NbTi accelerator dipoles are limited to magneticfields (H) of about 10 T, due to an intrinsic upper critical field (Hc2) limitation of 14 T. To surpass this restriction, prototype Nb3Sn magnets are being developed which have reached 16 T. We show that Nb3Sn dipole technology is practically limited to 17 to 18 T due to insufficient high field pinning, and intrinsically to 20 to 22 T due to Hc2 limitations. Therefore, to obtain magnetic fields approaching 20 T and higher, a material is required with a higher Hc2 and sufficient high field pinning capacity. A realistic candidate for this purpose is Bi-2212, which is available in roundwires and sufficient lengths for the fabrication of coils based on Rutherford-type cables. We initiated a program to develop the required technology to construct accelerator magnets from 'windand-react' (W&R) Bi-2212 coils. We outline the complications that arise through the use of Bi-2212, describe the development paths to address these issues, and conclude with the design of W&R Bi-2212 sub-scale magnets.