Future optical/infrared telescopes will need to be much larger than today’s, if they are to address such key challenges as direct observations of Earth-like exoplanets and of the first stars formed after the big bang. In this paper I consider the most promising of the new sites, both on the ground and in space, and telescope concepts to take advantage of their complementary scientific potential. Ground based telescopes with adaptive optics will be capable of diffraction limited imaging, down to a short wavelength limit set by the amplitude and speed of the atmospheric turbulence. The best conditions are on the high Antarctic plateau, where recent measurements at Dome C show turbulence typically half the amplitude of the best temperate sites, with temporal evolution at half the speed1. Thus uniquely in Antarctica, diffraction limited imaging at optical wavelengths should be practical. Conditions there are also best for infrared astronomy, given the combination of minimal aberration and winter temperatures averaging as low as 200K at Dome A (the highest point). In space, well away from the warm Earth, conditions are even better, with 24 hour/day observing free from all atmospheric aberration, and the potential for passive cooling to 50K or less by use of a sunshield. L2 and the Moon's south pole are such optimal space locations. A telescope at L2 requires only a little fuel to stay on orbit, and can be accurately pointed despite solar torques by well established active methods based on star trackers, gyros and reaction wheels. By contrast, the Moon provides a completely stable platform where a telescope with no moving parts can remain pointed indefinitely along the spin axis, or a telescope on a hexapod mount can be oriented and tracked by reaction to the turning lunar surface. Solar shielding on the Moon requires a polar location such as the high rim of the Shackleton crater, adjacent to the south pole, where there is also nearly continuous solar power. Long term operation large telescopes in space should be possible at affordable cost if we adopt the strategy used on the ground, where the same telescope OTA and mount is maintained for decades while instruments are periodically upgraded. HST has already shown the power of this modus operandi in space. It makes sense because the optical image quality of any telescope cannot be improved once the diffraction limit is reached, while instruments need to be renewed to keep pace with scientific and technical developments. Thus if future space exploration results in long-term robotic or human infrastructure on the Moon, the Shackleton rim would be favored as an observatory site, especially for ultra-deep optical/infrared surveys. If, on the other hand, exploration is centered a new station in free space, out of the Earth's gravitational potential well, observatories at L2 would be more easily supported. When contrasting the performance of ground and space telescope options, an important trade is larger aperture on Earth versus lower background in space The thermal zodiacal background of space is typically 105 times lower than even the Antarctic background, and the optical scattered starlight background in space is much less, but because of the strong dependence of sensitivity on diameter a 100 m telescope at Dome A or Dome C would have sensitivity and power to study Earth-like planets comparable to that of NASA's proposed TPF coronagraphic and interferometric missions combined. For ultradeep field studies in the infrared, integration time is also important, thus a 20 m fixed telescope on the lunar south pole surveying just the south ecliptic pole region would have nearly 100 times the sensitivity of the JWST at L2.
Neither Dome A nor the Moon’s south pole has yet been explored, even robotically. If large telescopes are ever to be built at these optimum sites, smaller precursors must be built first to develop the required technology and to gain experience. On the Moon, a start which would yield already interesting science could be made with a 3-m class, fixed, robotically-deployed survey telescope. On the Antarctic plateau, a 20 m copy of the Giant Magellan Telescope3,4 would be a good scientific and technological precursor to a 100 m telescope in Antarctica.