Massive 3D mapping of our universe using radio waves from distant hydrogen gas has the potential to become our most sensitive cosmological probe, shedding new light on dark matter, dark energy, our cosmic origins and ultimate fate. To fulfill this potential, however, future radio telescopes require extreme precision, which corresponds to both huge collecting area and unprecedented control of systematic errors. Unfortunately, great sensitivity traditionally requires a great budget. The cost for a steerable single-dish telescope grows rapidly with area, which has bolstered interest in interferometers. Interferometric arrays can be made arbitrarily large, but for a generic array layout, the data processing cost unfortunately grows quadratically with area, making a million-antenna array astronomically expensive. The Omniscope project is developing a novel approach where the cost per antenna grows only logarithmically, and precision calibration and data validation can be fully automated. We are currently building and testing a dual-polarization 64-antenna prototype using this scalable architecture, which can be integrated into future efforts currently envisioned by the cosmology community.