Any future giant ground-based telescope must, at a minimum, provide foci for seeing-limited imaging over a wide field and for diffraction-limited imaging over ~1 arcminute fields corrected by adaptive optics (AO). While this is possible with a number of design concepts, our choices are constrained if we anticipate wanting to later add a second telescope for imaging with still higher resolution, and very high contrast imaging for exoplanet studies. This paper explores designs that allow for such future development. Higher resolution imaging by interferometric combination of the AO-corrected fields of two telescopes is possible without loss of point-source sensitivity or field of view, as long as the baseline can be held perpendicular to the source and can be varied in length. This requirement is made practical even for very large telescopes, provided both can move continuously on a circular track. The 20/20 telescope illustrates this concept. Telescopes so mounted can additionally be operated as Bracewell nulling interferometers with low thermal background, making possible the thermal detection of planets that would have been unresolvable by a single 20 m aperture. In practice, limits set by funding and engineering experience will likely require a single 20 or 30 m telescope be built first. This would be on a conventional alt-az mount, but it should be at a site with enough room for later addition of a companion and track. In anticipation of future motion it should be compact and stiff, with a fast primary focal ratio. We envisage the use of large, highly aspheric, off-axis segments, manufactured using the figuring methods for strong aspherics already proven for 8 m class primaries. A compact giant telescope built under these guidelines should be able to perform well on its own for a broad range of astronomical observations, with good resistance to wind buffeting and simple alignment and control of its few, large segments. We compare here configurations with adjacent hexagonal segments and close-packed circular segments. For given segment parent size and number, the largest effective aperture is achieved if the segments are left as circles, when also the sensitivity and resolution for diffraction-limited operation with AO is higher. Large round segments can also be individually apodized for high-contrast imaging of exoplanets with the entire telescope-for example 8.4 m segments will yield 10-6 suppression 0.05 arcsec from a star at 1 μm wavelength, and at 0.25 arcsec at 5 μm.