The inherent material imperfections of solid core optical fiber, for example, Kerr nonlinearity, chromatic dispersion, Rayleigh scattering and photodarkening, set fundamental limitations for further improving the performances of fiber-based systems. Hollow-core fiber (HCF) allows the light to be guided in an air core with many unprecedented characteristics, overcoming almost all the shortcomings arising from bulk material. The exploitation of HCF could revolutionize the research fields ranging from ultra-intense pulse delivery, single-cycle pulse generation, nonlinear optics, low latency optical communication, UV light sources, mid-IR gas lasers to biochemical sensing, quantum optics and mid-IR to Terahertz waveguides. Therefore, the investigations into the guidance mechanism and the ultimate limit of HCF have become a hot research topic. In the past two decades, scientists and engineers have fabricated two types of high-performance HCFs with loss figures of 1.7 dB/km and 7.7 dB/km for hollow-core photonic bandgap fiber (HC-PBGF) and hollow-core anti-resonant fiber (HC-ARF) respectively. In comparison with the twenty-years-old HC-PBGF technology, the HC-ARF that recently appeared outperforms the former in terms of broadband transmission and high laser damage threshold together with a quickly-improved loss figure, providing an ideal platform for many more challenging applications. While the guidance mechanism and fabrication technique in HC-PBGF have been well recognized, the HC-ARF still has a lot of room for improvement. At the birth of the first generation of broadband HC-ARF, the guidance mechanism was unclear, the fiber design was far from perfect, the fabrication was immature, and the optical properties were not optimized. In the past five years, we have developed an intuitive and semi-analytical model for the confinement loss of HC-ARF and managed to fabricate high-performance nodeless HC-ARF. We further employ our theoretical model and fabrication technique to well control and design other interesting properties, such as polarization maintenance and bending loss in HC-ARF. For a long time, the anti-resonant theory of light guidance has been regarded as being qualitative, and the leaky-mode-based HC-ARF have been considered to have worse performances than the guided-mode-based HC-PBGF. Our investigations in theory and experiment negative these prejudices, thus paving the way for the booming development of HC-ARF technologies in the near future.