This paper presents a method for efficiently generating fault-tolerant Gottesman-Kitaev-Preskill (GKP) qubits using generalized photon subtraction with adaptive Gaussian operations. GKP qubits are a promising candidate for fault-tolerant quantum computing, as they can be implemented using linear optics and homodyne measurements. However, the generation rate of GKP qubits has been limited by the probabilistic nature of heralding methods, which typically result in low success probabilities. The proposed method improves the success probability by utilizing adaptive Gaussian operations to synthesize GKP qubits from multiple quantum states. The initial state preparation involves photon number measurements and adaptive operations that allow any measurement outcome above a certain threshold to be considered a success. This threshold is lowered by using generalized photon subtraction, which enables the synthesis of GKP qubits from non-Gaussian states. The method involves generating non-Gaussian states using generalized photon subtraction (GPS), which is a heralding method that allows the generation of non-Gaussian states by measuring photon numbers. The GPS method is combined with adaptive Gaussian operations to synthesize GKP qubits. The success probability of generating fault-tolerant GKP qubits in a realistic system exceeds 10%, which is one million times better than previous methods. The proposed method is implemented using a system with 20 GPS units, which contain 40 squeezed light sources and 20 photon-number-resolving detectors. Each squeezer needs to emit up to 18 dB squeezed vacuum states, and each detector needs to resolve up to 20 photons. The success probability of the proposed method is significantly higher than that of the original Gaussian breeding protocol, which is a method for generating GKP qubits using Gaussian states. The paper also discusses the advantages of the proposed method, including the ability to use adaptive elements to improve the success probability. The method is compared with other heralding methods, such as cat breeding, and is shown to be more efficient in generating GKP qubits. The simulation results show that the proposed method can generate GKP qubits with a success probability of about 0.01% to 0.5%, which is a high probability for single-shot state generation. The paper concludes that the proposed method is a powerful technique for generating fault-tolerant GKP qubits at a practical rate, which is essential for the realization of optical quantum computers. The method is expected to accelerate the development of optical quantum computers by enabling the efficient generation of GKP qubits.This paper presents a method for efficiently generating fault-tolerant Gottesman-Kitaev-Preskill (GKP) qubits using generalized photon subtraction with adaptive Gaussian operations. GKP qubits are a promising candidate for fault-tolerant quantum computing, as they can be implemented using linear optics and homodyne measurements. However, the generation rate of GKP qubits has been limited by the probabilistic nature of heralding methods, which typically result in low success probabilities. The proposed method improves the success probability by utilizing adaptive Gaussian operations to synthesize GKP qubits from multiple quantum states. The initial state preparation involves photon number measurements and adaptive operations that allow any measurement outcome above a certain threshold to be considered a success. This threshold is lowered by using generalized photon subtraction, which enables the synthesis of GKP qubits from non-Gaussian states. The method involves generating non-Gaussian states using generalized photon subtraction (GPS), which is a heralding method that allows the generation of non-Gaussian states by measuring photon numbers. The GPS method is combined with adaptive Gaussian operations to synthesize GKP qubits. The success probability of generating fault-tolerant GKP qubits in a realistic system exceeds 10%, which is one million times better than previous methods. The proposed method is implemented using a system with 20 GPS units, which contain 40 squeezed light sources and 20 photon-number-resolving detectors. Each squeezer needs to emit up to 18 dB squeezed vacuum states, and each detector needs to resolve up to 20 photons. The success probability of the proposed method is significantly higher than that of the original Gaussian breeding protocol, which is a method for generating GKP qubits using Gaussian states. The paper also discusses the advantages of the proposed method, including the ability to use adaptive elements to improve the success probability. The method is compared with other heralding methods, such as cat breeding, and is shown to be more efficient in generating GKP qubits. The simulation results show that the proposed method can generate GKP qubits with a success probability of about 0.01% to 0.5%, which is a high probability for single-shot state generation. The paper concludes that the proposed method is a powerful technique for generating fault-tolerant GKP qubits at a practical rate, which is essential for the realization of optical quantum computers. The method is expected to accelerate the development of optical quantum computers by enabling the efficient generation of GKP qubits.