Major pharmaceutical corporations frequently sell deprioritized drug rights for nominal sums because analyzing vast scientific literature is difficult. Shkreli explains acquiring these discarded molecules and pivoting them from saturated conditions to unaddressed rare diseases creates massive financial returns.
Shkreli argues quantum systems run at slow hertz-scale clock speeds, lack quantum RAM, and accelerate only specialized integer factorization algorithms. Consequently, they cannot sort data faster or outperform standard GPU clusters on enterprise business workloads.
Shkreli notes matrix multiplication requires massive operations on standard graphics processors, but light-based analog optical computing executes these calculations instantly. Emerging startups developing photonic cores could dramatically outperform current silicon architectures, eventually sending existing GPUs to scrap heaps.
Despite optimistic predictions by tech leaders, Shkreli asserts AI software cannot meaningfully aid clinical drug discovery. Developing therapies requires complex real-world biological and regulatory validation that computing clusters have historically been unable to simplify or accelerate.
Limiting the artificial intelligence total addressable market to human labor substitution ignores corporate appetite for cognitive compute. Enterprises like McDonald's could spend billions deploying massive GPU fleets to resolve major strategic business decisions, generating high net present value from unlimited intelligence.
Highly intelligent entrepreneurs often face severe self-imposed pressure to excel across every discipline. Shkreli reflects that his refusal to admit he was an inferior hedge fund trader compared to his entrepreneurial talents directly caused his personal downfall and imprisonment.