AI is fundamentally reshaping the global computing stack, transitioning from legacy PC-era architectures to high-throughput "token factories." Michael Stewart, from Microsoft’s venture arm M12, emphasizes that current investment strategies must prioritize strategic signaling over simple SaaS bolt-ons. The industry faces critical "compute starvation," driving the need for efficient model routing and the emergence of wholesale token purchasing models for large enterprises. Beyond software, the data center of the future demands a radical re-engineering of semiconductor fabrication, moving away from standard foundry models toward AI-purposed infrastructure. Furthermore, material science serves as a vital frontier, where AI-driven discovery—supported by automated testbeds—promises to accelerate breakthroughs in energy and hardware efficiency. These shifts represent a profound, long-term disruption rather than a transitory bubble, necessitating significant capital allocation toward foundational compute and material innovation.
Sign in to continue reading, translating and more.
Continue