The technology really isn’t all that new; many DC/DC converters operate with efficiencies in the 90+% range with 95% not uncommon*, although advances in semiconductors and other parts continuously improves the efficiency (for example, instead of using diodes to rectify the output, use a MOSFET; a normal silicon diode might drop 1 volt (0.6-0.7 v is just when they start to conduct) while a schottky diode 0.5 volts and a 2 milliohm MOSFET 0.1 volt, or 50, 25 and 5 watts respectively at 50 amps; the MOSFET will be even better at lower currents since voltage drop is resistive, although it does require additional circuitry to switch it on at the proper times). The elimination of a bridge rectifier however does make a huge difference at low voltages, since even schottky diodes will drop around 1 volt (2 diode drops), or a nearly 10% efficiency drop at 12 volts.
*Personal anecdote, I recently made a 100 watt wall-wart type converter for a personal project (20 volts, 5 amps, with PFC), with an efficiency of about 93% (it still gets warm enough without ventilation that I put a small low-speed fan on it; it measures only 4.5 x 2.5 x 1.5 inches), achieved by using low on-resistance, high-speed FETs including a FET on the output for rectification (in this case, the FET can be directly driven from a winding on the transformer with minimal reverse current, although this decreases efficiency at low load since it always runs at near full-load duty cycle, but I also added a micropower standby power supply which supplies 5 volts at 10 mA (why do computers need multiple amps for standby, and wouldn’t it be easier not to have to switch between standby power and normal power when in sleep mode?) so it can be powered off).