Microsoft’s plan to fix its chip problem is, partly, to let OpenAI do the heavy lifting (techcrunch.com)

Microsoft has struck a pragmatic hardware pact with OpenAI: rather than rebuild a cutting‑edge chip program from scratch, Microsoft will adopt and extend OpenAI’s custom AI chip designs (co‑developed with Broadcom) after securing intellectual‑property rights under a revised partnership. CEO Satya Nadella says Microsoft will get access to OpenAI’s system‑level innovations and will continue to access OpenAI’s models through 2032; the only major exclusion is OpenAI’s consumer‑hardware plans, which remain its own. Bloomberg first reported the arrangement, which effectively lets Microsoft leverage OpenAI’s heavy investment in chip engineering to shore up its own underperforming semiconductor efforts. The deal matters because designing high‑performance AI accelerators — including chip architecture, system integration, and firmware — is extremely expensive and complex. By inheriting proven designs, Microsoft can accelerate deployment of optimized hardware across Azure, close gaps with rivals like Google and Amazon, and focus internal resources on systems integration and scale. Technically, the move implies closer hardware‑software co‑design (chips tailored for large model workloads), faster time‑to‑market for custom accelerators, and potential shifts in supply chains given Broadcom’s role. It also creates strategic dependency on OpenAI’s IP: Microsoft gains speed and capability but ties some future chip strategy to a partner that retains certain hardware freedoms.