The design of a transformer at 10s of MHz is not the main issue in making a switching power supply operate at high frequency like that. There are plenty of ferrites that can be used as the core material that won't be particularly lossy at that frequency. The biggest issue will be to minimize capacitive coupling between the primary and secondary and between parts of each winding. This might call for a toroid with each winding not quite spanning half the toroid so that there is some gap between then and the ends of each winding are apart.
The real problem will be switching losses. When the whole switching cycle is only 100 ns or less, then a 10-20 ns switch transition time is significant. If the switch is a FET, then then charging and discharging the gate 10 M times a second represents significant current. Fast parts generally cost more and require more power to drive. While the inductor can be made nicely small, the loss of efficiency at this frequency will limit size due to heat dissipation problems.
It sounds worth looking into a resonant design. That might actually put parasitic capacitance accross each transformer winding to use with some cleverness.