“NVIDIA's CUDA platform was available for approximately 10 years before the AlexNet model was trained on it, demonstrating a long-term, visionary investment.”