Why everyone from OpenAI to SpaceX is building their own chips (and turning up the heat on Nvidia) - TechCrunch

For years, NVIDIA has dominated the AI chip market, but signs are emerging that this era of total dependence may be ending. Companies like OpenAI, Google, Apple, and SpaceX are increasingly investing ... Read original →

Tech giants designing their own AI chips isn’t merely about reducing reliance on NVIDIA—it’s a strategic race for computational sovereignty. Technically, this shift accelerates adoption of chiplet architectures and pressures EDA, advanced packaging, and sub-3nm nodes. Geopolitically, U.S.-China tech decoupling forces firms to diversify supply chains across Taiwan, China, and South Korea to mitigate single-source risk. In market dynamics, NVIDIA’s CUDA moat remains strong short-term, but Broadcom—via OpenAI’s Jalapeño—could replicate its data-center switching playbook by vertically integrating AI training silicon. Over the next 12–24 months, open-source IP and RISC-V will lower design barriers, enabling domain-specific AI accelerators that erode general-purpose GPU growth, making hardware-software co-optimization the new competitive frontier.