On the demand side - many of the intended customers of SBIR grants (government agencies who need a specific technology solution) can get what they want faster, and with lower risk, through an existing contractor. You could gamble with a two person company funded by a few hundred thousand government dollars, or you could go to a Raytheon or a United Technologies and get the same thing made by a team of experts who will build you exactly what you want. In terms of deliverables, the risk profile is much lower (and the resulting upside to your mission is much higher) if you go with the big contractor. SBIR exists to fund some of the riskier technology efforts, but also to give the government some cover in terms of perceived equality. They can point to SBIR and say "Look, the little guys had a chance to bid too! We even gave them money!"
On the receiving side, I've met more than a few researchers who have gotten SBIR grants for their work, and effectively used them to pay for research/proof of concept that they have already done. I.E. the SBIR grant (Phase I) pays for work that they've done over the past year or so. Many of the folks who take Phase I grants that I've spoken to have no interest in commercializing and productizing their work. They are interested in basic science, not product development.
I'm not sure how to get over what seems like a pretty serious disconnect here.