They've likely not exceeded the SR-71 by very much since then. The last several decades have seen the trend where costs are more frequently taken into account

They've likely not exceeded the SR-71 by very much since then. The last several decades have seen the trend where costs are more frequently taken into account when deciding whether or not to pursue a line of scientific research. This is akin to the new direction the lazy chip makers have taken with multicore-multichip instead of continuing to work at improving the individual chips as much.