small board with a 100 or so SMD placements. You will see that $5 is complete fantasy.
RPI Foundation could've just marketed this as "cheap development platforms for you to try out your code, but only 1 per customer". Instead they've tried to maintain that this is some sort of "real product" that is commercially available at otherworldly low prices.
For what's it's worth we targeted between 1 and 10 million units as a one-time bid (without them having to continue to guarantee availability ), and also targeted a ridiculously long fulfillment period, like a year - which would be enough to build any capacity they want, to fulfill that one bid. With the understanding that in the meantime (while the bidders' wait the year or however long) a newer version could be announced, no harm no foul and the bidders would not get that, and also that this is a way for the bidders to support the foundation, without getting support (such as supplier support) in return.
Finally, we thought that Broadcom might not want to flood the market with cheap chips, so we positioned it as a genuine set of orders around 1000 pieces, by entrepreneurs. (For real.) This would then lock these small-time entrepreneurs in with the Broadcom family (there are other ARM suppliers), and Broadcom could therefore upsell them on the rest of their full production solutions, since the small sellers would not have unlimited access to more chips and if their small marketing takes off they will want to place large orders for chips that are actually available, on their custom PCB.
At any rate, whatever the issue was, it wasn't capital.